tldr
Tobi Lütke and John Collison sit down at a Cheeky Pint for a 100-minute conversation that covers the full arc of Shopify's evolution — from a snowboard shop to a commerce operating system processing billions. Tobi makes a compelling case that the next wave of commerce will be driven by AI agents acting as personal shoppers, that internal tools shape company culture more than any memo, and that stablecoins will fundamentally change how payments work. The throughline: Tobi is still, after 20 years, genuinely captivated by the problem of internet commerce — and that obsession is Shopify's moat.
Key Takeaways
- Fall in love with problems, not solutions. Tobi's framing: "The best gift in life is finding a beautiful problem you can never solve." The distinction between problem-people and solution-people is one of the most important hiring signals — and very hard to detect.
- Internal tools are culture. Shopify built GSD (Get Shit Done), an internal system that reflects how they think about work. The argument: the tools you build internally shape behavior more than any cultural doc.
- Agentic commerce is the next frontier. AI agents will act as personal shoppers — browsing, comparing, purchasing on your behalf. This changes the entire acquisition funnel. Discovery becomes agent-mediated, not search-mediated.
- Shop Pay's 16% conversion lift is a compounding moat. Payment UX isn't glamorous, but it's one of the highest-leverage things Shopify has built.
- Stablecoins will expand what can be purchased online. Tobi is bullish — not on crypto speculation, but on stablecoins as a practical payment rail that opens up new transaction types.
- Stripe + Shopify = Apple + TSMC. Tobi frames the 13+ year partnership as symbiotic infrastructure — each company makes the other stronger, and neither could easily be replaced.
- Taste scales, but barely. Tobi references Midjourney as a rare example of scaling aesthetic taste. Most companies lose taste as they grow. Keeping it requires the founder to stay hands-on.
- Shopify's codebase: ~20M lines of Ruby + 8-10M lines of TypeScript. Possibly the largest Ruby and TypeScript application in the world. A reminder that monoliths can work at enormous scale if the team is disciplined.
Timestamps
| Time | Topic |
|---|---|
| 0:00 | Intro — "The best gift in life is finding a beautiful problem you can never solve" |
| 1:00 | John Collison welcomes Tobi, Shopify's scale: ~20M lines Ruby + 8-10M TS |
| 2:30 | Companies as technology — creating social acceptance for missions |
| 4:00 | Gold mining executives vs. internet commerce — falling in love with problems |
| 5:00 | The unmeasurability of R&D and engineering productivity |
| 7:00 | GSD (Get Shit Done) — Shopify's internal tool culture |
| 9:30 | Rituals, project updates, and how GSD shapes company behavior |
| 11:00 | The dangers of internal tool culture — wrong paths and scope creep |
| 13:00 | Vendor software has a worldview — buy vs. build tradeoffs |
| 14:30 | Infrastructure thinking — environments that cause people to accomplish more |
| 16:30 | Software shapes organizations faster than policies |
| 17:00 | The Shopify origin: "You've done an online store, just like that?" |
| 18:00 | Small businesses in recessions and industry upheaval |
| 19:30 | Mobile commerce shift — merchants can't keep up alone |
| 21:00 | How Shopify thinks about SMB vs. enterprise |
| 23:00 | Merchants who started on Shopify and scaled to billions |
| 25:00 | Stripe Billing's time travel engine — complexity in billing |
| 27:30 | Infrastructure as composable primitives |
| 28:30 | Flash sales and requests-per-second thresholds |
| 30:00 | Supreme drops and Kylie Jenner launches — peak traffic war stories |
| 32:00 | Database locking, transaction architecture, and pool-based inventory |
| 34:30 | The transition to agentic commerce — AI personal shoppers |
| 36:00 | Shopify's role: infrastructure for agent-mediated discovery |
| 37:00 | Personalized ads as a force for good — Meta creating businesses |
| 39:00 | Agent-mediated commerce benefits the long tail of merchants |
| 40:00 | Consumerism vs. quality — "People throw things away because they hate them" |
| 42:00 | Product search and embeddings — commerce's under-invested AI frontier |
| 45:00 | Midjourney and taste in AI — driving toward a distinct aesthetic |
| 46:00 | Impulse purchases and the future of agent-mediated buying |
| 49:00 | Shop Pay and the history of accelerated checkout |
| 50:30 | HTTP 402 Payment Required — the web was built for commerce |
| 52:00 | Shop Pay's conversion lift and the direct-to-consumer boom |
| 54:30 | Kevin Kelly's 1000 True Fans and personalized advertising |
| 55:30 | Stablecoins on Shopify — pragmatic crypto, not speculation |
| 57:00 | Expanding what can be purchased with stablecoins |
| 59:00 | Making stablecoin payments invisible and frictionless |
| 59:30 | The Stripe–Shopify partnership — Apple and TSMC analogy |
| 1:01:30 | Karl Popper and finding beautiful problems — staying focused |
| 1:03:00 | "What problem do you understand better than GPT-5's training data?" |
| 1:05:00 | VS Code as craftsmanship — Erich Gamma's fourth editor |
| 1:07:00 | Founders building good product then losing touch with craft |
| 1:09:00 | The tsunami of demand when someone ships something authentic |
| 1:12:00 | Tobi's apprenticeship, no other job, learning what not to do |
| 1:14:00 | Coinbase board — watching Brian Armstrong navigate regulation |
| 1:17:00 | Avoiding well-trodden mistakes — make new mistakes |
| 1:17:30 | Going direct inside a company — product reviews as information flow |
| 1:19:00 | Identifying future leaders by who builds context fastest |
| 1:20:00 | Change reveals character — advantages accrue immediately, disadvantages over time |
| 1:22:00 | Programming's lost immediacy — "Have we made programming too hard?" |
| 1:23:00 | Computer science vs. craft — "We have significant science envy" |
| 1:25:00 | AI solving technical debt — entrepreneurship as on-ramp |
| 1:27:00 | Growing up with computers — "So much would change" |
| 1:30:00 | Benjamin Bloom's 2 Sigma Problem and parenting with mastery learning |
| 1:32:00 | Canada's future — "Make big plans, hold everyone accountable" |
| 1:34:00 | Canada's resources advantage for the next manufacturing age |
| 1:37:00 | Motor racing as feedback loop — technical, visceral, honest |
| 1:40:00 | Formula 1 as entertainment product — Liberty Media's savvy transformation |
| 1:42:00 | Authentic content in racing — managing while on camera |
Relevance to SupportWire & FeatureOS
- "Internal tools are culture" maps directly to how we build. Tobi's thesis that the tools you build internally shape behavior more than any values doc is a challenge: are FeatureOS and SupportWire shaping how our customers' teams actually behave? If the tool doesn't change workflows, it's furniture. If it does, it's infrastructure.
- Agentic commerce = agentic SaaS adoption. Tobi's prediction that AI agents will mediate commerce applies equally to how companies discover and adopt B2B tools. FeatureOS needs to be agent-readable: structured data, clear API endpoints, machine-parseable pricing — because the next wave of buyers might not be humans browsing your website.
- "Composable primitives, not monoliths." Shopify's shift from selling features to providing composable infrastructure is the same move SupportWire should make. Don't sell "a support tool" — sell primitives (ticket routing, knowledge retrieval, sentiment detection) that agents and workflows can compose.
- The taste gap is real and widening. Tobi's point about Midjourney scaling taste while most companies lose it is a direct competitive advantage for FeatureOS. Most feedback tools look and feel generic. If we maintain craft at the pixel level — every empty state, every interaction — that's a moat competitors can't easily replicate.
- "The world is a museum of passion projects." Distribution doesn't come from features — it comes from building something people are starved for. Tobi describes the "tsunami" that happens when authentic craft meets real demand. The FeatureOS changelog, the SupportWire onboarding — these are surfaces where passion either shows or doesn't.
Detailed Breakdown
[00:00] The Philosophy of Beautiful Problems
Tobi opens with his core belief: the best thing that can happen to you is finding a problem so rich and complex that it never runs out. Internet commerce is his. Even after 20 years, the problem keeps spawning "enlightened problem children" — sub-problems that are each fascinating in their own right. John draws the parallel to Stripe's experience with payments.
[07:04] GSD — Internal Tools as Culture
Shopify built an internal system called GSD (Get Shit Done) that manages how work flows through the company. Tobi's thesis: the tools you build internally are more powerful culture-shapers than any written values doc. How a company tracks, prioritizes, and ships work reveals what it actually values — not what it says it values. This section is a masterclass in why founders should care deeply about internal tooling.
[16:32] The Shopify Vision — Ambitious Merchants, Cutting-Edge Tools
Tobi describes Shopify's target customer: the ambitious merchant who wants to build a real business, not just dabble. Shopify has taken merchants from zero to a billion dollars in revenue — and they're still on Shopify. The goal is to be the commerce operating system that grows with the merchant, from first sale to IPO scale. The platform's job is to remove every piece of friction between "I have a product" and "someone bought it."
[27:49] Flash Sales, Supreme Drops, and Peak Capacity
A deep dive into Shopify's infrastructure war stories — handling flash sales for theCHIVE, Supreme drops, and Kylie Jenner's product launches. These events create traffic spikes that would bring down most platforms. Shopify treats them as engineering challenges to be solved, not edge cases to be tolerated. The capacity to handle these moments reliably is a competitive advantage that's invisible until it matters.
[34:36] Agentic Commerce — AI Agents as Personal Shoppers
This is the section that matters most for anyone thinking about the future of e-commerce. Tobi's thesis: AI agents will increasingly act as intermediaries between consumers and merchants. Instead of browsing a website, your agent browses for you — comparing prices, reading reviews, checking availability, and purchasing. This fundamentally changes how commerce works: discovery, conversion, and loyalty all get re-mediated through agents. The merchants who make their products and APIs agent-friendly will win. The ones who rely on human-only interfaces will lose.
[49:15] Shop Pay — The Quiet Compounding Machine
Shop Pay delivers a 16% conversion rate increase for merchants. Tobi frames this not as a feature but as compounding infrastructure — every percentage point of conversion improvement multiplied across millions of merchants creates enormous aggregate value. The bet is that owning the payment experience (rather than handing it off to a third party) gives Shopify control over one of the most critical moments in commerce.
[55:29] Stablecoins and the Future of Payments
Tobi is pragmatically bullish on stablecoins — not as speculative instruments, but as payment rails. His argument: the most important thing that needs to happen is expanding what can be purchased with stablecoins. Once that happens, stablecoins become a serious alternative to credit cards for certain transaction types, especially cross-border. This isn't a crypto-bro take — it's a payments infrastructure take.
[59:20] The Stripe–Shopify Relationship
Thirteen-plus years of partnership, described by Tobi as analogous to Apple and TSMC. Stripe handles the payments infrastructure; Shopify handles the commerce experience. Each company's success amplifies the other's. Tobi and John discuss how this kind of deep, long-running partnership is rare in tech — most companies either try to build everything themselves or treat partners as interchangeable vendors.
[1:04:02] Craftsmanship and Software Taste
Tobi riffs on what it means for software to have taste. He cites VS Code and Midjourney as examples of products where you can feel that someone cared. His observation on Midjourney: they managed to scale taste by owning a particular aesthetic — a rare achievement. Most companies lose taste as they grow because taste requires opinionated decision-making, and organizations default to consensus. The founder's job is to be the taste bottleneck that doesn't become a bureaucratic bottleneck.
[1:12:23] Coinbase Board Learnings
Tobi sits on the Coinbase board and shares what he's learned watching Brian Armstrong navigate Regulatory Battles. The key insight: watching another founder deal with a fundamentally different set of challenges (regulation, government relations) expands your own leadership toolkit. Board service isn't just governance — it's a masterclass in seeing how different types of companies and founders operate.
[1:17:23] Staying Ungovernable — Organizational Design
Tobi's philosophy on corporate governance: stay as "ungovernable" as possible. Meaning — resist the natural tendency of organizations to layer on process, approvals, and committees. The goal is to preserve the speed and judgment of a small team even as you scale. This doesn't mean no structure; it means every piece of structure must earn its place by making the company faster, not slower.
[1:25:02] Entrepreneurship and Programming as On-Ramps
Both founders discuss how programming and entrepreneurship serve as on-ramps to impact — especially for people who don't come from traditional backgrounds. Tobi references Benjamin Bloom's "2 Sigma Problem" (one-on-one tutoring is 2 standard deviations better than classroom teaching) and Kevin Kelly's "1,000 True Fans" as frameworks that shaped his thinking. The argument: the tools for building businesses have never been more accessible, and that accessibility is itself a force multiplier.
[1:31:50] Canada's Economic Future
Tobi offers candid thoughts on Canada's economy and what he'd tell Prime Minister Mark Carney. The gist: Canada has the talent and the resources, but the regulatory and cultural environment doesn't encourage enough risk-taking. Too many smart Canadians leave for the US. The fix isn't more government programs — it's removing friction for builders.
[1:36:40] Motor Racing as a Feedback Loop
The conversation closes with Tobi's passion for motorsport. He frames racing as a perfect feedback loop — you make a decision, the car immediately tells you if it was right or wrong. No politics, no ambiguity, just physics. It's the same thing he loves about programming and building products: tight feedback loops that reward good judgment and punish bad decisions instantly.
Notable Quotes
"The best gift in life is finding a beautiful problem that you can never solve. And even if you accidentally solve it, if you're so unfortunate to solve it, hopefully it has plenty of enlightened problem children."
"I built Shopify to avoid people having to use shitty software, not to add to the pile."
"Companies are technology by which you create social acceptance for people pursuing missions together."
"Some people fall in love with solutions. Some people fall in love with problems... very hard to know who is who."
"A personalized ad is a wonderful thing... Meta and Instagram have created more businesses than most government policies."
Books & Resources Mentioned
- Benjamin Bloom — The 2 Sigma Problem (one-on-one tutoring vs. classroom)
- Kevin Kelly — 1,000 True Fans
- Erich Gamma — Design Patterns
- Charles Calomiris — Fragile by Design
- Formula 1: Drive to Survive (Netflix)
One Thing to Act On
Make FeatureOS and SupportWire agent-friendly. Tobi's thesis on agentic commerce applies beyond e-commerce — AI agents will increasingly mediate how people discover, evaluate, and adopt SaaS tools too. The question isn't just "does my website convert humans?" It's "can an AI agent understand what my product does, compare it to alternatives, and recommend it?" Think about structured data, clear API docs, and machine-readable product descriptions now — before this becomes table stakes.
#shopify #stripe #commerce #agentic-ai #stablecoins #internal-tools #craftsmanship #founder-mindset #payments #taste
Raw Transcript
Auto-captions from YouTube. Folded by default — expand if you need to grep the source or pull an exact quote.
