0:02 The biggest moat is going to be which
0:04 companies understand something that's
0:05 super hard for other people to
0:07 understand. And if your answer to that
0:09 is I don't know, then you maybe could
0:10 get vibe coded away.
0:12 >> Block was one of the first to make a
0:14 pretty drastic decision in cutting 40%
0:16 of the workforce. What led up to that decision?
0:17 decision?
0:18 >> There's been this correlation between
0:20 the number of folks at a company and the
0:22 output from the company for decades and
0:24 decades. I think that basically broke
0:26 and what we were seeing is that one or
0:28 two engineers who was on the tools is
0:32 able to be 10 20 100x more productive
0:34 over time. It's like pretty obvious that
0:35 these systems are just going to be so
0:37 much better than like having a thousand
0:39 humans who are doing that work. I I do
0:41 believe that fundamentally for a given
0:44 product or for a given road map, you're
0:45 going to need fewer engineers, fewer
0:47 designers, fewer PMs. I think that's
0:49 like very very clear.
0:51 >> So you show up on Monday, 40% of the
0:52 company's gone. What's the most
0:53 meaningful difference in how you're operating?
0:53 operating?
1:00 what does it actually look like for a
1:02 large public company to restructure
1:04 itself around AI? Owen Jennings is the
1:06 business lead at Block where he oversees
1:09 product operations and customer support
1:12 across Square, Cash App, and Afterpay.
1:15 Before this role, he was the CEO of uh
1:17 of Cash App during its critical scaling
1:20 period. And recently uh block executed a
1:22 roughly 40% reduction in force and
1:24 they've been pretty candid about AI
1:26 being a critical component of that
1:28 decision. Owen has gone through the AI
1:30 transformation at scale across product
1:32 lines and business units and so we're
1:34 going to dig into the that decision
1:37 around the riff how block has adapted
1:38 the current and future state of the
1:41 business. So thank you so much Owen.
1:51 Awesome.
1:54 Um, so you know, Jonathan, I think did
1:55 an amazing job kind of setting the
1:57 stage, you know, for this conversation.
1:58 Uh, you know, talking about how
2:01 important it is to be founderled.
2:03 Uh, you know, Block was one of the first
2:05 to make a pretty drastic decision in
2:08 cutting 40% of the workforce. Um, maybe
2:11 walk us through kind of what led up to
2:13 that decision and how you thought about
2:16 it. Sure. I I think I would pro it
2:17 probably starts two or three years ago.
2:20 I think one thing about Jack is I I find
2:23 Jack to be generally right and generally
2:27 early. Uh sometimes very early. Um and I
2:28 think that's flowed through Twitter,
2:31 Square, Cash App, Bitcoin, etc. And so
2:34 we were pretty early on the agentic
2:36 development side. We actually launched
2:38 Goose, which was the first agent
2:42 harness, at least that I know of, um, in
2:45 early 2024. And that started to augment
2:47 how we approached software development,
2:49 uh, how we thought about internal
2:52 tooling. And I would say that over the
2:56 over that period 24 and 25, it was like
3:00 pretty meaningful progress. Um, and then
3:02 late November, first week of December,
3:04 it was just there was a binary change.
3:07 you basically have Opus 46, you have uh
3:10 codeex 53 and essentially you get this
3:13 shift where I think the the the tools
3:16 and the foundational models were pretty
3:18 good at writing code especially for new
3:20 ventures and kind of like green space.
3:24 Um it became clear almost overnight
3:26 maybe in a couple of weeks that now
3:27 they're incredibly capable working with
3:31 existing complex code bases. Um and so
3:33 there was a
3:35 massive paradigm shift where at least
3:38 from my perspective there's there's been
3:40 this correlation between the number of
3:42 folks at a company and the output from
3:45 the company uh for you know decades and
3:46 decades. I think that basically broke
3:48 the first week of December and what we
3:50 were seeing is that one or two engineers
3:53 or a designer and an engineer who was on
3:55 the tools quote unquote as we say is
3:59 able to be 10 20 100x more productive.