0:00 The best gift in life is finding a 0:01 beautiful problem that you can never 0:03 solve. And even if you accidentally 0:06 solve it, if you're so unfortunate to 0:08 solve it, hopefully it has like plenty 0:10 of enlightened problem children. 0:12 Consumerism is like a thing that like is 0:15 being thrown around. But where does it 0:16 come from? They throw away things 0:18 because they they hate the things they 0:20 that they have. The thing that solves 0:23 consumerism is quality products. A 0:25 personalized ad is a wonderful thing. I 0:27 am scrolling past something that's 0:28 monetizing the free application that I 0:30 appreciate using. 0:31 >> You should tell the EU this. 0:33 >> Yes. Well, I sure I do actually. Some 0:35 people fall in love with solutions. Some 0:37 people fall in love with problems. And I 0:39 I I just like fundamentally I'm a person 0:41 who appreciates. So people fall in love 0:42 with problems. But like it's very hard 0:44 to know who is who. Um because well if 0:46 you don't change anything they look 0:47 exactly the same. 0:50 >> This is the point you only have half an 0:51 Irish problem. 0:53 >> Oh well. This is our our MVP uh when we 0:57 weren't even sure that this would be a 0:59 thing. Toby Luca launched Shopify in 1:01 2006. We've been working closely with 1:03 them for over a decade and I always 1:05 enjoy getting together with Toby and I 1:06 always learn a lot. Cheers. Cheers. 1:11 >> Okay. Many place I could start. What's 1:12 the size of the code base in as much as 1:14 you know vendor stuff whatever? I think 1:16 the sort of core part of Shopify um with 1:19 all the uh like like minor stuff like 1:22 the identity system and these kind of 1:23 things is like 20 million lines of code 1:26 and then the uh TypeScript written admin 1:29 interface is like another I think 8 n 10 1:32 um recent it's apparently one of the 1:34 largest TypeScript applications and 1:35 definitely the largest Ruby application. 1:37 So they're pretty 1:38 >> we pushed it into pretty uncharted 1:39 territories for for for our tech stacks. 1:42 It does so much like it's like a full 1:44 fintech business in there with capital 1:46 and these kind of things and all the 1:47 work we are doing together and my 1:49 company adopts the complexity the real 1:52 real life messy complexities of the 1:55 commerce world into itself to then front 1:58 it with um approachable interfaces 2:00 things that uh don't uh you know just 2:02 making all the pieces fit together and 2:04 um so that that requires a lot of work. 2:07 >> So there's like thousands of projects 2:09 going on in any given time. So um going 2:12 through all of them is work 2:14 >> but we have a good system. We have like 2:15 a product operations team that prepares 2:17 for it. We have an internal system 2:19 that's just for the reviews which is 2:20 actually kind of cool. 2:21 >> What does the system do? 2:23 >> So I'm fascinated with companies. I 2:25 think companies are underappreciated by 2:27 for the 2:28 >> like just we don't think of that 2:30 generally. It's like they you know 2:32 companies are technology um themselves. 2:34 Um they are a technology by which you 2:36 create um part of what they create is 2:39 social acceptance for 2:41 you know people maybe tens of people 2:44 hundreds of people thousands tens of 2:45 thousand people depending um to spend 2:47 all their day all all their day pursuing 2:49 a mission together you know it's like 2:51 >> it's really just like universities 2:54 exist doing giving you time that you can 2:57 use for like hopefully intellectual 2:59 inquir in inquiry into some topic and 3:02 everyone's cool with that. Although if 3:03 you try to do the same thing, you know, 3:05 just with your laptop on the internet 3:07 all day long has not nearly the same 3:10 social acceptance 3:11 >> may also be valuable. And then companies 3:13 are ones that allow people a lot of 3:14 people together to uh you know try to 3:16 make a like pursue a mission or create 3:18 something hopefully world class. So um I 3:22 think copies are underststudied um and 3:24 um also the sort of incentive system 3:26 behind companies tends to be that um 3:29 they just there's no absolute need for 3:32 um uh perfecting them. There's just like 3:34 you've got to be better than everyone 3:35 else and at some point in many 3:37 industries like if you get ahead it's 3:40 really really hard for the others to 3:41 catch up and so so just there sort of a 3:44 point equilibrium at which point you 3:46 just 3:46 >> because you're good enough and so there 3:48 isn't pressure to get better beyond some 3:49 point. 3:49 >> Yeah. I mean I just I mean take an 3:51 absurd example I talked with some 3:53 executives in the gold mining space 3:55 right it's just like really sort of 3:56 interesting because I'm like what is 3:57 your day like like the same questions 3:59 like it's like much more interesting 4:00 >> what is a gold mining executive 4:03 >> it's just like I mean my conclusion 4:05 which was maybe what he was trying to 4:06 guide me to was um but it's essentially 4:09 government relations or investor 4:11 relations you know like if you're you 4:13 know like you know how much is in ground 4:15 you know sort of it's a capital play but 4:18 somehow if the investor does like you 4:20 better when um you get a higher multiple 4:22 at which point it behooves you to 4:24 purchase all other gold mines really 4:25 quickly. So I imagine this involves 4:28 playing golf that's probably with 4:30 technology there. So is one of the 4:32 conundrums for you as someone who thinks 4:33 about companies and organizations a lot 4:36 the lack of measurability of R&D which 4:40 is probably you know half of what 4:42 Shopify you know uh spends its money on 4:44 is R&D and so if you're running a 4:46 factory it's pretty easy to measure the 4:47 inputs and outputs and you know how many 4:49 widgets are we producing per hour and 4:50 what's the labor efficiency and 4:52 everything like that it's very 4:52 measurable whereas when you're building 4:54 software it's like I don't know we put a 4:56 bunch of engineers on it and hopefully 4:58 in like 6 months they have a great 4:59 product out there in the 5:01 But the intermediate measures yeah you 5:03 can measure like pull requests per 5:05 engineer per day or whatever and 5:06 everyone knows that's like a very broken 5:07 proxy. I had lots of thoughts on this 5:09 specifically I would love to actually 5:11 know how you think about this too. So 5:13 traditionally when you look back in 5:14 business history about when people got 5:16 like a little bit scientific about 5:17 business or like actually started to 5:18 improve them like clearly it all started 5:20 sort of vibes based and then you know 5:22 Frederick Taylor came up and uh you know 5:25 brought the wonderful new technology of 5:26 a stopwatch to the production line and 5:28 like um timed the various steps and um 5:31 of course uh produced this is I think 5:33 Bethlehem steel and end of 1800s and 5:35 drove massive efficiency gains and so 5:38 that's like the hero's journey of 5:39 business books right like it's like 5:41 >> and it's still so much of the corporate 5:42 culture today, right? You know, it's how 5:44 org charts tend to be structured. It's 5:45 how the financial statements are. You 5:47 know, it's what the kind of business 5:48 that financial statements are optimized 5:49 for is kind of, you know, a factory 5:52 >> in a world where almost no company was 5:54 terribly conscientious and and no 5:56 company was like metabolized all the 5:58 value available to them. 5:59 >> Yes. 6:00 >> Someone starting to hill climb some 6:03 source of uh you know efficiency gains 6:06 or or just like becoming better company 6:09 um was like a breathtaking change, 6:11 right? like once someone starts hill 6:13 climbing everyone has to or like gets 6:15 you know gets left behind. So um that 6:17 became the story and it's it's 6:20 remarkable to think that um that was 6:22 probably good enough for like probably 6:24 80 years or 90 years of like from this 6:26 point on right like it's like just like 6:27 >> it's a lean basically which is kind of 6:28 the 2.0 of that. 6:30 >> Exactly. So you said so the car industry 6:32 with exactly Toyota and so on. We just 6:35 there's some new ideas came came on but 6:38 they were all around um the sort of 6:41 drive for efficiency. 6:42 >> The downside of a drive to efficiency is 6:45 that it requires you to act on on on 6:47 quantifiables right like you you can 6:49 only measure what you can measure for 6:51 like sematically. So that's not the 6:54 entire like the space of how to make 6:56 better companies. It's like there's a 6:57 huge amount of things that are 6:58 unquantifiable, but taste 7:00 >> Mhm. like quality is is hard to 7:03 quantify. 7:04 >> Well, not only that, cuz those are the 7:05 obvious examples of uncquantifiables, 7:06 but the big one is I'm curious like 7:08 there's a team at Shopify that's doing a 7:11 great job of their product and there's a 7:12 team that's lost and has kind of a 7:14 woolly direction and actually their dev 7:15 tools suck and so it's very hard for 7:18 them to make forward progress and 7:19 they're thrashing if the lots of keep 7:21 the lights on work. just how do you 7:22 distinguish between the team that's 7:25 struggling and the team that's executing 7:27 really well? Yeah, I think I think 7:28 that's I mean this is why um I the 7:32 question my my answer is I should go to 7:34 I don't know. Here's what I found works 7:37 is um have rituals by which we talk 7:39 every um I mean at at the latest every 7:41 eight weeks, right? Um and so let the 7:44 teams uh talk about the progress against 7:46 the goal. It's like the new like we have 7:48 we have an internal system called uh uh 7:51 GSD getting [ __ ] done. Um it's a it's a 7:53 it's the central registering registry 7:56 clearing house. Um it's our wiki, it's 7:59 our like feeds and stuff like this, but 8:02 it's also has every project in it. And 8:04 so it's built around uh you know teams 8:07 updating everyone else registering like 8:09 here's something we learned. 8:10 >> And so like that was step number one. 8:12 Let's get all like get the actual state 8:14 of a business into a legible um internal 8:18 system by which you can reason about it. 8:21 where is every project? What's the 8:22 deadlines? What has changed? And so this 8:25 is what these reviews are about. So we 8:26 go like this is what we have on the 8:28 screen. They they get a little TL draw 8:30 that they can drag images and mockups 8:32 and everything into. But that's their 8:33 area. But like all the metrics and all 8:36 the things are like around the around 8:38 the edge on the screen right in front of 8:39 everyone and we can talk about what um 8:41 um that and how many people are on a 8:43 team and this kind of stuff. and giving 8:45 everyone an opportunity to just talk to 8:47 me um for even if it's a only quickly is 8:51 incredibly valuable and I just learned 8:53 tons of things about it and this is 8:55 something that's in the calendar and I 8:58 don't know it's exhausting and much a 9:00 lot of work but I don't know how to do 9:01 it in any other way because that is the 9:03 thing by which people can say hey it's 9:07 feels like progress is really well and 9:08 here's what we're building and here's 9:09 something we learned what should we do 9:10 about it is 9:11 >> is it like people want to get into 9:12 running the couch to 5k program start 9:14 with get off the couch, go walk down the 9:17 block, go walk down two blocks, you 9:18 know, and things like that. It turns out 9:20 that from an organizational point of 9:21 view, having a centralized source where 9:23 you track all the projects and you list 9:25 out your goals and you post updates 9:27 based on it like that sounds too simple 9:29 to work in the same way that like 9:30 getting off the couch is too simple as a 9:32 step to running a marathon, but it turns 9:33 out both are true. it. I think it's 9:36 honestly it's not a complex idea and 9:38 it's um it's it's extremely valuable for 9:41 15 other reasons other than it powering 9:44 the um uh these reviews but yeah it's 9:48 it's not ter it's it's not terribly 9:50 complex and it it works. No, we took 9:52 inspiration from GSD, your system as we 9:54 built out ours similarly. And having a 9:57 centralized 9:59 internal source of truth for projects 10:01 that are going on is 10:03 surprisingly 10:05 helpful and it feels like it shouldn't 10:06 be, but it is. 10:07 >> Yeah. You know, it's funny because so 10:10 again, it's an internal system. Shopify 10:11 has a culture of building internal 10:14 software. It's um I mean I tend to point 10:17 out Shopify itself started out as 10:19 internal software to um um power my 10:22 snowboard store. Um so um it would be 10:25 funny to not do this. And so I think 10:27 fundamentally uh like companies are all 10:31 pretty bad or um like all the companies 10:35 of today are pretty bad compared to the 10:36 companies we will have in the future. 10:39 And um I don't know how to how I would 10:42 possibly run Shopify without having GSD. 10:44 And I think there's going to be some 10:46 productizations of GSD that'll be 10:48 available to everyone. It's probably 10:50 going to be renamed 15 times or whatever 10:52 become a category of software that's 10:53 going to exist. 10:54 >> And then um everyone's going to be 10:56 wondering how the hell did we build 10:58 software before we had it. 11:00 >> Yes. 11:00 >> And I try to have Shopify live in that 11:02 particular state um a little bit 11:05 earlier. So, 11:06 >> has your proclivity for developing 11:08 internal software ever led in just like 11:09 a really hilariously overbuilding 11:12 internal software direction? 11:13 >> Oh, yeah. Absolutely. All the time. 11:14 >> Like what's the most ridiculous 11:15 internally developed piece of software? 11:17 >> You have like a a static sort of or 11:20 internal geiosities like it's like 11:22 static site hosting which is like like 11:24 it's the most incredible and like um 11:27 Yeah. Like it's like a like '90s web um 11:30 um with all the pros and cons. 11:32 >> Mhm. 11:32 >> Yeah. like it it's it's I mean it it 11:34 send us down wrong paths like it's it's 11:36 very often I say internal tools culture 11:39 this is actually a little bit of a 11:40 dangerous statement I shouldn't make it 11:42 like as such we have a 11:43 >> um very strong uh appreciation of good 11:46 software and and good in the context of 11:48 we have to use it like some software is 11:51 really good like if you like Shopify is 11:53 really really really really good 11:54 software if you're um running like a D2C 11:56 business or with products 11:58 >> some people uh try to stretch it to um 12:01 subscription based groceries 12:03 stores, something I maybe that even 12:06 works, but I we clearly didn't build it 12:08 with a rest in mind. And I actually, 12:09 although I appreciate all my customers, 12:11 I'd rather not you anyone use the 12:13 software if it's bad for them because 12:15 that's like I I built Shopifies to avoid 12:19 people having to use shitty software, 12:20 not to add to the pile, right? So So 12:22 sometimes this means given our scale and 12:24 our our minimum quality bar, uh that we 12:28 have to build it ours. And in those 12:29 cases, we do that. Sometimes we build it 12:31 in where we have no business doing it 12:33 and that's where things go sometimes 12:34 wrong. 12:35 >> Well, what I think is interesting is you 12:36 think it's important to build your own 12:38 HR software and you would recommend 12:40 people at least think so about that. 12:42 Maybe you can share the worldview there. 12:44 >> Not sure exactly recommend it. Um but we 12:47 have ideas about how to do HR in terms 12:49 of compensation and so on that they're 12:50 just different from what's normally 12:52 implemented. This is most true in 12:55 project management software like again 12:57 GST. If you use software by others, you 13:00 you have to buy into their vision. Like 13:02 some vendors 13:05 tell you the software can do absolutely 13:06 everything and and that should 13:08 >> software have a worldview. So you're 13:09 adopting workday's worldview. 13:11 >> That's right. 13:11 >> When it comes to your HR, which may or 13:13 may not be what you want to do. 13:14 >> That's right. And and I think that's um 13:16 that's important to realize and it's 13:18 important to use to your advantage. Um 13:20 and and and I think this is why it's 13:21 really really important to buy software. 13:23 like the people who buy the software, 13:24 other people who should use the software 13:26 in the end because they have a better 13:27 view of what needs to be uh what needs 13:29 to be had the um so in in in our case in 13:33 HS we we had some um uh things that were 13:36 not doable with um workday and obvious 13:39 systems and so we built it ourselves. I 13:40 mean this is a this led to a journey 13:44 that made might have gone a little bit 13:46 too far on um how should this be done. 13:49 This is one of the areas where like 13:50 probably Shopify is the weirdest bit of 13:52 software ever made. But like 13:53 >> but is it something like the internal 13:56 systems you use will subtly affect the 13:59 decisions that you make? Everyone's 14:01 probably been in meetings where it's 14:02 like, oh well, we can't do that because 14:04 the system doesn't support it. And so 14:06 again, do you want to have a workday 14:08 designed compensation system? Probably 14:11 not. You like you wouldn't frame it that 14:12 way. Uh which is why again it's 14:14 important that people think about yeah 14:16 what is the vision they want? Uh, and 14:18 does the software? 