4:01 And so that's really what led us to make
4:03 the the decision a few weeks ago. We
4:06 spent Q1 discussing like what does this
4:08 mean fundamentally? What does this mean
4:09 in terms of how we're going to build
4:10 products, how we're going to build
4:13 software for customers, and then also um
4:14 how we're going to run a company. What
4:16 is it going to mean to actually run a
4:19 company? And we spent Q1 as an executive
4:22 team uh with Jack um working through
4:24 that. Uh and ultimately that's what led
4:26 us to this place where where we we did a
4:28 reduction in force that was you know
4:30 slightly greater than than 40%. And that
4:32 wasn't even uh you know to the to the
4:34 conversation we were just having the
4:35 tools were flowing through really
4:38 meaningfully on the development side and
4:40 so the cuts were way larger on the
4:41 development side. If you think of
4:43 something as outbound sales or account
4:46 management um the cuts were you know
4:48 fairly dimminimous. Um and so that was
4:51 really what we were reacting to. C can I
4:52 push you a bit on this a little bit? I
4:54 mean, Alex when he kind of introduced
4:56 the, you know, the conference uh just,
4:58 you know, an hour ago talked about the
5:02 zer period. Uh, you know, how much of
5:05 the riff was sort of overhang from 2021
5:08 kind of overhiring versus AI and and
5:09 kind of like the product actual
5:10 productivity gain is going to be in the
5:12 business? Like if you look at where we
5:14 were from a from a gross profit per
5:17 full-time employee basis from like 2019
5:19 through 2024, we're basically like right
5:21 in the middle of the pack with all of
5:24 the um uh with all the competitors. Um
5:25 if you look at last year, I think we
5:26 were kind of I don't know second
5:28 quintile or something like that. I think
5:30 it's basically like Nvidia and Meta that
5:32 are ahead of us. Um, and then when you
5:34 look at the composition of what we did,
5:36 if you thought it was like croft and
5:37 bloat and so on and so forth, then like
5:40 this riff would have acred to the
5:42 operational teams and like like that
5:44 sort of stuff. Those were really really
5:46 meaningful cuts on the development side.
5:49 You don't make really really significant
5:50 cuts on the development side if you're
5:52 not seeing a technology and a tool
5:54 that's just fundamentally changed how we
5:57 build. I mean, we're we're like we're
5:58 not writing code by hand anymore. That's
6:02 over. That's done. Um and so so anyway,
6:05 everyone has their narrative. Um it's
6:07 largely not true.
6:08 >> Um so maybe just walk through like
6:10 tactically how did you actually execute
6:12 you know this this transition you know
6:14 culturally you know operationally in the business.
6:15 business.
6:18 >> So I think so we were um the the the
6:20 nice part about this riff uh relative to
6:22 some other you know things that have
6:23 happened at block or at other companies
6:25 is we were coming from a position of
6:27 strength on a on a profitability and
6:29 operating income side. And so sometimes
6:31 when it's really financially motivated,
6:33 you know, the CFO or the CEO says,
6:35 "Okay, we need to do a 16% riff in order
6:37 to like hit this hit this target." And
6:39 um that wasn't the case at all. We said,
6:43 "What should the org look like given how
6:45 these AI tools are flowing through now
6:47 and what we expect to happen in the in
6:48 the coming months and quarters." We had
6:51 some core principles. Um the first one
6:53 was reliability. When you do something
6:54 this size, worst case scenario is you
6:56 have an outage or you go down. So that's
6:58 like P 00 not acceptable at all.
7:00 Obviously, you know, things have been
7:01 great over the past several weeks, which
7:04 is fantastic. Second is building trust
7:07 with customers and um compliance and
7:09 navigating the regulatory environment.
7:11 We all operate in a super complex
7:13 nuanced regulatory environment. That's a
7:14 non-negotiable. We have to make sure
7:16 that we're that we're doing doing right
7:17 there. For instance, like we we
7:19 basically did not touch our our
7:20 compliance team and our compliance
7:22 technology team. Even if the tools are
7:24 there, it's like let's not take any
7:26 risks. And then third was let's continue
7:28 to drive durable growth. So there's
7:30 things that are on the road map that we
7:31 already know that we're building. We
7:32 need to continue to do that. We know
7:34 that it might be a squad of three people
7:36 instead of a feature team of 14 who's
7:38 building that. We want to make sure
7:39 we're continuing to build those features
7:41 and that we're continuing to make
7:43 longerterm bets. And then we built up
7:46 the org from scratch. And in some areas
7:49 like um the regulatory council team or
7:51 the SDRBDR team, the org looked pretty
7:54 similar to how it looked in January. Um
7:56 on the development side, it looks
7:59 completely completely different. Um and
8:01 then you know from a from an execution
8:04 perspective um you know we thought very
8:05 deliberately obviously I've been in the
8:07 company 12 years. A number of folks who
8:09 we parted ways with our friends and
8:11 colleagues for for you know more than a
8:14 decade. um we were in a position where
8:16 we're able to be generous in terms of
8:18 you know the the severance packages that
8:19 we gave. We didn't cut people's
8:22 technology access instantly which can
8:25 suck. Uh we chose to have an all hands
8:27 with everybody at the company. So Jack
8:31 and the executive team were um you know
8:32 looking each other in the eyes and
8:34 explaining this decision and explaining
8:37 the the drivers behind it. And um I I
8:39 think that that it was on a Thursday. I
8:40 think like the Friday, Saturday, Sunday
8:43 there's a lot of shock uh dealing with
8:47 ambiguity. Um and then what we've been
8:50 doing is uh we massively reduced the
8:51 number of meetings we have probably like
8:53 70 or 80%. So I now have time to like
8:54 build and work and it's not backto-back
8:56 meetings. We're also meeting with the
8:58 company every week. So we have like a
9:00 one or two hour all hands with Jack
9:02 every every Monday. It just feels like
9:04 we're we're smaller, we're leaner, we
9:06 have fewer layers, we have larger spans,
9:08 and it's it's been back to building.
9:10 >> So, you show up on Monday, 40% of of the
9:12 company's gone. Like, what how is what's
9:13 the most meaningful difference in how
9:14 you're operating? I don't know, maybe
9:16 it's in the EPD or elsewhere.
9:18 >> Um, I think that there's a there's a
9:20 there's a few different components to
9:23 this. I think the biggest thing is so
9:25 one concern that I have with like how
9:27 some of these org changes might flow
9:29 through the tech industry is that and
9:31 and it gets back to the to the founder
9:33 point. If you're not founder le and you
9:36 don't have the the ability to be bold,
9:37 then you're going to probably take a
9:39 more incremental approach. And so the
9:40 way that that's going to feel is like
9:42 you do a 15% riff and it's like, oh,
9:44 it's fine. and then you do another 15%
9:45 riff and then culturally that's just
9:47 like devastating for your team because
9:48 there's always this like pending riff
9:51 looming looming over your over your
9:53 shoulder. Um this was obviously a
9:55 decision to go in a different direction.
9:57 I think one of the benefits that we got
10:00 from this is like we were already seeing
10:04 a a very meaningful increase in AI tool
10:06 usage especially on the development
10:08 side. This is just a massive forcing
10:11 function. Like if we're building um
10:13 okay, we're building Moneybot and we
10:15 want to roll Moneybot out to 50% and
10:16 there used to be a team of 15 people
10:18 working on it and now there's a team of
10:21 four people plus $2,000 on the tokens.
10:23 That's this is like un unlimited access
10:25 to tokens and you can use fast mode on
10:27 cloud code. Um so now you have four
10:28 people plus the tools. It's like okay
10:31 well you need to have eight instances of
10:32 goose up and you need to shift your
10:35 workflow from sequentially working
10:37 through a PR submitting it getting a
10:41 review making the change to I have 14
10:43 agents who are building PRs on my behalf
10:45 right now and I'm going to context
10:47 switch between all of those and it's not
10:49 just uh on the software development side
10:51 it's for PMs too it's for growth
10:54 marketers too the biggest shift myself
10:56 included I I have you know countless
10:58 agents running right now that I have to
11:00 I have to go check on. Uh it's it's not
11:03 um it's less of a linear workflow and
11:04 it's more of like in the background
11:06 there's 10 or 20 agents who are doing a
11:07 whole bunch of stuff and then I have to
11:09 check in on the work and nudge it and
11:11 change it or what have you and then I
11:13 can commit it to GitHub and I can I can
11:14 get the markdown file. We can put it in
11:16 the source of truth and we can move on.