14:19 >> I'm to make maker through I the the pit 14:22 in the land parties of the 90s that set 14:24 up the internet sharing for everyone, 14:25 right? Like then that was still 14:26 difficult. I just like I'm I'm a tools 14:29 uh tool maker infrastructure thinker in 14:32 my entire life. And um I deeply believe 14:34 in environments um that uh um cause 14:38 people to accomplish bigger and better 14:40 things than what they even imagined they 14:42 could have done. Um, Marshall Mlan says 14:44 like the first we make the tools and 14:45 then they they they shape us or 14:47 something along those lines. I have a 14:48 terrible memory for for for exact 14:50 quotes. And um but the sentiment is 14:52 right. And so I I think this is powerful 14:56 and is something I want to channel at 14:58 every level. Uh again, I want I want the 15:01 people who have a utilitarian problem of 15:04 hey, I need an online store. 15:05 >> Mhm. to accidentally catch Shopify 15:07 because it's the go-to software, but 15:09 then Shopify to inspire them to build 15:12 much bigger and better companies than 15:13 they thought they did and like elevate 15:15 their own ambition in what they are 15:17 building just like my ambition increased 15:19 from building snowboard stores to uh 15:22 helping millions of businesses be built. 15:24 And so um it's it's the environment that 15:27 I think we have a lot more um power over 15:31 and um it's it's the sort of missing 15:33 ingredient people talk about you know 15:35 people talk about incentives shaping 15:37 companies and then policies shaping 15:40 companies right so those are the tend to 15:41 be the two tools and I I think sort of 15:45 generally like if people are really 15:47 insightful they they talk about culture 15:49 as well but um it's the environment 15:52 that's even more powerful than all those 15:54 And it tends to be as a software 15:56 company, especially if you feel like you 15:57 have agency over the tools, that you 15:59 actually have more power over the tools 16:00 and therefore the environment than you 16:02 have over policies like because that 16:06 those can be moving forward by the by 16:08 the speed of a deploy like um you know a 16:10 new change to to the GSD software can is 16:12 immediately part of environment where a 16:14 roll out of a new policy is going to be 16:16 a conservative uh town hall and uh you 16:18 know a lot of um convincing. So in in 16:21 funny ways it actually acts as the as a 16:24 fast and the most effective way of uh 16:26 evolving a company forward 16:28 >> having opinionated software that you use 16:31 to tweak the environment. 16:32 >> That's right. So the obvious question is 16:34 if software shapes us and kind of shapes 16:38 the actions of organizations 16:40 um and you should think about what the 16:42 vision embedded in software is. What is 16:45 the vision of Shopify for your 16:48 merchants? I mean you said one part 16:49 which is that they should be more 16:50 ambitious for uh their business success 16:52 and they can probably be more successful 16:54 than they might realize. What else? 16:56 >> What I tend to visualize is like what 16:58 are the things you know um if um you you 17:02 you're sitting um with a friend over 17:04 drinks in a Irish pub and um 17:07 >> I'm with you so far 17:10 >> and the question is like hey um you've 17:13 done an online store, you've done an 17:14 e-commerce business um you've done a 17:16 retail business with Shopify. would you 17:18 do this again? But the answer to the 17:20 question is like hopefully hell yes 17:22 because and then there comes a list of 17:24 things that we that we would like to uh 17:26 do like amongst those is like it just 17:29 makes it easy like it it allows me to 17:31 build the business not feeling like I 17:34 have like you know when you try to paint 17:36 a painting and you have like mittens on 17:38 like it's like it's just like this is 17:40 sort of a you know there's perfect 17:41 vision in your mind but you can't get it 17:43 out because the tools or 17:44 >> we've all used software like that. 17:45 >> Yes. Exactly. So like another thing is 17:48 that um it it it just kept it keeps me 17:50 cutting edge right like it's we going 17:52 through an enormous platform shift like 17:54 Shopify predates smartphones right like 17:56 a software so we have seen multiple 17:58 platform shifts and and and usually what 18:00 happens when you study any kind of um 18:03 industry uh upheaval or like a recession 18:06 or anything it's hardest on uh like the 18:10 small businesses the really big 18:12 businesses get bailed out anyway and 18:14 everyone in between has usually the line 18:16 of credits or whatever capital to to to 18:18 make it through. 18:18 >> You're talking about an economic 18:19 recession. 18:20 >> Economic recession. So like you 18:21 something happens in the world, the 18:23 small businesses are the ones who are 18:25 wiped out. The the failure rates during 18:27 um recessions is enormous and that's a 18:30 huge drag on economies because like 60 18:32 to 80% of all people who work work for 18:35 small businesses. I I think people felt 18:37 this very viscerally in uh co with 18:40 restaurants which were obviously 18:41 particularly hard hit by co but I think 18:43 people could see how little uh how kind 18:46 of shallow the balance sheets were 18:48 because entire you know streets turned 18:50 over with new storefronts like it was it 18:52 was a very visual representation for 18:54 people of how fragile a lot of the small 18:55 businesses are. 18:56 >> Yeah. And like I mean many many places 18:58 never recovered. It's like it's um it's 19:00 a very organic process by which an area 19:02 becomes um like sort of have the 19:04 critical mass of interesting stores that 19:06 like then have has people come there and 19:08 that just like this builds up over very 19:09 very long periods of time. The entire 19:11 entire city's centers can shift in in 19:14 these times and so it's an very 19:16 precarious um balance and um they they 19:20 have very uh tight margins. They they 19:22 often like there's there's so much that 19:24 makes them fragile by like 19:26 systematically 19:27 >> that at least um all their sales 19:31 disappearing because suddenly most 19:34 people purchase from their mobile phones 19:36 uh that 19:37 >> just their website doesn't work on 19:40 >> that's it's not their job to stay 19:42 current with these kind of things. So 19:43 like just like inoculating from all 19:44 these kind of shifts is is a really 19:46 important part of a business. 19:47 >> So you're talking about some of the 19:48 disadvantages that small businesses 19:50 have. I was realizing recently, it 19:52 actually really hit me when uh I went to 19:55 buy something from a very large famous 19:59 uh brand name redacted to protect the 20:01 guilty but um very big, very storied. 20:04 And I was going through the e-commerce 20:05 experience and it was horrendous. It was 20:08 so bad. And I realized that thanks to 20:10 Shopify, we are now I don't think it's 20:12 hyperbole to say we're living in an 20:14 inverted world where the very large 20:16 retailers are worse off in their 20:18 e-commerce experience. They have, you 20:19 know, worse systems and lower converting 20:22 websites than small businesses which 20:25 have these like amazing super snappy and 20:27 they're just like more technically 20:28 performant if you measure the latency if 20:30 you measure the you know full funnel 20:31 conversion and everything like that on 20:32 any metric by which you know if you were 20:34 with your stopwatch and you went from 20:36 Bethlehem Steel to you know the internet 20:38 and we're kind of measuring things with 20:39 your stopwatch the small business are 20:40 better off. one I thought that was just 20:42 interesting that in this case we have 20:44 this inversion where on a very tangible 20:47 technical level the the Rebel Alliance 20:49 is doing better than the uh the large 20:50 established companies but then it made 20:52 me wonder there's kind of a question for 20:54 Shopify where you have succeeded on the 20:56 small business side and I know that's 20:58 probably an to you it's like oh no we're 20:59 just getting started blah blah blah but 21:01 you have very significant fraction of 21:03 the small business e-commerce market 21:06 with a very successful product there and 21:07 so how do you think about Shopify 21:09 expansion from here where you You can go 21:10 international, you can go up market to 21:12 large companies, you can go into new 21:14 modalities like Aenta Commerce and like 21:15 there's many vectors of expansion, but 21:18 it does feel like there's a take stop 21:20 moment because step one is deliver a 21:22 great experience for small businesses. 21:24 >> Yeah, thank you. Um, by the way, me too. 21:26 I I I also have purchased from Nike. So, 21:30 >> it wasn't Nike, but yeah, those kinds of 21:32 people. 21:34 >> Um, yes. uh you know that I I I love the 21:38 this inversion because the inversion is 21:40 super this is one of very few spaces 21:42 where this is true right like and um 21:44 that is exactly what we set out to do 21:47 and it's it's an access to the capital 21:49 market 21:50 fundamental belief is like people should 21:52 use amazing software that really really 21:54 um like is like perfectly fit to the 21:58 problem they have 21:59 >> you don't blow 22:00 >> exactly but uh we can go deeper and 22:02 deeper and deeper deeper into into the 22:04 space right So it's just really really 22:06 quick. It's really easy to underestimate 22:08 the size of everyone underestimates the 22:11 size of the internet and everyone 22:12 underestimates the size of retail. We 22:13 sit in an intersection of both those 22:15 things, right? Like I guess my favorite 22:16 stat is that the um um the the companies 22:19 um the industry supplying wooden pallets 22:23 to warehouses is a $60 billion a year 22:26 industry, right? Like just like that 22:28 gives you sort of a um idea for for for 22:31 how how large retail actually is. um um 22:34 at least directionally. And so what we 22:36 have uh done from a company perspective 22:39 is we were very very clear that our 22:41 mission is to to to make 22:43 entrepreneurship simpler. 22:44 >> Um and that's where uh you know like our 22:47 heart is um and um many of our customers 22:50 have gotten quite big like we we ended 22:52 up racing one of our customers to a 22:54 billion dollars of revenue. Um um and um 22:57 >> so so there's some some of our customers 22:58 have like in the top 10 of largest 23:00 customers many of them have started on 23:01 Shopify. the reason why they're still 23:03 around is because we actually again we 23:05 take this so seriously that we're 23:06 actually building with them while they 23:08 grow and um um you know it turns out 23:11 >> again like all these stories of software 23:15 is for a certain segment of a market are 23:17 skills issues and and that's exactly 23:18 what they sounded like when I when I 23:20 first heard about it. It's convenience 23:22 to the company making the software to 23:24 say we are focused on one segment and by 23:26 the way we're always segment like 23:27 focused on the richest segment funnyly 23:29 enough 23:30 >> and we just said no we are like starting 23:32 with SMB and then you some of them grow 23:34 and hopefully build good software and 23:36 then we invite the others over. The most 23:38 common thing that people said with 23:40 Shopify in the early days was always 23:41 that this will never work in a real life 23:43 because like in the real world retailers 23:46 are messy and you know just like have 23:48 all these uh business rules that we 23:51 didn't implement and our opinion was 23:54 always like well the real world sounds 23:56 like a terrible place. we like ours 23:57 better and we just invite everyone over 23:59 and 23:59 >> well isn't there also something where 24:01 like the real world is messy but we find 24:03 this with stripe billing where people 24:05 say oh everyone's billing system is 24:06 different and like the dimensions of 24:09 their billing that they want to encode 24:11 and and things like that. Therefore, you 24:13 know, there's a certain size beyond 24:15 which, you know, stripe building will 24:16 cap out and people need a custom system. 24:18 And our view is absolutely billing is 24:21 really messy and it's also hard to test 24:23 because it involves time travel, right? 24:24 It's like, you know, what happens if 24:26 people do a mid period adjustment that 24:28 needs to be prorated, whatever. 24:29 Therefore, do you want to be using your 24:32 homegrown version of this that was 24:33 developed by some guy who had never 24:35 built an e-commerce or billing system 24:37 before and has since left the company 24:39 and, you know, now it's kind of poorly 24:40 maintained? Or do you want to buy it 24:42 from people who have been thinking about 24:44 this problem space and building a 24:45 generally abstracted frame and you know 24:47 our stripe billing now has a time travel 24:49 engine where you can just move time 24:51 around to test all these kind of 24:53 different cases because obviously you 24:54 need that and in Shopify's case 24:56 similarly yeah we've been thinking about 24:58 inventory management for a very long 24:59 time and you know how you model 25:01 inventory and all these kind of things 25:02 and so I feel like it's funny that 25:03 people sometimes say oh this will never 25:06 work at the large end because our needs 25:08 are super complex therefore we should be 25:10 using a piece of software that only has 25:12 one user and is poorly maintained by you 25:15 know one tiny corner of the company. 25:17 >> Yeah, that's right. I think um look when 25:19 I when I first um started programming um 25:22 in my apprenticeship um my me told me um 25:26 that in the software world you have um 25:31 about 2 years time anytime you start a 25:34 important piece of software to try to 25:35 nail it try to nail the problem main 25:37 because afterwards it just magically 25:39 like someone puts them in the codebase 25:40 and you're never going to change a 25:41 thing. 25:41 >> Yes. 25:42 >> And that's just like that's the state of 25:43 like software in the '90s. Um and um 25:46 frankly just the way the software works 25:48 like 25:49 >> everything that got to scale was like 25:51 finished in '90s with cement in it 25:53 >> and then sold in the 2000s and in a lot 25:55 of decision makers life that was the 25:57 time when software adoption really 25:59 started and their priors are loaded up 26:01 with software that corresponds to 26:03 roughly what my uh what Jung back then 26:05 was telling me. Now we've gotten a lot 26:07 smarter about building software. We have 26:09 you know CI and automated testing 26:10 general and like we just like we know 26:12 how to build software in a way we by no 26:15 means perfect at it and we don't even 26:16 know how to measure it as we talked 26:17 about earlier but like state of art has 26:18 increased significantly and we make 26:20 better software and that's a completely 26:21 different thing than what people are 26:23 comparing it again 26:25 >> but it's hard to tell them right like 26:27 it's you know the first 80% of every 26:29 e-commerce business are the same. It's 26:31 like you need a checkout, you need 26:32 credit cards of course and um you're 26:34 taking credit cards for us and um you 26:36 know like you need the hosting the 26:38 checkout and office got fix and an admin 26:40 interface. So it's the last 20% are 26:42 actually quite unique and and and they 26:44 exist in a particular very predictable 26:46 problem domain um like what does the 26:48 website look like? It's like what my 26:49 brand looks like and um also business 26:52 rules like we have over time uh just 26:54 like your time traveling uh um back test 26:57 system figured out where do we allow 27:01 programming or programming enabled uh 27:03 extensibility in the system so that 27:06 these last 27:07 >> 20% can uniquely be created for everyone 27:10 who wants them and and I think 27:12 >> um that's what it means then I think I 27:15 think that's a place you can only get to 27:17 >> if people work on a problem really care 27:20 because like you kind of have to fall in 27:22 love with a problem to want to look at 27:23 it from so many sides that at some point 27:25 you understand it well enough 27:27 >> to um not solve the problems that people 27:30 state but build infrastructure in a form 27:33 of primitives that can be composed to 27:36 solve all the problems and that's a 27:37 completely different level but it's what 27:39 I I deeply appreciate I I 27:41 >> massively admire it every single time 27:44 any category gets to this uh quality of 27:47 decision making and infrastructure. 27:51 Another thing I enjoy on the um software 27:55 quality side that both Shopify and 27:58 uh Stripe sell on that's just a a funny 28:02 domain to pursue is peak load where uh 28:06 stripe sells on just you can put a lot 28:09 of transactions through the system at 28:10 once and obviously Shopify similarly it 28:12 feels to me that how you displace other 28:13 e-commerce systems is if you are getting 28:16 very spiky sales you can get you know 28:19 put a thousand requests per second 28:20 through the system and that's totally 28:21 fine and most systems just at a at a 28:23 technical level can't do that. I guess I 28:26 found it funny because I think we kind 28:27 of grew up together kind of expanding 28:30 the um you know requests per second uh 28:32 thresholds and uh and you guys pushed us 28:34 pretty well there. But I'm like where 28:37 did this come from in e-commerce 28:39 culture? Because like the marketers 28:40 saying everyone will come to the website 28:42 and click you know buy at the same time. 28:43 The engineers are like no please do not 28:45 stop encouraging that. But it's become 28:47 like a real part of e-commerce culture 28:48 like the drops and things like that. 28:50 Maybe you can fill me in on the drop 28:51 history because you seem to have started 28:52 it. 28:53 >> Yeah. Like it's it's really funny. It's 28:55 I this I'm I'm just as amused by this as 28:58 you are because it's like no engineer 29:00 would ever suggest this, 29:00 >> right? Because it's it's it's it's a 29:02 it's a very very strange thing because 29:04 but but but it works with fantastic 29:06 business build 29:08 around it. The one that did it for us 29:11 first um is like some somewhere in 2010 29:13 was a 29:15 >> um this is called the the chive which 29:17 was like like some I mean it's a website 29:20 it's like some community 29:22 >> and had like some t-shirt with Bill 29:24 Murray on it 29:26 >> and that thing every single time it went 29:28 on sale it took Shopify on. It was just 29:29 like unbelievable. We actually couldn't 29:32 we didn't know the demand level because 29:34 we just couldn't observe it like like 29:37 upstream the routers could not send 29:40 enough traffic and it's like 29:42 >> first of all we should try we had to 29:43 like like catch up very quickly and um 29:46 so it was one of those 29:48 >> things where um uh there was a very 29:50 specific decision moment where um we 29:52 realized I mean this is obviously we 29:54 shopify down is the worst possible thing 29:56 like this is like every like the entire 29:57 company exists to never make it make it 29:59 so that anyone went down no less the 30:01 entire thing. Um so um we were like well 30:06 the traditional business thing is like 30:08 this is super expensive from reputation 30:10 from money and we can't even in our 30:12 business model make ever make money back 30:14 from the chive given what the amount of 30:16 resource they consume. Um so so we 30:18 clearly need to fire the customer and we 30:19 decided no absolutely not we using this 30:21 as a as a gem and um we're going to use 30:24 them like we're going to build the 30:25 system in such a way we can deal with 30:26 this because this is exactly what good 30:28 engineering looks like. 30:30 And um that was for tutors uh because um 30:34 uh I think the sort of pattern ended up 30:36 start like really sticking. It was like 30:39 Supreme which really drove this further 30:41 the street wear world. 30:42 >> I remember Supreme and I remember Kylie 30:44 Jenner those like Lipkits I don't really 30:47 just that that's for embedded in my 30:49 memory from kind of 2013 2014 Stripeish 30:53 that that was definitely like that was a 30:54 board level topic. 30:55 >> Yes. And we had we had we had board 30:57 level discussions and our side too. It's 30:59 actually would be funny to to to know 31:01 the diff we probably sweated exactly the 31:03 same amounts of on both sides because um 31:05 obviously we have an incredibly 31:07 long-standing and I think at this point 31:10 storyried partnership. It's tremendously 31:12 difficult to um uh synthetically uh test 31:18 the all the pieces that go into commerce 31:20 because it's it's just like so complex 31:22 of so many things. There's so many APIs 31:24 there. so much we have to like the 31:26 shipping to of course payments and 31:29 payments is like depending on the mix of 31:31 bins it's like a different uh profile of 31:33 on lot right um Kylie Jenner like we 31:36 store off a lot more debit cards and 31:38 that just kind of exercises different 31:39 parts of the of the infrastructure then 31:41 and so anyway these kind of these kind 31:43 of things you find out and um um in and 31:47 on our side we just like 31:48 >> yeah like no sin uh was forgiven like 31:52 every single sin like was uh revealed 31:54 and 31:55 found us to be um wanting very publicly. 31:58 So 31:58 >> is there anything particularly 31:59 technically hard to scale? 32:01 >> Well, I mean so if you if you open a um 32:04 uh book on databases like sort of page 32:06 probably chapter one is like uh an 32:09 example of how a transaction works is 32:11 like I mean taking either moving money 32:14 on a ledger or uh uh moving an item out 32:17 of inventory and into an order, right? 32:19 Like 32:19 >> that's the textbook definition of 32:21 collectionality. 32:22 like two textbook definitions in a row. 32:25 Um so you are fundamentally have to 32:28 serialize at those points. This is like 32:30 the lock contention moment. Um there's 32:33 architectures which we employ that like 32:35 you can get around this. There's like 32:36 pools of 32:38 >> products like we we take products out of 32:40 an integer into like a like a like a 32:42 striped thing and like there's all sorts 32:45 of things that um do and then try to 32:47 refill this as fast as possible. But the 32:49 point is that um you you really deal 32:51 with log contention and lo contention is 32:52 difficult. You you 32:54 >> it's it's very very hard to um uh scale 32:57 if everyone 32:58 >> at the same time wants to decrement an 33:00 integer in the database just actually um 33:02 then then also the transaction involves 33:04 having to go to interchange and move 33:06 money. I was reading some histories of 33:08 Oracle recently and but I think it's 33:09 kind of interesting because um uh you 33:12 know proprietary databases were the 33:14 leading databases like the LAMPstack 33:16 only really took over in the 2000s but 33:18 in the 1990s if you were like building a 33:19 thing you actually needed to go buy your 33:20 database and there was these proprietary 33:21 database wars um in the ' 70s and I was 33:24 wondering are there lessons for us for 33:26 the kind of current open source or so 33:28 current AI wars uh and model wars from 33:30 the the database history but it was 33:32 striking talking about kind of the early 33:34 feature wars of Oracle versus its 33:36 competitors and locking uh was such a 33:40 big differentiator in those early 33:41 versions and like role level locking 33:44 level 33:44 >> but it came in like Oracle six or 33:46 something you know I mean you would 33:47 think it's like an Oracle 2 uh uh uh 33:50 feature but no it took them a few revs 33:52 to get there and row level locking was 33:55 like the bee's knees and they released 33:57 it 33:57 >> yeah and uh very glad they did 34:01 >> so I mean yeah I mean this is predates 34:04 Shopify a little bit but like um Yeah, 34:06 com tech company creation was like um 34:10 day one you send a million dollars to 34:12 Larry to get your Oracle database and uh 34:15 I think uh you know day like 180 you 34:18 public or something like very different 34:20 time from today. 34:21 >> Yes. I mean I guess we're kind of 34:23 getting back to that a little bit in AI 34:25 where you know um uh uh step one send 34:28 $300 billion to Oracle for a data 34:30 center. 34:30 >> Yeah, you're you're you're a couple days 34:32 past 180 on your journey to becoming a 34:34 public company I have to say. like 34:36 >> Yes. Yes. Um I want to Anyway, moving 34:39 on. Um I want to talk about Aenta 34:42 Commerce. I had a really nice shopping 34:44 experience recently where I wanted to 34:47 buy a integrated travel adapter and 34:50 power brick and so I did some product 34:52 research in chat GPT and it was really 34:56 nice. There was no uh ads getting in the 34:59 way or anything like that. Actually 35:00 found a product that I hadn't come 35:01 across from Momax. It was really good. I 35:03 think I was telling you I was excited 35:04 too. Okay. Yeah, I was speak And what's 35:06 the verdict? 35:06 >> It's very good. 35:07 >> It's very good. Okay. Yeah, we'll put a 35:08 link. And it kind of l me out and and I 35:10 purchased obviously super easy, one 35:12 click with shop and everything like that 35:14 and it felt very futuristic to me both 35:16 in the product discovery and the the 35:18 completion side of things, 35:20 >> but it feels like filling out web forms 35:25 is not a value ad activity um for 35:29 people. Like when you travel, you 35:30 probably don't directly book the hotel 35:32 yourself on the website. you know you 35:33 have an assistant who does it or 35:34 something like that because choosing the 35:36 hotel might be desirable for you but 35:37 actually filling out the web form is not 35:39 a value ad activity for you and so what 35:42 is your vision for how agentic commerce 35:45 happens specifically because it seems 35:47 relevant to your interest 35:48 >> it's relevant to my interest I mean it 35:50 will be extremely valuable and it will 35:51 be something that that that um will be 35:54 done a lot and might even be like a 35:57 majority of commerce or like on on the 35:59 internet I don't know um but like uh I 36:02 think my role in this is going to be 36:04 like largely into infrastructure 36:06 providing, right? Like I I um uh this is 36:09 the our job is to keep our like all the 36:12 millions of business of Shopify 36:14 extremely current. And so um 36:16 >> um what um we want to do is make sure 36:18 that like everyone's plugged into the 36:21 various chat uh systems and there's an 36:24 >> an MCP connected to every Shopify store 36:26 and uh you know global catalog that's 36:28 really nicely presented. So it's like um 36:31 we we're building software specifically 36:32 for um the open AIS and the clouds and 36:36 the the um complexities to make it 36:39 really really easy for these products to 36:41 show up and reason about them and um and 36:43 so because we think it's really really 36:45 important that they can show up. So like 36:46 there's an infrastructure angle in it 36:48 but like how it's being used. So I think 36:51 the best way to predict the future in 36:52 most cases is just like frankly look at 36:54 what rich people 36:56 buy for themselves, right? like you know 36:58 Uber is a wonderful company is also 37:01 extremely predictable because switch 37:02 people had car service and uh you know 37:04 so it's figuring out a way to scale that 37:07 in such a way from cost basis to so that 37:09 everyone can use it is just like yes 37:11 that works. So uh personal shoppers 37:14 right these are the places where uh you 37:16 know memory and understanding people and 37:19 personalization are just like pay so 37:22 many dividends and are so clearly also 37:24 good like it's like people have make so 37:27 much out of like data and and 37:29 advertising you know a personalized ad 37:31 is a wonderful thing right like if a 37:33 personal ad that tells like 37:35 >> you know I am scrolling past something 37:37 that's monetizing the free application 37:39 that I appreciate using 37:41 >> and The fact you told the EU this. 37:43 >> Yes. Well, I I do actually. And and and 37:46 the fact that it's highly personalized 37:47 to me means like at this moment where 37:50 this platform monetizes because they're 37:52 giving me a gift here that I get to use. 37:55 >> It's win-win. Yeah. 37:56 >> It's it's win-win using this moment also 37:58 that is now not just like a [ __ ] like 38:01 Ford 150 truck or I don't even know if 38:04 that's how you say that, but like it's 38:05 like I'm I'm not in the market for 38:06 trucks, right? So I uh uh instead I see 38:09 a travel adapter like the one you 38:10 usually send me you know like which is 38:12 like 38:13 >> I am your personal shopper in this case. 38:15 >> But I I I the funny thing is you send me 38:17 the link and I if I would put that link 38:19 in chat GBT which has my uh sort of you 38:21 know lots of my history and ask 38:24 >> is that the kind of product that I would 38:25 appreciate it would like just return yes 38:28 right because it's like it solves the 38:29 problem in its entirety. It's like a 38:31 travel adapter which like for any 38:32 country to any plug to any other um all 38:36 sorts of stuff good good product right I 38:38 want it 38:38 >> okay so we're getting down the road here 38:40 I mean it feels like step one is that's 38:43 going to happen with agentic commerce is 38:45 I feel like existing aggregators if just 38:48 aggregation is their raise on debt then 38:52 they're going to have a lot of uh 38:55 questions that they need to answer 38:57 because chat GBT and the AI apps end up 39:00 as a new aggregation point. And so it 39:03 seems to very much benefit the tale. It 39:05 benefits all the Shopify merchants just 39:07 like, you know, we had a new set of 39:09 content creators emerge on YouTube that 39:11 effectively compete with kind of TV. 39:13 They're both just entertainment and the 39:15 tail becomes much more powerful when 39:17 you've over-the-top distribution through 39:18 YouTube. Similarly, it feels like once 39:20 you have over-the-top distribution 39:22 through AI apps, it becomes you have a 39:25 much more powerful position as a tail 39:28 merchant because you know MOMAX is a 39:29 tiny brand uh but they can just get 39:31 recommended as the best product. So 39:33 that's step one that I think happens. 39:36 >> I think a slight modification I would 39:38 just make to this. I don't think the 39:40 tail part matters. I think merit 39:42 matters. It's like it's just like it's 39:44 it's kind of an important point because 39:46 like 39:47 >> you know often when you talk a lot about 39:48 commerce and products 39:50 >> at some point just because of 39:52 >> prior discussions people are like 39:54 >> yeah but aren't you just feeding 39:55 consumerism right like and and so so 39:57 this is sort of like the negative take 39:59 right and and again consumerism is like 40:01 a thing that like is being thrown around 40:03 but where does it come from? It comes 40:04 from people throwing away a lot of 40:06 things that they probably shouldn't. 40:08 >> But they throw away things because they 40:09 they hate the things they they have, 40:12 right? They these things don't work. 40:14 They they fall apart. They don't do the 40:16 thing. They're like cheap, plasticky, 40:19 this kind of thing. 40:20 >> The the thing that solves consumerism is 40:24 quality products, which you want to keep 40:26 using, right? Like you want to be 40:27 excited about this and like you want 40:29 this travel ad to last forever. And so 40:31 you never have to think you you 40:33 literally never have to Google what plug 40:36 is being used in Singapore, right? Like 40:38 and you just know that the thing that 40:39 you have of you can deal with everything 40:41 to everything. Um so problem solved. And 40:44 so um um so that's I think I think 40:46 that's that's an important modification 40:48 on on on point. Yeah, I guess implicit 40:52 in my view is that there's good stuff in 40:54 the tail whereas the head has 40:56 distributional advantages and so it is 40:58 more meritocratic to bubble more stuff 41:00 up from the tail. That is I think the 41:01 reconciliation of those two worldviews. 41:03 Okay. So that's kind of step one that 41:04 happens. I think you're getting to step 41:06 two which is 41:08 >> it feels like there's a lot of changes 41:10 potentially that can happen around 41:12 personalization about agents actually 41:15 acting for you. And maybe one thing I'm 41:16 curious if you have thoughts on it. Just 41:18 seems like a lot of search in e-commerce 41:19 is really bad where it's keyword based 41:22 search still and you can't do things 41:24 like let me know if this becomes 41:26 available in gray or uh do you what 41:31 other products do you have like this and 41:32 you want some kind of embedding space 41:34 that thinks about similarity between the 41:35 products and no one can do this. So when 41:37 are we getting better personalizing and 41:39 search I just want really good search of 41:41 the products out there. I I this I I'm 41:43 obsessing about this now and like we we 41:45 have like we should have been 41:46 >> tell the secret Shopy robot. 41:48 >> Yeah, we Shopify should have solved this 41:50 earlier and hasn't and I I'm I'm pretty 41:53 mopy about this. Um I uh like you guys 41:56 are okay. 41:57 >> We we yeah um we will I I think we will 42:00 solve this now. So like the thing is 42:02 that um so just it's at a really really 42:05 fascinating I really deeply appreciate 42:07 it as a field. It's like so different. 42:09 It has its own has its own lore. It's 42:12 like everything is interesting is my 42:14 fundamental opinion and search has been 42:16 a wonderful area to like actually learn 42:19 a lot more about the history now and 42:20 like the the fundamental thing is like 42:23 there is a a generic uh bias in in 42:26 search and um a text is king and like 42:30 frankly the pe the the people who worked 42:32 in search are like much more interested 42:34 fundamentally more interested in 42:36 searching through papers um than um 42:38 through products 42:39 >> and so there's very few of the best 42:42 people in search in in in the world of 42:44 people who know most have actually ended 42:45 up working specifically on product 42:47 search which is really a should be seen 42:50 as its own completely different domain 42:52 that just sort of looks like it but it's 42:54 really clearly deeper with lots and lots 42:56 of 42:56 >> complex features and so on. So we like I 42:59 I got to like build up like over the 43:01 last year like a like you know a really 43:03 really really really great search team 43:05 and um exactly look at the you know 43:07 embeddings and these kind of things and 43:09 it's just there's so the amount of 43:11 improvements that just like 43:13 >> unexplored is just staggering. Didn't 43:15 you do search across? I mean, I find it 43:17 interesting um when Shopify does things 43:20 broader than the Shopify universe cuz 43:22 like Shopify shop the shop app can track 43:25 packages not bought from Shopify 43:27 merchants, right? 43:29 >> Yeah. And so shouldn't you do the search 43:31 thing across the whole web and not just 43:33 Shopify? 43:33 >> Oh, should we? I I think so because 43:36 again the thing that feels like the big 43:37 opportunity to me is um clustering was 43:41 always so hard to do because all the 43:43 features were developed manually and so 43:44 you had to like go say that the dress is 43:46 red and everything like that and it 43:47 feels like have you ever seen LM as just 43:49 a recommener engine where like I like 43:50 feeding in these are all the books that 43:51 I've read recently uh give me a 43:53 narrative non-fiction book you know that 43:55 someone who's read all these books might 43:57 like and it does extremely well but 43:58 you're kind of using the LLM to access 44:01 its world model and it feels Like now 44:04 you can have an interesting kind of 44:06 product world model and access it but 44:09 people like for the people using it 44:12 whether a merchant is on Shopify or not 44:13 is probably an incidental detail they 44:14 don't care about. Yeah, I totally agree. 44:16 And and you you're talking about how, 44:17 but like the what is the really 44:19 important thing. I I just want I want 44:20 the LLM to be proactive and just tell 44:22 me, hey, there's like this outfit would 44:24 look good on you and here's a preview, 44:25 right? Like and and and just like here's 44:27 a full cost and you want it, right? Like 44:28 this is like it's it's the personal 44:30 shopper outreach kind of component which 44:31 I think is just like 44:32 >> really valuable and this is not going to 44:34 be like some kind of 44:36 >> like you you'll get this from everyone 44:38 because the it's really easy to put 44:40 together. It's hard to get the data and 44:42 the embeddings and the sort of 44:43 understanding about um you know like we 44:46 have to figure out 44:48 it's funny the most important thing 44:50 again in the in the AI space is also the 44:53 unquantifiables it's the vibes people 44:55 call it there like in it's like it seems 44:57 like midjourney specifically is one 44:59 company that managed to actually scale 45:01 taste um because 45:02 >> very distinct aesthetic in the images 45:04 they produce. 45:04 >> Yeah. And maybe this is because they 45:05 they drove towards an particular 45:07 aesthetic that they just sort of kind of 45:09 fed they had ownership over. But it's 45:10 one that's clearly appreciated. 45:12 >> But David is also like deep founder mode 45:15 type how he operates. 45:16 >> Yeah. Exactly. And so like I mean it's 45:18 when when something 45:20 outperforms it's usually the product of 45:23 uh a place that doesn't make decisions 45:26 by consensus, right? As we know. So 45:28 >> yeah. Um so so you know how do we like 45:31 we need to figure out taste um um uh in 45:34 in this product sense. 45:36 >> Yes. 45:36 >> And um that we will we'll figure this 45:38 out. Uh someone's going to figure it 45:40 out. Maybe everyone's going to figure it 45:41 out. Um maybe it's already figured out 45:43 in the neurons and we just don't give 45:44 enough information for it to reason 45:46 around taste 45:47 >> and um so putting like 45:50 >> I think it will be very very good 45:51 personal chopper. The thing that I think 45:53 the aggregators still um to to just like 45:55 get back to your um point on the 45:57 aggregation side, there's many reasons 45:59 why people buy things um I I just like 46:02 lots of lots of things are input 46:03 purchases like it's like the things at 46:05 the um next to the checkout counter in 46:08 the supermarket. 46:09 >> I don't think people are going to use 46:11 Chad GBT to buy mass bars or something 46:14 like candy or so like let's just 46:16 >> don't do so online. 46:17 >> Right. 46:18 >> Exactly. But like is there an online 46:19 impulse purchase that is not suited to 46:22 happening in an AI app? 46:25 >> Interesting question. Nothing comes to 46:27 mind immediately. 46:27 >> I mean I guess online impulse purchases 46:29 are like things where you're scrolling 46:31 through Instagram and it's like you know 46:32 all the co stuff. It's the sharpest 46:34 knife ever developed. You know 46:35 >> I I it's I wonder if 46:37 >> umbrella that changes color in the rain. 46:39 >> I wonder if also truly inputs. I mean 46:41 they they sort of feel like input 46:42 purchase but they tend to be like pay 46:44 very few people part with money unless 46:46 they are like holy [ __ ] this is the 46:47 thing I always wanted like this is 46:49 actually usually 46:49 >> tapping into demand that's been there 46:51 for a long time it 46:52 >> it's sometimes a little bit the opinion 46:54 when you look at log files you're going 46:56 to get a very incorrect view of a world 46:58 um it's very often the longest uh um 47:01 consideration phases like for years I 47:04 wanted a travel adapter that can deal 47:05 with everything to everything and the 47:07 fact that like it goes in a chat from uh 47:09 from from from you or in an in Instagram 47:12 ad, doesn't matter. I'll I'll insta 47:14 purchase it at this point. 47:15 >> Okay. You were just ready to do that. My 47:16 other purchase I made recently that I 47:18 really liked is um uh I was going on a 47:21 long day hike and if you read the 47:22 accident reports is often people 47:23 dayhiking who get into trouble because 47:25 people who are going out on a multi-day 47:26 hike, they have like their tent with 47:28 them and they like they're prepared. You 47:29 know what I mean? And then people go 47:31 >> problem with bicycle helmets too, right? 47:33 Like same same issue. 47:34 >> It's a short journey exactly just to the 47:35 store that gets people in trouble. Yeah. 47:36 So dayhikers they get in trouble. Uh and 47:38 so they because they think they don't 47:39 need any preparedness and then they end 47:40 up out after dark and you know a rescue 47:42 party and everything like that. And so 47:44 it was like a it's called a bivvie you 47:46 know like a little sleeping bag for 47:48 emergency uh purposes and uh it's made 47:50 out of the reflective foil. Uh and so 47:53 again it's if you end up stuck um uh an 47:56 emergency foil sleeping bag just so you 47:58 don't die of exposure but it folds up to 48:00 it's like that size and so you can just 48:02 stick it in a dayack and uh now you are 48:04 somewhat more prepared for the uh the 48:06 elements than you were before. Um, I 48:07 don't know. Do you have a favorite 48:08 recent purchase? 48:10 >> I I like the I just really like these um 48:14 um I mean someone going just like too 48:17 ridiculously further and and and and and 48:19 to just like almost like products that 48:22 just celebrate craftsmanship in in a way 48:24 even in places where just like no one 48:25 else does. Like 48:27 >> it's like I I I 48:29 example but like I I you know take all 48:31 these vitamins in the morning and like I 48:33 bought these like 48:34 >> uh CVS vitamin pill cases at some just 48:37 for like for travel and and then like I 48:40 found on Shopify like this like Japanese 48:43 uh perfectly CNC built like unbelievably 48:46 beautiful pillcases like it's just like 48:47 it's like the entire website is just 48:49 like a love letter to 48:50 >> beautiful pillas you know precision 48:53 engineered like you like tolerances of 48:55 sub million. 48:56 >> What's it made of? 48:57 >> Like metal, which is like absolutely 48:58 like it feels like it feels so um 49:02 >> anyway whatever it I'm I'm really 49:04 excited about it. I I itch like let's 49:07 put a link in link. Um I forgot the name 49:10 sadly. Um and uh now I like gift those 49:13 things because they're just beautiful. 49:14 >> Yeah. Yeah. No, that's cool. 49:16 >> Um 49:18 uh how's shop doing? Um we referenced it 49:21 there. 49:21 >> Shop is awesome. It's um you know it 49:25 started out this life as um a feature 49:27 called remember me which uh like is 49:29 literally the name what was on 49:31 >> we also had a thing that was called 49:32 remember me. That's that's like the 49:33 original name for link. 49:35 >> No way. 49:36 >> Yeah. 49:37 >> That's that's trippy. 49:38 >> That's very funny. 49:39 >> So the um which you know is like I mean 49:43 >> they're kind of the same thing. I mean 49:43 shop is much broader but they're they're 49:45 similar things. 49:46 >> Yeah. Actually shop feels more narrow. 49:48 It's like it's on it's on it's on 49:50 comments specifically. 49:51 >> But you guys do all the shipping 49:52 tracking. 49:52 >> Yeah. Yeah. 49:53 >> And so so um yeah, so yeah, shop started 49:56 like shop then I mean shop is Shopify's 49:59 um uh brand for buyers, right? Like so 50:04 it's um you know Shopify is what you use 50:06 to 50:07 >> be on shop. 50:09 >> Um and at least this is how we think 50:11 about it. So um you know there's a shop 50:13 app and uh you know shop pay. Um, so I 50:16 mean it again it started 50:18 because it just is a ridiculous thing 50:20 that we asked anyone asked anyone to uh 50:24 fill out their address with their thumbs 50:26 when you purchase something that is 50:27 insane. This was meant to never be a 50:30 thing because you know like in the I 50:32 don't know when they came up with HTTP 50:34 but there's the 50:36 >> 402 error code payment payment required 50:39 right like um it's like built into the 50:41 HTTP standard and no one ever 50:42 implemented it 50:43 >> and so moving money around was supposed 50:45 to be built into the platform and um uh 50:48 wasn't and then um you know credit card 50:51 forms were like hilariously insecure 50:53 over the normal web and then we came up 50:54 with HPS and made it reason reasonable I 50:57 suppose and then it became good enough. 50:59 Um there's all these efforts Apple Pay 51:02 like I think actually PayPal was around 51:04 but then sort of didn't do anything you 51:06 know. So there's so many people like 51:08 doing this and like they just like okay 51:11 well 51:11 >> clearly this is going to be solved soon 51:13 but like might as well like just put a 51:15 remember me thing in there so we can 51:16 solve this now um as almost a polyfill 51:19 for the future of the internet and then 51:21 um no one did. I mean very few people 51:23 did and it just sort of didn't happen 51:24 and then now shop is like incredible. 51:27 super glad to have it. It's like 51:29 >> the most converting thing we've ever 51:30 done. 51:30 >> Yeah. What seems interesting to me about 51:32 it is I think it's a significant part, 51:34 it's not the only part, but it's a 51:35 significant part of the inversion where 51:37 again now the small stores are 51:40 significantly better off than the large 51:42 stores online. And that's just a 51:44 striking 51:45 >> it's striking and it's like it's it's 51:47 16% conversion increase by by 51:49 >> Yeah. 51:49 >> it being round, right? Like it's um 51:51 >> um so it's it's a remarkable 51:55 product. Um now partly because again 51:57 it's power of branding. It's purple 51:59 always and people learn to trust it and 52:01 associate it with a store that's like 52:02 actually 52:03 >> like recent and cared for and uh doesn't 52:06 you know they now they don't have to 52:08 before they get to check out they now 52:09 know they will not have to type in the 52:10 address of their thumbs like some kind 52:12 of 52:13 >> heathen from the '90s. 52:14 >> Yes. 52:15 >> Just like 52:16 >> you mentioned personalized advertising 52:18 and being a big fan of it. It seems to 52:20 me that there's a um real symbiotic 52:22 relationship between Instagram where 52:25 people are kind of scrolling in an 52:26 aesthetic mood looking at nice things 52:30 and then Shopify stores where they need 52:33 customers. Uh and so especially during 52:36 the pandemic there was this huge growth 52:37 of direct to consumer brands who really 52:40 cracked the Instagram 52:43 >> cact math and they were able to grow in 52:46 a big way there. Then of course the 52:47 Apple at changes happened where for a 52:50 brief period it got harder. It seems 52:52 like Meta has mostly solved that. But 52:53 anyway, I'm just curious what's the 52:55 current state of like is this worldview 52:57 have true that there's a lot of Shopify 52:59 um merchants that have been able to 53:01 really scale up thanks to personalized 53:04 direct you know direct advertising. 53:06 >> Yeah, 100%. It's it's a predom it's a 53:08 main um um uh growth channel for is is 53:11 is advertising as a on platform. the 53:13 meta platform specifically is just so 53:15 incredible um for this like I think the 53:18 the the the meta Shopify LAU has created 53:21 more businesses than like uh um you know 53:25 any government policy in history like so 53:28 it's like I mean if you just compare to 53:29 television and uh uh consumer packaged 53:33 goods before the television channel 53:35 required broadcast and and so the only 53:38 things you could really advertise were 53:39 like mass market high margin products 53:41 that could afford the advertising 53:43 washing powder and 53:44 >> exactly so it's like you know laundry 53:46 laundry laundry laundry laundry laundry 53:46 detergent and toothpaste and so um uh 53:50 the Instagram Shopify is the precise 53:52 inversion of this is like highly uh like 53:54 highly merit to meritful products 53:57 >> it's the first time you can advertise 53:58 niche products 53:58 >> exactly and and the niche products 54:00 actually doing better that's the crazy 54:01 thing like advertising directly uh 54:05 snowboards is like impossible there um 54:07 um because a billboard for snowboards 54:10 the brand halo of burden means you you 54:13 can't compete with that if there's 54:14 intersections for like like snowboards 54:17 only for powder snow on the you know 54:19 like US west coast and uh like in in 54:23 certain environments with like just like 54:24 you know like there this is not the best 54:26 example usually there's like three 54:27 different intersections of something 54:29 >> and it goes back to you know Kevin Kelly 54:31 wrote about this in in the early 2000s 54:33 he's like it's like he wrote an essay 54:36 thousand true fans and the internet is 54:37 about finding your thousand true fans 54:39 and everything else will uh will happen 54:41 >> and and um restoring that idea or 54:46 bringing this idea into retail has done 54:49 a lot of good things, right? Like it 54:51 just like created better products. It 54:52 created this symbiosis and like meta 54:55 allowing you to discover one of those 54:58 people who has a potential to be one of 55:00 those thousand true fans. It's an 55:01 incredibly powerful thing that just like 55:04 again led to enormous good and 55:07 employment and all these kind of things. 55:09 So it's a very very powerful 55:10 combination. Um there's all sorts of 55:12 other ways how people uh like like grow, 55:15 but this is a very very large part of 55:17 everyone's strategy. 55:18 >> You're saying this enables 4K for 55:20 commerce. There's this like higher 55:22 resolution, more specialization as 55:24 opposed to the blurry generic products. 55:26 >> Yeah, it's exactly it. Find find the 55:27 people who are really interested in who 55:29 appreciate the products. 55:31 >> I like that. 55:32 >> How is the stable coin rollout going on 55:34 Shopify? So, you know, we're obviously 55:36 doing lots of stable coin uh stuff with 55:38 lots of people, including you guys, but 55:39 I feel like 55:40 >> you guys are the 55:42 >> one of the biggest 55:44 >> bets on 55:46 stable coin payins. Everyone should just 55:48 be able to, you know, use stable coins 55:49 as a payment method that we've seen. And 55:51 so, how's that going? 55:52 >> I'm I'm I'm obviously like just like 55:54 you, huge fan of uh the concept of 55:57 stable coins. I love the work you you're 55:59 doing there and congrats on the launch 56:01 of Tobo. Um, and the internet just needs 56:04 its own currencies. I think it's 56:06 important uh at least it needs its own 56:07 infrastructure to move uh things by 56:09 internet speeds. Um, and I'm excited for 56:12 it because I mean sort of I think a 56:14 valid criticism of crypto in the past, 56:16 as exciting as it is, has been that um 56:20 the utilitarian 56:22 difference between a um real US dollar 56:26 and a US dollar crypto coin is like that 56:31 the US dollar crypto coin is mostly like 56:33 it's sort of like gift certificates to 56:34 scams. It's like there's not that many 56:37 things you can actually purchase with it 56:39 that are of high utility value. It's 56:41 like obviously this is I think 56:43 >> the economy was too circular. 56:45 >> Exactly. It's like most of it was like 56:46 you you purchase other coins that then 56:49 >> you might have 56:50 >> I mean hopefully you diamond handed into 56:53 some kind of upswing but like it's like 56:55 that's not usually what happened. 56:56 >> Um and I think that's valid criticism. I 56:59 think the the most important thing to 57:00 happen there is um we need to expand um 57:05 what can be purchased the stable coins 57:07 to uh the things people want to purchase 57:09 like outside of like just the sort of 57:11 insula crypto uh cryp fintech. Um 57:15 >> but yeah like we working hard on getting 57:17 all Shops to just take 57:20 >> what does the merchant response mean? 57:21 very positive again because like through 57:24 like the work that bridge is doing we 57:26 can settle into US dollars directly. So 57:29 the merchants 57:31 can they have just gained a new ability. 57:33 >> Mhm. 57:34 >> With without having to make a strategic 57:36 choice to go into new industry like if 57:38 >> there's demand for being able to spend 57:40 stable coins and uh they can now sell to 57:43 everyone who wants to particularly do 57:44 that. And so it's like it's totally 57:47 transparent and that's the beautiful 57:48 thing amotizing 57:50 >> um R&D over millions of businesses with 57:53 us centrally because we can just like 57:55 build up the system 57:56 >> um and everyone gets benefit of it and 57:59 so um you know obviously it's in testing 58:01 like the conversion rates have to not be 58:03 negatively affected obviously by its 58:05 presence. So there's some like we have 58:07 to make absolutely sure this is the case 58:09 and like lots over abundance of caution 58:11 on every every roll that goes from the 58:13 checkout because it's like the probably 58:15 I mean it's you know busiest check out 58:17 in the internet basically. So um we we 58:19 might not want to 58:21 >> yes 58:21 >> reduce um conversion rates ever. 58:24 >> Yeah. I think it'll also be interesting 58:25 to watch the consumer training aspect of 58:27 it where like QR codes is always my 58:29 example of this where like the tech has 58:30 existed since what the '90s but it was 58:32 really only during the pandemic when 58:34 like we install the software in people's 58:36 brains to understand how QR codes work 58:39 >> but there was a moment and now they're 58:41 kind of everywhere for everything and 58:42 obviously they're in Asia before they're 58:43 in the US and everything like that. I 58:45 think there'll be something similar with 58:47 stable coins where they'll be offered at 58:48 the point of checkout way before people 58:51 become super familiar with them. But 58:54 we're we're probably only a small number 58:56 of years away from from some mass market 58:59 familiarity. 58:59 >> I think it's going to be very very uh 59:02 natural for people but also in a way 59:03 that they pro they will just not think 59:05 about it. It's like it's a normal thing. 59:06 It's like doesn't matter like it's going 59:09 to be so well uh toolled up to 59:11 >> bridge between these things that 59:13 >> um it's you know just it's using crypto 59:16 rails while using the internet and until 59:19 you need it out of the internet it's not 59:21 like you don't have to bridge anything. 59:23 So 59:23 >> yes. Yes. You've written a few times the 59:25 um kind of the stuff Stripe and Shopfire 59:27 are doing together and you know we've 59:29 caught up at various points on and you 59:31 even talked at the meta level about the 59:32 Stripe Shopify relationship where 59:34 there's some I know is it like Apple 59:36 TSMC or you know something there where 59:39 it's like deep intertwined building 59:42 in a stable way but also kind of 59:45 expanding over time and no confusion as 59:47 to who's ultimately the one building the 59:49 end product versus who's kind of 59:50 providing some of the underlying 59:52 infrastructure. But I'm curious what you 59:54 think has worked well because yeah, a 59:58 lot of corporate relationships do not 60:00 last. I mean, I guess if you dated 2012, 60:02 which I think is accurate, maybe 2011 um 60:05 uh 2012 like 13 years, it's it's 60:07 certainly more tenure than some. 60:08 >> Yeah. And um I think uh it's almost more 60:11 impressive when you think about it as 60:13 market cap expansion, right? Like I mean 60:14 we were both a very greedy company. Um 60:17 well also I mean we have had a highly 60:21 functioning excellent uh symbiote 60:23 partnership and probably there's like I 60:26 mean how many more digits on the market 60:28 cap of both companies like we've 60:30 expanded many many many times here many 60:33 orders of magnitude 60:34 >> and um that's pretty rare because like 60:36 again 60:37 >> partnerships especially in tech tech is 60:40 extraordinarily bad at partnerships it's 60:43 um um and and uh I mean it It's it's 60:47 always treated as a um um prisoners 60:49 dilemma that uh at some point someone in 60:52 the company realizes defecting gives you 60:54 five points on the on the mouth and that 60:56 sounds really really good with the 60:58 short-term focus. So everyone must trust 60:59 each other and so it's pretty rare and I 61:01 think it's it's it's it's cool that we 61:03 pulled it off and I mean I I appreciate 61:05 the partnership. It's it's it's been 61:06 really really fantastic being able to uh 61:09 just focus on my space instead of having 61:12 to, 61:13 >> you know, duplicate 61:15 things that Stripe would always have 61:17 done better. I think a lot about main 61:19 quests and side quests and I I I 61:21 >> I 61:23 I have such good I guess founder market 61:26 fit with commerce. I just like I I I 61:28 really care about this and I really I 61:30 have 61:30 >> I can spend my entire like it's one car 61:33 popper calls it finding like like the 61:35 best gift in life is finding a beautiful 61:37 problem that you can never solve 61:40 >> and even if you accidentally solve it if 61:42 you're so unfortunate to solve it 61:44 hopefully it has like plenty of um uh 61:47 enlightened problem children that you 61:49 can then tackle right like so um and and 61:52 and so so so finding one of those is 61:53 good and not having to like meander into 61:55 some complete adjacency 61:57 to support it is is actually wonderful. 61:59 >> Yeah. Yeah. I think there's also 62:01 something about um when you talk about 62:03 this idea of focus. 62:05 I think act second acts are kind of a 62:08 bit over valorized in Silicon Valley 62:11 somehow where everyone talks about oh we 62:14 want to do like a new AWS is kind of the 62:16 coolest thing ever where you do a second 62:17 thing that is completely unrelated to 62:18 the first thing versus Nvidia is the 62:21 largest company in the world and they 62:22 just started making GPUs in the '90s and 62:25 turns out a lot of people want GPUs and 62:27 I mean they did obviously lots of 62:28 incremental expansion from there like 62:30 calling them GPUs is is doing a 62:32 disservice but it's in a very hill 62:34 climbing expans expansion way I think 62:36 >> they went they went vertical I think it 62:38 said horizontal business books tell you 62:40 to go horizontal because again the chief 62:44 competency of most businesses ended up 62:46 being having having a stopwatch like 62:48 that's it was sort of a management 62:49 system world but we are actually in a 62:52 different world now it's like this this 62:53 is commoditized it's like easy to access 62:55 and probably GPT5 is better at it than 62:57 most managers 62:58 >> and um it's now actually like well what 63:01 problem do you understand better than 63:02 what's in the training tokens for GPT5 63:05 what do you care more which is a human 63:06 capacity specifically and so like Jensen 63:10 just cared more about GPUs and and and 63:12 and you know saw more potential in that 63:15 >> is there a bit of MBASM 63:17 um here too where if one was trying to 63:20 you know when Shopify was a thousand 63:21 times smaller raise money for Shopify it 63:24 would just sound ludicrous to say we're 63:26 just going to do e-commerce software and 63:27 it turns out that's a really big market 63:29 and here's our current projection and 63:30 here's when it's a thousand times larger 63:32 it's like get out of here come on 63:33 whereas if you say, "Oh, we'll expand to 63:35 do this and we'll expand to do that. 63:36 We'll build a CRM or whatever." It's 63:38 easier to justify, you know, expanding 63:40 into other TAMs. 63:41 >> Like I got rejected from plenty of VCs 63:44 in 2008 when I went around and some of 63:46 them said they would invest and then it 63:48 was usually terrible terms of like 63:50 ratchets or multiple and I had to move a 63:52 company to Silicon Valley which again I 63:54 would have if I wouldn't have told me to 63:56 do like once you tell me to do something 63:58 I don't want it anymore. Um and um uh 64:02 but anyway, so like like and I I caught 64:04 up with many of the partners later in 64:06 one case. We had really interesting 64:07 conversation and just like he asked me 64:09 what what did we miss back then 64:10 >> and I just like well the reason why you 64:14 didn't invest is like the like you said 64:16 like there's 40,000 online stores in the 64:19 world right now in 2008 and well you can 64:22 get half of them but that's not actually 64:24 that big of a business. Um and um what 64:27 they missed is like well Shopify itself 64:30 was the solution to that problem. Like 64:31 we the reason why no one was building 64:33 online stores is because it was 64:35 >> the friction was so high. It's like 64:37 >> um and so it's a little bit of that, 64:40 right? Like it's you got to like Yes. 64:42 It's it's not an interesting business 64:44 plan. It doesn't the pitch isn't very 64:45 good and specifically coming from me. 64:48 >> Yeah. Yeah. Coming from me is probably 64:49 especially bad. But like it's also kind 64:51 of man the best products in the world 64:55 come from people just like really really 64:56 really really uh being deep on something 64:59 like in a programming world we all using 65:02 cursor which is basically VS Code. VS 65:04 Code is Eric Gamma's fourth editor right 65:07 like it's like he spent his entire life 65:09 making these things craftsmanship. 65:10 >> Yeah. It's like he he wrote the design 65:12 patterns book that many of us read at 65:14 some point literally about programming 65:16 text editors, right? Like so it's like 65:19 yeah sometimes you just need to dedicate 65:21 a career to uh a field and 65:25 >> get deeper into it than anyone else and 65:27 uh celebrate craftsmanship along the way 65:29 and kind of try to build the best thing. 65:32 >> Yeah. But again, it's hard for people to 65:35 give credit to that idea in a model or 65:38 like people people are not capable of 65:40 being sufficiently optimistic about a 65:42 business that really works. They it just 65:43 it blows through all their projections. 65:45 >> Yeah. The question is also like like 65:47 we've seen the story so many times that 65:48 like some step function changes like at 65:51 some point like maybe the shope and like 65:53 um like all these kind of things with 65:55 Shopify it somehow it became something 65:57 very different than what it otherwise 65:58 sort of what what you would have 65:59 visualized. And I mean this is in no 66:02 area more true than uh in the Nvidia 66:05 case where like I mean these CUDA cores 66:07 the tensor cores were around on the they 66:10 were on these cards which you put which 66:12 everyone they were part of the bomb um 66:15 >> and very little use by any of engine 66:18 like 66:19 >> carax new engine probably used them for 66:21 something useful but they were just 66:23 there and cuda was not even like like 66:25 that was around for 10 years before Alex 66:27 net got trained. Yeah, 66:29 >> like it's an incredible vision to just 66:31 like doggedly pursue this because you 66:33 know in the future there will be value 66:34 if you do this. Um 66:36 >> and um it's it's makes you wonder I 66:39 think that's probably true for almost 66:40 every space. I think every space driven 66:42 to that level 66:44 >> could become so big. 66:46 >> Isn't there something here where more 66:49 founders should be willing to follow 66:51 their nose of what's just a good 66:55 product? Because again I think that what 66:59 can happen as companies scale is 67:00 founders get successful by building good 67:02 product and then they hire a bunch of 67:04 people and finance people and run the 67:06 planning process and everything like 67:07 that and it's like okay well what's the 67:08 revenue projection for this and you know 67:10 things like that whereas actually like 67:13 when when we built um stripe atlas for 67:16 incorporation there was no amazing 67:18 projections behind that that one day 67:20 this would be a giant business and it's 67:21 not a giant business for us by any kind 67:23 of revenue metric but we think it's very 67:25 valuable because people start their 67:27 business on Stripe and we help them with 67:28 something foundational of like 67:29 incorporating the company and get it off 67:30 the ground and it all kind of ties 67:32 together in this very immeasurable 67:34 nebulous way and kind of like you're 67:36 saying with CUDA you know that was hard 67:38 to justify at that moment in time but it 67:41 seemed like the right technical decision 67:42 and there's something maybe where 67:44 founders should be a bit more willing to 67:47 follow their noses on what is right for 67:49 the product and right for the user and 67:51 it'll all come out in the wash 10 years 67:53 down the line. Yeah, I'm a great 67:55 believer that if you if you build things 67:56 that are of value, 67:59 then people will figure that out. It's 68:01 like it's very very hard to keep people 68:03 away from things that make um their 68:05 lives significantly easier, better or 68:08 more joyful. Uh and 68:10 >> um I mean I I I get the front row seat 68:12 to this on my platform, right? It's 68:14 crazy how fast some success stories. One 68:17 of my favorite things we um we do is we 68:20 um started making these really sort of 68:22 nice um um awards for people when they 68:26 do hit certain sales milestones and um 68:30 uh you know some of those I when I go 68:33 travel I check out who's just got them 68:36 and just like sometimes deliver them and 68:37 Harley does this much more than than me 68:39 and like it's it's such a cool thing to 68:41 like to to to talk with them and man 68:45 like some of these businesses that hit 68:47 like a like 100,000 or million sales are 68:49 like 68:50 >> 18 months old. It's just like just 68:53 cracked the code with something of and 68:55 and and with the most you know talk 68:57 about thousand true fans. It's like like 68:59 millions. It's like an army created 69:02 because 69:03 >> people just so starved for this like 69:06 authentic like unabashed like something 69:10 needed to be done. Some something was 69:11 wrong involved. something. It involved a 69:14 product not existing and that product 69:17 not existing was such a bother to 69:19 someone that they made it their mission 69:20 to create it and then they they they 69:22 went through trials and tribulations 69:25 usually they often documenting the 69:27 journey. 69:27 >> Yes. 69:28 >> Um taking like allowing other people to 69:30 write shotgun to sort of over time try 69:33 on their vision of the world which again 69:35 is not to change the world but it's to 69:38 add the thing that was missing. They're 69:39 just like it's it's a makes things 1% 69:42 better, not like sort of everything is 69:45 new. And they just agree with that and 69:47 they love authenticity of it. They love 69:49 that someone that they know have a 69:51 little bit of a relationship with like 69:53 kind of pursued this thing that they 69:55 think it was worth doing and then the 69:57 product comes out and and and they tell 69:59 everyone and then more people buy it and 70:02 and and it's just like there's so much 70:04 good that comes out of this and that was 70:06 all value which was there for everyone 70:07 to take, right? like every product that 70:09 sells well was a discovery. It's like it 70:11 was sort of invented but also kind of 70:13 discovered and so um um um but but it's 70:17 that person that did it and they deserve 70:18 the chronolic value that comes to them. 70:20 It's such everything about this is like 70:22 it's the best parts of um capitalism all 70:24 rolled up um for everyone to inspect and 70:27 review 70:28 >> and we get to have a front row seat from 70:30 data and front row seat sometimes with 70:32 giving these um uh sort of celebration 70:34 uh artifacts 70:36 >> and um 70:37 >> anyway so this is always the greatest 70:39 thing is like just like the people who 70:41 are using the platforms are actually 70:43 really inspiring people 70:44 >> well you know I I feel like it must be 70:46 particularly satisfying for you because 70:47 there's all these edge case failure 70:48 modes of capitalism of you know rent 70:50 extraction or igopy or regulatory 70:52 capture everything like that and you 70:53 just have a front row seat to the pure 70:57 competitive differentiation 70:59 uh people succeeding on the merits and 71:02 uh it's a it's kind of very pure form 71:05 >> I mean I feel like you have that too 71:07 it's it's and it's like we should like 71:10 write postcards from you know to the 71:13 rest of it's like it is like there's 71:14 something really to this like you know 71:16 just don't this real world thing right 71:19 like it's like the the one everyone says 71:22 your product doesn't wouldn't work in 71:24 like that sounds so dystopian is like it 71:27 it does really sound quite dystopian 71:29 like so I I don't want to have anything 71:31 to do with it. I just much rather like 71:33 hey just build this whatever world 71:36 whatever that is um and invite everyone 71:38 over and if it's good and works and it's 71:41 than people to join like it's actually 71:43 that simple and I think it's better to 71:45 um like live in a world with like not 71:48 like people going direct uh people 71:49 communicating right people being real 71:51 people it's there's no there's very very 71:54 rarely is there a uh PR department or 71:56 like even there's probably a marketing 71:58 strategy but the marketing strategy 72:00 works like based on you know just like 72:02 new ideas and organically and um with 72:05 very few people being highly enabled 72:07 building products that they care about. 72:08 >> We had Mackenzie from Amber uh on here 72:11 which is um SAS for um financial 72:14 infrastructure for farms but she was 72:16 saying one of the big trends in 72:17 agriculture right now is direct to 72:19 consumer stuff where you know you're 72:20 actually transacting directly with your 72:22 farmer. Uh and I'm sure a lot of them 72:23 are running on on Shopify as well. 72:28 >> What have you learned on the Coinbase 72:30 board? I never had another job really in 72:32 my apprenticeship and um um mostly 72:35 learned what not to do uh as a company 72:37 and then um sort of had the opportunity 72:39 to start Shopify just trying little 72:41 opposite and that's basically the 72:42 operating system I've been running on. 72:44 It's been amazingly successful. So thank 72:45 you Semen but like it's it's it's it's 72:48 so hard especially building it in Canada 72:50 like in Ottawa and Toronto. It's funny, 72:52 I always had this effect of I came to 72:54 Valley a lot um when I when I had time 72:57 and um money, you know, I had all these 72:59 meetings and then talked with people and 73:01 um they described to me how real 73:03 companies do these things and then I 73:04 went home and tried to implement it and 73:06 only years later did I realize um very 73:09 few of the companies describing what 73:10 they're doing um actually did the things 73:14 that they told me. Everyone gave me sort 73:16 of a idealized highlight reel and I 73:18 compared that to my average and then I 73:20 implemented how I think you could do a 73:22 better job on the stuff that described 73:24 and sometimes ended up with uh good 73:26 things and I was very very happy with 73:27 this and then um now obviously I'm 73:30 describing when this goes really right 73:33 and then just as often this went really 73:34 wrong like so they just like Shopify 73:36 ended up being this like sort of 73:38 lopsided thing that like is like um 73:41 probably ludicrously overengineered and 73:43 but advanced at certain things and then 73:46 like incredibly primitive at things that 73:48 no one fails usually because in a in a 73:51 in a more liquid uh labor market where 73:53 people more people build companies the 73:56 information is just more readily 73:57 available. What was awesome was like 73:59 joining Coinbase as a board because like 74:01 it's like here's another founder run uh 74:04 company by an incredible entrepreneur 74:06 that's like put together like again not 74:09 by committee but by um you know like a 74:12 clear desire to build an like an an 74:15 important company important institution 74:17 and just like holding it up and seeing 74:19 like what they were incredibly strong at 74:21 um compared to like like it gave me I 74:24 learned so much about my own company and 74:25 I could like you know 74:27 >> a reference 74:28 have a reference point, right? Like, so 74:29 it's it's it's it's quite a gift. I I I 74:31 I don't know when it's the right time 74:33 for an um successful entrepreneur to 74:36 join another company. I wouldn't do it 74:37 too early, but like it's it eventually 74:39 becomes quite valuable. And so, Cro is 74:42 such a remarkable company. you had Brian 74:43 on. He's such an incredible um 74:46 entrepreneur. 74:47 >> Very first principles 74:48 >> and that was a company under brutal 74:51 attack by the regulators in a way that 74:53 like I very few companies of all time 74:56 have ever faced. That's like that for 74:59 like 75:00 >> the government wanted them dead um and 75:02 pretty much settle right so I learned so 75:05 much about the posture that uh Brian 75:07 took to it. he will never get enough 75:09 credit for the yes 75:10 >> for the courage and like the the 75:12 intestinal fortuitity he's shown there 75:14 um 75:15 >> um because many of the details are 75:17 internal to the company and the board 75:19 but like god damn it he's like good like 75:22 I I do not think I could have 75:24 >> navigated my company through a similar 75:27 barrage 75:28 >> yeah Brian strikes me as someone who is 75:30 very hard to manage up to you know 75:32 that's always the risk for CEOs that 75:34 people are trying to you know construct 75:35 some pmpkin village and you know he was 75:37 describing 75:38 the policy uh work that they were doing 75:41 and he was saying, you know, well um you 75:44 know, the company's meant to be the good 75:45 cop and the trade group will be the bad 75:46 cop and then being the trade group and 75:48 it's like, wait, these are also the good 75:49 cops. And so he was saying, okay, here's 75:51 our new bad cop plan. But again, he was 75:52 just kind of refusing to go along with 75:54 the the things that everyone was serving 75:56 up to him as the uh as the plan. 75:58 >> I think I think great entrepreneurs are 75:59 ungovernable. And um uh it's the 76:02 problems often arise when people somehow 76:05 manage to figure out how to Yeah. I 76:08 think studying failed companies is 76:09 actually often more useful than um you 76:12 agree. Yeah. Like there what you find 76:14 invariably is that the company knew 76:17 exactly the stuff that was the problem. 76:19 Uh it's just like somehow 76:21 >> CEOs got disconnected from that. 76:23 >> Yeah. It's not so much that I want to 76:25 study failures. It's that I really want 76:28 to avoid making the same mistakes that 76:32 other people have well-intentionally 76:34 made and like found, oh yeah, no, that's 76:36 a blank canyon up there. We should at 76:38 least be able to make new exciting 76:39 mistakes. Um Rob the guy here who runs 76:41 corp dev uh you know he's very 76:43 experienced in it and he's done hundreds 76:45 of acquisitions uh you know previously 76:47 but it makes him way better as a corp 76:49 dev lever because it turns out that 76:52 there's lots of ways for deals to go 76:54 sideways and if you just have seen a lot 76:56 of them before you can keep tabs on them 76:58 and be finding ways to steer away from 77:00 them things like that and I think just 77:02 avoiding the welltrodden mistakes and if 77:04 you're going to make mistakes at least 77:06 make new mistakes that haven't been 77:07 discovered yet was kind of underrated as 77:09 a It's totally underrated and and but 77:11 but completely vital because like you 77:13 you die to a thousand paper cuts as a 77:15 company. So like you you kind of have to 77:17 like first avoid like it being becoming 77:19 a thousand and then second you will make 77:22 mistakes make them original so there's 77:24 alpha in it so that you acrew some 77:25 additional knowledge that now you is 77:27 unique to you. 77:28 >> How do you like to be ungovernable? How 77:30 do you like to avoid being managed up to 77:32 >> uh probably can't share all your secrets 77:33 because part of them are part of the 77:34 special. 77:35 >> I just like think going direct a lot 77:37 inside of a company too. is like just 77:38 like I I I mean this 77:40 >> the turning of the org chart and 77:41 information flowing up being polished at 77:43 each step. 77:43 >> Yeah. I I'm just like I I don't I don't 77:45 think like up and down is sort of not 77:47 the right mental model is in and out is 77:49 good. It's like uh like certain 77:51 decisions are best made at the outskirts 77:53 of a company closest to the customers 77:54 and um 77:56 >> um but I do I'm a big believer in many 77:58 decisions have to be brought inwards as 78:00 well. Um like this is why you know 78:01 product reviews are like very 78:02 centralized. It's like a way of 78:05 efficiently moving information uh 78:07 completely in the bend of or charts. I'm 78:09 a fan of functionally organized 78:10 companies. It's um the trade-offs. It's 78:13 like it's it is significantly harder for 78:17 executives. 78:18 it's very unpopular with executives 78:20 because obviously they just don't own 78:22 like entire shebang that they can make 78:25 all decisions in and that's requires a 78:27 lot more collaboration which feels wrong 78:29 and feels slow but like you make vastly 78:32 better products in um um functional 78:36 organizations at least for Shopify and 78:39 um you know that that that helps. I just 78:42 like know a lot about the company like 78:43 like being in so many details like I I 78:46 try to be like in AI stuff first uh so 78:49 that I can 78:50 >> you know it's it's I always see it as a 78:52 gift when something new happens. 78:54 >> Yeah. 78:54 >> Because especially like I I'm very good 78:57 at quickly figuring out stuff and then I 78:59 I I sit there with knowledge and I can 79:01 now like look at okay who else gets 79:05 there fast and those are my future 79:06 leaders. Like that's a very implicit 79:08 test. 79:09 >> Oh that's interesting. you're kind of 79:10 testing some adaptability or something 79:11 like that. 79:12 >> Yeah. Like the biggest difference 79:13 between people inside of any company or 79:15 just like in general in industry is like 79:18 um some people fall in love with 79:19 solutions, some people fall in love with 79:20 problems. And I I just like 79:22 fundamentally am a person who 79:24 appreciates people who fall in love with 79:25 problems. They are hard to distinguish 79:27 during fairly stable times, right? Like 79:28 it's very hard to know who is who. Um 79:30 because well, if you don't change 79:32 anything, they look exactly the same. I 79:34 like inducing change in general to to 79:35 sus this out. I like 79:37 >> I don't know. Right. 79:38 >> How do you induce change? That sounds 79:40 euphemistic. 79:41 >> I don't know. Just like literally like 79:42 that like do stuff. Um it's I think 79:44 reorgs are sort of a standard way we all 79:48 reox work because reorginging is 79:50 valuable. It doesn't matter what your 79:51 organization structure is afterwards. 79:53 The b well I mean does but like like 79:55 there's some benefits to any 79:56 organization structure 79:58 >> shakes things out of the stasis. 79:59 >> Well and and the advantages of any 80:01 organization structure occur immediately 80:03 where the disadvantages only appear over 80:04 time. So if if you change a lot um you 80:07 do multiple things you you're also like 80:10 >> I think heightening your misalignment 80:11 with the people you don't want in your 80:12 company is an enormously important 80:14 thing. So like if you reorg 80:16 >> Yeah. So exactly. So if you if you reorg 80:18 a lot you end up um the people left out 80:22 who don't like that which um is a 80:23 perfect like it's probably the sane 80:25 position to take even. 80:26 >> Yeah. 80:27 >> If a company thrives under a lot of 80:28 change and is exposing itself to a lot 80:30 of change then uh it's important that 80:32 you are clear about that. I think you 80:34 try to make your company as different 80:36 from every other company as possible, 80:38 which is actually I think the opposite 80:39 of what most companies, most CEOs 80:41 optimize for 80:42 >> and probably most professional 80:43 management, 80:43 >> right? Um I I think a lot people want to 80:46 make their company as similar as 80:48 possible to all other companies. But I 80:50 think that's I mean that to me that's a 80:52 lack of diversity, right? Like there 80:53 should be diversity in companies because 80:55 there is diversity in people and people 80:57 need to fit find their product. 80:59 >> In what ways is Shopify most different 81:01 to other companies? I mean the sort of 81:04 uh you know the internal tool culture is 81:06 maybe maybe an an example of this. It's 81:09 quite clearly um R&D run in a way that 81:12 quite a few companies are but they don't 81:14 acknowledge or obscure it. Um we 81:16 actually say it and celebrate it and say 81:18 we actually are on the opinion at this 81:20 point that everyone should be able to 81:21 code at least sort of since VIP coding 81:22 is around. This is clearly this has gone 81:24 from a, you know, have to spend all of 81:27 your teenage years in a dark room 81:29 cultivating a a rare skill to um 81:33 spending a couple weekends with YouTube. 81:34 Like, let's go. Like, it's this is not a 81:36 problem anymore. 81:37 >> Speaking of teenage years, I'm always 81:38 very interested in the on-ramp to 81:41 entrepreneurship and the on-ramps to 81:43 programming. M and I'm curious, some 81:46 people decry that um programming has 81:48 gotten harder to unramp into where you 81:51 know the first stuff I built was like 81:53 janky PHP web pages but obviously PHP is 81:55 actually very nice because you know 81:56 you're just writing code directly in the 81:58 web page and you refresh it that's the 81:59 execution environment and now 82:01 >> it's a high watermark of our industry. 82:02 >> Well, so is this and so have we made 82:04 programming too hard to get into? 82:07 >> Totally. Okay. to to tell my wonderful 82:09 board member David Hine Hansen like the 82:11 high watermark of uh the world of 82:14 programming at least web programming was 82:16 clearly when we uh FTP a PHP file up and 82:20 it was instantaneously deployed how long 82:22 does it take code to deploy these days 82:23 like this is like all choices now right 82:25 like I mean we made some choices which 82:27 are worth making do again 82:29 >> yes 82:30 >> but like it there was an immediacy and 82:32 and a visceralness to programming and 82:35 that that that clearly has been lost so 82:37 Lots of things have gotten incredibly 82:39 amazing and I always said like we we 82:41 used to say that um after two years it's 82:44 like someone puts a new in the code base 82:45 no one changes anything we we figured 82:47 out how to avoid those horrors. I think 82:50 we have a we have a very mature um 82:52 discipline. We are um we've always as 82:56 programmers I'm an engineer too. We've 82:58 always had like a identity crisis. We 83:01 always wanted to be someone else. Um um 83:04 we have significant science envy. Um we 83:06 we we even attempt to call it computer 83:09 science. Um 83:10 >> there are some people who do computer 83:12 science in the machine learning world. 83:13 So it's it's a real thing. But clearly 83:15 that's not 83:15 >> it's a vocational art. 83:17 >> It's not most computer science degrees. 83:19 Yeah, and this is also how we eliminated 83:21 things like uh aesthetics and beauty and 83:23 stuff is from our code bases or like 83:25 like from talking about it. They are 83:26 beautiful code bases with beautiful 83:28 code. It's like this code that like 83:31 tells a story and clearly communicates 83:33 what it's there for to to to everyone 83:35 who reads it ideally in is is is an 83:38 artwork of sorts. And so yes, I think we 83:40 have aggressed. I think we have also 83:42 advanced, but I think we don't have the 83:44 clear um vision to drive 83:48 um simplicity. And I think I I I 83:51 personally really deeply appreciate and 83:54 have hired always prioritized hiring 83:56 electrical engineers um into programming 83:58 roles 83:59 >> just because they just have a better 84:01 idea for the costs of inefficiency, 84:04 waste really. um you know like they like 84:06 you put too many chips on a thing, it's 84:07 too too expensive to manufacture, you 84:10 will just never sell it, it's not a good 84:12 product. Um where the zero marginal cost 84:15 of copying software just means all sins 84:17 are forgiven. Um and and in effective 84:19 code still runs and does the thing, 84:20 right? Like so 84:21 >> you really have to fit the whole system 84:22 in your head as an electrical engineer. 84:23 >> That's right. 84:24 >> And so anyway, so um I I think we're 84:26 immature. 84:28 >> We don't know what grade is. um um we 84:31 leave a lot of this up to people and 84:34 therefore the outcomes are so different 84:36 um depending on who's on the project. I 84:38 mean again we all getting more 84:39 conscientious about it and I think we 84:41 are in a counter swing to the sort of 84:44 horrors of architecture astronauts that 84:47 like have caused a lot of damage to I 84:49 think our our industry over the last uh 84:52 like especially like in the 2010s. So we 84:55 we we're getting in a better spot and 84:56 then I don't know what AI it was but the 84:58 but interesting thing is it I don't even 85:00 think it's going to end up mattering 85:01 because AI is going to solve all 85:03 technical depth. So it's like maybe 85:05 engineer you're right. 85:06 >> Exactly. Yeah. 85:07 >> Same question about on-ramps to 85:09 entrepreneurship. You know as a kid you 85:10 were selling things to your classmates 85:12 and what I observe is a lot of 85:15 entrepreneurs have the same origin 85:17 stories where they were entrepreneurial 85:20 early either something like that where 85:22 they had some business in school super 85:24 common as Patrick did web development 85:26 and you know when you're a kid getting 85:27 paid thousands of euros to build a 85:30 website is like infinite money I didn't 85:32 know such large amounts of money um 85:34 exist um and obviously to a small 85:35 business you're like wow I'm totally 85:37 taking advantage of this kid from 85:39 and you know few thousand euros and then 85:42 another common one seems to be drop 85:44 shipping arbitrage you know this whole 85:45 thing of like buying on AliExpress and 85:47 selling on eBay or something like that 85:48 but you maybe have a window into this 85:50 with Shopify or something do you have a 85:51 sense of what the right like if a 13 85:54 14year-old is listening to this and 85:56 interested in making money what are the 85:59 areas where you think are kind of good 86:00 to to pursue today 86:01 >> yeah I think just these reps are the 86:04 important thing I the mechanisms change 86:06 and there's like entrepreneurship clubs 86:08 are really like using Shopify a lot and 86:11 like it's it's I love that and I this is 86:13 one of those areas I would love to go 86:14 much deeper on like just because it's 86:16 like such a it's it's it's it would be 86:18 so fun to really really um uh you know 86:21 build directly for them. The super 86:22 important thing is just go through life 86:24 like understanding that like everything 86:26 around you is built. I you you wrote one 86:28 of my favorite tweets of all time 86:29 probably like actually maybe the high 86:31 water marks like on on this topic like 86:34 >> Museum of Passion Projects. 86:35 >> That's right. And I I love that because 86:37 it's like if you can teach that to 86:40 students it's worth more than most of 86:42 the entire curriculums of uh subjects. 86:45 And so I think the commonality behind uh 86:48 um entrepreneurs is that very early in 86:50 in life um they somehow something 86:54 happened that told them that the world 86:56 is dynamic and it changes. We some of us 87:00 I had the enormous advantage of living 87:03 exactly sort of at the time when we went 87:06 of computers to no computers like in 87:08 homes and like I had one early and lots 87:11 of my neighbors didn't so I had a 87:13 counterfactual there of and being just 87:15 absolutely fascinated with them and then 87:17 networks came in my teens internet um so 87:21 I had a front row seat to a world like 87:24 like the the world around me in the 87:26 small sleepy town in Germany 87:28 um was irreconcilable with what was 87:31 becoming possible like it like it was so 87:33 clear that so much would change and so I 87:35 think that dissuade me of sleepwalking 87:38 through world it just made me like this 87:39 like okay like 87:41 >> that's fun like I'm sitting on a piece 87:43 of information 87:44 >> yes 87:44 >> that uh is uh everyone around me doesn't 87:46 believe yet um but is clearly true and 87:48 I'm going to bet on and so I'm going to 87:51 take reps looking at things around me 87:53 and figuring out how they could be done 87:54 better and like that process is uh I 87:58 think the most valuable process to 88:00 cultivate as an young adult or teenager 88:02 and um 88:04 >> so the meta process is more important 88:05 than the actual domain you choose. I 88:07 >> I agree. Yeah. Absolutely. I I think if 88:09 you spot opportunities, see what you can 88:11 do with it is is a really important 88:13 thing. I think we overemphasize 88:15 um those stories and deemphasize 88:19 the process that causes this. It's it's 88:21 building the habit of looking at things 88:23 and thinking about uh a do I know how to 88:25 make a better vision of this which um 88:28 you should do anyway. 88:30 >> Um and then we definitely underestimate 88:32 and under celebrate and underlineize um 88:35 >> the Beatles Hamburg days 88:37 >> where we're going to and acting on this 88:39 impulse would mean I'm going to spend 88:42 the next 5 years of my life doing this 88:44 and I'm willing to pay what that costs. 88:47 And for you, was that just doing a lot 88:48 of programming as a teenager? 88:50 >> Programming is how I just like made 88:52 things different. And uh um you know, 88:55 like I it's like I went to these 88:58 classical language schools. I was bad at 88:59 Latin. Um we we started with Latin. I 89:01 only got English in like an eighth 89:03 eighth grade. Um hence my accent. 89:05 >> But your Latin accent is actually 89:06 >> Yeah, I imagine. Well, no, because no 89:08 one knows what how that actually sounds 89:10 and it's definitely not German. So So 89:13 everyone just like it goes. Yeah. 89:15 Anyway, I I I just hated doing the 89:17 vocabulary tests. I was bad at it. So, 89:19 at some point, I wrote software for 89:20 teaching myself. Basically, similar to I 89:23 suppose what's called space repetition 89:24 now, but like just like it's like there 89:27 little basic shell script, right? Like, 89:29 and it's stupid, but like something 89:32 about making this and then using it made 89:35 it interesting enough that I did it. And 89:36 so, like it's very clear moment there 89:38 like went from I think I got a D on a 89:40 test which was like starting to be 89:42 problematic. I just like A's for rest of 89:43 my life, right? Like it's just like that 89:45 is a crazy experience right like this is 89:47 a thing that didn't exist that like fits 89:49 me personally maybe my first internal 89:50 tool so um uh yeah it changes you 89:54 >> exactly it made you passionate about 89:55 building your own software you mentioned 89:58 uh space repetition the other 90:00 interesting you know uh educational 90:02 topic dur mastery learning and the bloom 90:04 two sigma effect how do you put that 90:06 into practice do you with your own kids 90:08 >> yeah so um my kids um I have three boys 90:11 um 15 14 11 and they go to a school, a 90:16 boy school, which uh we are quite fond 90:18 of. Um and um I mean they're my kids. 90:22 They're quite nerdy. So like it's just 90:24 like I met them. 90:25 >> Yeah. Yeah. So I love them to bits. I 90:27 can't wait to go home. Um it's such a 90:29 fun time. Uh everyone should have kids. 90:31 Um and um the particular mechanisms by I 90:36 I don't I don't know. It's like the most 90:39 important principle of uh like parity 90:42 principle I have is everything's 90:43 interesting. This is a sentenc. 90:48 No one is allowed to say they're bad at 90:49 something. They're only allowed ever 90:50 allowed to say um they are not good at 90:53 something yet. Um so that's basically 90:56 the entire set of principles. 90:57 >> We're not good at gross margins yet. 91:00 >> Exactly. Yes. We choose not. Yes. Um you 91:03 know and making it a joke is funny too, 91:06 right? Like it's just like it's it it's 91:08 I want them to have those things in 91:09 their prior because like it's it's 91:11 really easy to say something is 91:13 annoying, but like it's much much harder 91:15 to figure out what you would have to do 91:17 to make it interesting for yourself. And 91:19 I think that's there just isn't anything 91:21 that is not actually interesting. I I 91:23 you spend a long time in banking. It 91:25 doesn't sound terribly interesting to me 91:26 initially, but like I've learned enough 91:28 about it that I like I've read books 91:29 about double entry accounting and the 91:31 history thereof and it's just like 91:33 absolutely fascinating. Did we recommend 91:35 the Calamirus book to you? Um, Fragile 91:37 by Design, which is a history of various 91:39 countries banking system. 91:40 >> I'm literally just reading it. 91:41 >> Oh, you are? Okay. The Canada chapter is 91:43 particularly good. It's like a history 91:44 of the Canadian banking system. 91:46 >> Like very fond of can Canada right now, 91:48 which is like um need some new I think. 91:51 >> Yeah. Yeah. Well, just I found it 91:53 interesting how it kind of explains the 91:54 path dependence of how each country gets 91:56 to where it is. Speaking of Canada, 91:57 what's your advice for Mark Carney? 91:59 >> Yeah. I mean, it's like funny. Our 92:01 partnership is very very uh storied and 92:03 layered and somehow your board members 92:05 know my prime minister. 92:08 So that that's fun. Look, I love Canada. 92:10 I live now more in Canada than Germany 92:12 uh for long for longer. I think it's a 92:14 it's a beautiful project. It's a 92:16 beautiful implementation of different 92:17 implementation from United States or 92:18 from Europe of classical liberalism. Um 92:21 it's very much ve into um socialism 92:24 under under Trudeau. He pushed that very 92:26 very hard. Um um I 92:31 am very much hoping that uh Mark is 92:33 helping all of Canada to figure out that 92:35 that's a successful discovery of 92:37 something that doesn't work because 92:39 obviously that kind of pushed socialism 92:41 always leads to poverty. It's like 92:43 >> problem socialism is eventually run out 92:44 of other people's money. 92:46 >> Exactly. And so it's just like and it 92:48 doesn't need it because it's such an 92:49 incredibly like I I mean I I built K's 92:52 largest company now at this point from 92:54 by market cap as an immigrant. Um I 92:57 think the advantage I had as an 92:58 immigrant is I wasn't sort of aware that 93:00 we are supposed to um uh you know make 93:03 small plans so I made big plans um you 93:05 know just like and held everyone 93:06 accountable to that and turns out you 93:08 can build worldass companies if you do 93:10 that um and um there doesn't need to be 93:12 any kind of inferiority like they are 93:15 and also list like on on from first 93:17 principles is like it's just like I mean 93:19 first of all it is attached to like next 93:21 door to the best consumer market on 93:23 planet earth until 15 minutes ago it was 93:25 we good friends and um it's uh um you 93:29 know it it has in the ground basically 93:32 as much like it's number one two or 93:34 three on every single resource uh that 93:38 is of value in in the in in the next 93:41 manufacturing buildup age. So like like 93:43 we should probably get on extracting 93:45 some of those resources and do 93:46 something. I think it can be the richest 93:48 country in the world if it if it chose 93:50 so and um it it just needs to make a 93:52 decision that it wants that and um I I'm 93:55 very optimistic about Mark um you know 93:59 pushing into that direction and I I hope 94:01 he's going to push very very far because 94:04 I think that's 94:05 >> there's no there's no advantage of 94:07 waiting. So taking economic growth and 94:08 GDP growth seriously which has obviously 94:10 plateaued somewhat in in Canada in 94:12 recent years and then being willing to 94:15 make the potentially controversial 94:17 policy decisions and kind of changes 94:18 that are required to actually enable 94:19 that. Well, I mean the things that are 94:21 controversial policy decisions are not 94:23 are just like um controversial for path 94:25 dependence, not for for not real like 94:27 it's like people like Canada has sounds 94:29 crazy to tell people but has like this 94:32 incredible heartburn over over pipelines 94:35 which are like things that go 94:36 underground next to highways and just 94:38 like it's just like no one knows that 94:40 there's a pipe 94:41 >> like the only thing a pipeline does is 94:43 you don't need diesel trucks. That's the 94:45 only thing and you can save a lot of 94:48 >> Yes. 94:48 >> diesel. Um um it also changed the cost 94:51 basis and you like Canada has all these 94:53 resources and they need to somehow get 94:54 to uh to ports and so it just like feels 94:57 like these like tiny little 94:58 implementation details. It's it's like 95:00 as if computer manufacturers would 95:01 somehow get down on copper cables, 95:03 right? Like it's it's it's a weird thing 95:05 and um 95:06 >> um so so depoliticizing just like 95:08 implementation details that 95:09 professionals can do. It's odd because 95:12 >> Canada for instance doesn't have his own 95:14 nuclear power. Like there's like where I 95:16 live in Ontario, it's like the grid is 95:18 entirely clean. It's like the Bruce 95:20 nuclear power plant which I constantly 95:21 build more reactors for which is like 95:23 every time under budget and before 95:25 schedule like six, seven gawatt now and 95:29 just humming and then the rest is it's 95:31 the cleanest grid in the world. 95:33 >> Yeah. 95:33 >> And so like 95:35 >> the funny thing is it doesn't even need 95:36 it because it has all the other stuff 95:38 too that that could just use but like we 95:40 good at this too. So like I think 95:41 there's like the blueprint is obvious. 95:43 What I actually So my my main advice to 95:45 him and like is is always like I I want 95:47 him to um in fact ideally talk to you 95:50 about the Shannon economic zone because 95:52 I feel like it's like the blueprint for 95:53 success in Canada because we can I think 95:55 we can build data centers or 95:57 >> spaceports and stuff like it's much 95:58 faster. And 95:59 >> my my hope is that Canada can build up 96:02 an economy and and a a meritful uh 96:06 market and so on. uh which makes it like 96:08 can can get back to doing what K does 96:11 best which is like succeed by helping 96:13 America succeed and uh you know just 96:15 like work together and let's go. 96:17 >> Yeah. And as you say energy could be a 96:19 big part of that. As you mentioned 96:20 nuclear obviously think about hydro 96:21 where obviously Canada has a ton of 96:22 hydro already like you were saying the 7 96:25 gawatt nuclear um power station but I 96:27 think lary is a 7 gawatt project and 96:30 could be a lot more. I think there's 96:32 other phases that we're not done. 96:34 Absolutely massive on every single angle 96:36 you look. It's it is every single time I 96:38 learn about a new resource that we will 96:40 probably need a lot of, if I look, it's 96:41 like just like there. 96:43 >> Yeah. 96:43 >> Like it's just unexloited. 96:45 >> Yeah. Um maybe last question. Um we have 96:48 your uh Daytona car here. What is 96:52 >> the draw of motor racing to you? 96:56 >> Yeah, I I I mean I just love it and it's 97:00 I'm surprising to me. It honestly is 97:03 it's this is one of the weirder weird 97:05 >> sorry is or isn't surprising. 97:06 >> It is surprising to me. So it's like you 97:08 know so so I I 97:10 >> I I love like kind of figuring out my 97:13 own limitations or propensities like 97:15 just like I think it's like some it's a 97:17 game everyone is playing I think to some 97:18 degree maybe less conscientiously. 97:21 >> I uh try on every hobby that anyone I 97:24 admire gets into because it's like maybe 97:27 I'm interested in it too. I wouldn't 97:29 know. Like again everything's 97:30 interesting. motorsports was like not 97:32 really something I high likelihood or 97:34 batting average on on liking because I 97:36 didn't even have a driver license after 97:37 coming to Canada um after my German one 97:39 expired and um so so a couple friends 97:42 took me to like I had no went to the 97:44 track and I did went out and I just like 97:46 immediately fell in love with it. It's 97:47 it's there's something just incredible 97:50 and I love competing against myself. I 97:52 guess 97:53 >> I think that's that's the thing I I just 97:55 like I race in races and um I don't even 97:58 try to win. I'm I'm like I'm I still 98:01 competing against myself and I see all 98:02 the other cars as obstacles. Um uh and 98:05 um so there's just like this beautiful 98:08 thing about it's technical and you 98:11 connect it with this like four uh uh 98:15 basically playing card size patches of 98:18 rubble to the ground and there's only so 98:20 much grip to get out of them and like 98:24 trying to get to like 99.9% of like 98:27 lateral acceleration or turning. um 98:29 understanding the weight transfers of 98:31 the cars and all these kind of things 98:33 and and and and then trying to get to 98:34 the best lap time you can and like the 98:36 the difference between that lap time and 98:38 the theoretical possible lap time. It's 98:40 just the you know exact report card of 98:43 all your inadequacies as a driver and 98:46 like next lap you try to do it slightly 98:47 better and better and better and just 98:49 hone this craft. It's like just like the 98:51 mey and the the just like it's it's a 98:54 it's just a brilliant uh thing which I 98:56 discovered for myself. So, so started 98:58 taking SE more seriously over a couple 99:00 years and um now uh like did the toner 99:04 this year and uh hopefully do the big 99:06 24-hour 99:06 >> which is a 24-hour relay race, right? 99:08 Like you swap with the other drivers. 99:10 >> Yeah. Yeah, that's right. So So one car 99:12 24 It's actually um if you watch the 99:13 Formula 1 uh uh cinema movie that that 99:16 starts at that race, which was also 99:18 really fun to see. It's like I I've done 99:20 the getting up in the middle of the 99:22 night, walk exactly down that first 99:24 scene where Brad Pit walks like down to 99:26 the pits and um 99:27 >> oddly all pit stall was even exactly the 99:30 one you went to there like um in in the 99:32 movie and then uh get in the car and 99:34 you're very very tired in the middle of 99:35 night. So I have to say 99:37 >> I found out recently they sold 99:38 sponsorships for the uh fake F1 teams in 99:42 that movie and made a lot of money on 99:45 >> incredible bit of product placement. I 99:47 mean like if you think about it from the 99:48 one itself is like such a madeup thing, 99:50 right? Like it's like it's so synthetic. 99:51 It's like we just decided to uh create a 99:54 document which is a set of rules. 99:56 >> Yes. 99:56 >> And somehow because that document 99:58 existed like 99:59 >> 20 teams move cars around planet which 100:01 is like also one of the bull cases like 100:02 like I I don't even believe in 100:05 >> AI unemployment from to begin with. It's 100:07 not how things work. But if you were 100:09 worried about us, 100:11 I think F1 is kind of proving that we 100:13 can just make up completely incredible 100:15 pursuits, especially with 100:17 >> AI and robotics. We can make it cheaper 100:18 and then like let's go. 100:20 >> Yeah, there's a great I mean the history 100:21 of F1 is super interesting. There was a 100:22 great Colossus podcast on uh the history 100:25 of F1 describing basically the Bernie 100:26 Ecklestone era which came with you know 100:28 the the emergence of TV obviously and 100:31 there's even things like the different 100:33 race schedules were not standardized and 100:35 there was like multiple like not every 100:37 team would show up to every race or 100:38 whatever and so he really kind of 100:40 heranged everyone into kind of a 100:41 standardized thing but then it feels 100:43 like the modern era especially with 100:45 Netflix I think they very successfully 100:47 internalized that they're selling an 100:49 entertainment product and some people 100:51 will watch every race every weekend, but 100:54 like I'm not going to watch every F F1 100:56 race, but I will watch Drive to Survive. 100:58 Like that is an appropriate short form 101:00 amount of F1 for me. And so I think 101:02 they've actually very sav internalized 101:04 that they're F1 is an entertainment 101:07 product and uh it needs to be kind of 101:09 consumable in all the appropriate uh 101:11 form factors 101:12 >> and and they just gave access to the 101:14 sort of intrigue and politics and and 101:16 and just characters behind it. Like 101:18 Formula One drivers are crazy people, 101:20 right? It's a nutty thing to basically 101:21 drive a billboard around at 350 km now. 101:24 So like the characters are larger than 101:26 life and like absolutely like driven. 101:28 They're almost 101:30 >> zero sum competitors you can no one's 101:32 there to just competing against 101:33 themselves. And part of how it works, I 101:34 remember hearing Zack Brown, the uh 101:37 McLaren team principal, talk about how 101:39 uh Netflix is everywhere and you know 101:41 the paddocks and around and things like 101:42 that. And he was saying, "Yes, you're 101:44 aware that you've like signed a thing 101:45 that like Netflix microphones will be 101:47 everywhere, but they're really savvy 101:48 where they'll have the like in a spy 101:50 movie, they'll have the directional 101:51 microphones trained on you from like a 101:53 super far away distance and you just end 101:56 up forgetting that like you're on a hot 101:57 mic your entire day while trying to 102:01 actually manage a race." And so I think 102:03 they get authentic content because yeah, 102:06 in theory people know that they're on 102:07 camera and on a microphone all the time, 102:09 but you actually just can't live your 102:11 life that way. You know, you just forget 102:12 and so they actually get kind of 102:14 interesting content. 102:15 >> I completely agree. And it's like the 102:16 season is long. They do many races. It's 102:18 actually amazing what they do. Um, you 102:20 know, like very fantastic athletes, they 102:22 lose like five, 10 pounds like some of 102:24 these 102:25 >> Singapore night race where it's so so 102:27 it's pretty amazing and like it's it's 102:28 it's a fantastic sport. I actually love 102:30 that what Netflix did because like it I 102:32 mean Liberty Media is one of those 102:33 companies you kind of have to like Jer 102:35 Malone um like I mean Cable Cowboy is 102:38 already one of a great book. 102:39 >> It's such a great book and then like 102:41 >> you know he wrote a book he's like it's 102:42 coming out soon imminently. 102:44 >> I did not know it. 102:44 >> Yeah. He has an autobiography so 102:46 >> that'll be that'll be fire. 102:47 >> Yeah it'll be really good. I agree. Well 102:49 look thank you Toby. 102:51 >> This is fun. 102:51 >> This is super fun. So 102:52 >> it's really really fortuitous. So you 102:54 sort of an Irish pup just popped up in 102:56 your office. Yes.