11:19 >> Yep. So we have a lot of you know public
11:20 companies in the audience. We have a lot
11:22 of founder businesses in the audience.
11:24 Do you expect other companies to kind of
11:26 follow a similar path and and I guess
11:28 what conditions need to be in place for
11:29 that to be successful?
11:31 >> I I I don't I don't necessarily want to
11:33 like I I talked at the beginning about
11:38 um the ground work that happened in 23
11:42 24 and 25 like we built this agent
11:45 substrate goose and then we built a lot
11:46 of tooling at the company on top of it.
11:48 We have an agentic operating system
11:52 internal only called G2 where anyone can
11:55 automate any deterministic workflow. So
11:57 anyway, I think there's work to do to to
11:58 be successful. I would expect many
12:00 companies are doing that work. Some of
12:03 them are incredibly um far ahead than
12:07 than others. Um and so I I I don't know
12:09 what to expect. What I will say is like
12:11 to the extent that I I do believe that
12:13 fundamentally for like a given product
12:16 or for a given road map, you're going to
12:18 need fewer engineers, fewer designers,
12:22 fewer PMs. I think that's like very very
12:25 clear based after like December. Um that
12:27 doesn't necessarily mean that there's
12:29 going to be fewer engineers, designers,
12:32 and PMs in the world. Um, it's like the
12:35 classic Jevans paradox thing where I I
12:36 think that there's probably now just a
12:39 superset of things that that can be
12:41 built. Um, so I don't know, you know, a
12:43 given tech company might be might be way
12:45 smaller, but there might be 50 or 100
12:47 more tech companies or you're going to
12:48 start getting this development working
12:51 in in sectors and and areas where that
12:54 hasn't historically been the case. Um,
12:55 but I I'm not here to to predict the
12:58 future. I'm focused on block.
13:00 >> Uh, fair. you you talked a bit about
13:00 kind of the some of the AI
13:02 infrastructure you build. Maybe you can
13:04 get go in a bit more depth uh you know
13:06 both in how it's impacting the kind of
13:08 technology or I'm also curious about you
13:10 know how are you using AI in other parts
13:11 of the business you oversee ops customer support.
13:12 support.
13:15 >> Yeah. Um so I got asked at a investor
13:18 conference uh last week like how is AI
13:20 like flowing through block and to me
13:22 that it's like asking um how are
13:25 computers flowing through block? uh like
13:28 it's it's a uh fundamental in-built
13:30 thing that has changed on in like a
13:33 binary way over the past 18 months and
13:35 then feels like it changed all over
13:37 again in the past four months. Um so
13:39 I'll break it down into
13:42 internal and then external and how we're
13:44 thinking about our products, what we're
13:46 putting in customers hands and then I
13:48 can talk a little bit about the the
13:49 future and where we think things are going.
13:51 going.
13:53 So on the internal side, I think the
13:56 biggest difference is the shape of the
13:58 of the org. So we used to have kind of
14:00 like a classic hierarchical uh
14:03 structure. It it was functional. Um
14:04 which was great, but it was like fairly
14:06 standard if you like averaged through a
14:09 bunch of medium-sized tech companies. Um
14:12 and so you would have kind of eight
14:14 server engineers, four client engineers,
14:16 a PM, a designer, and you would work
14:18 linearly through your road map. Now we
14:21 have um small squads. So squads of like
14:24 one to six people. Um so meaning
14:25 meaningfully smaller than the other
14:27 teams would be. And we have way more
14:30 flexibility and and fluidity where a
14:32 given squad can work a few cycles on
14:34 this product, get it live and then a
14:38 cycle on this other product. Um which is
14:39 different than how things worked a year
14:40 or two ago where it's like I'm on the
14:42 banking team. I'm going to be on the
14:44 banking team forever. We also have way
14:47 fewer layers. So on the development
14:48 side, I think we probably cut our layers
14:51 by I don't know 50 or 60%. Like on the
14:53 product side, I only have I think two
14:55 layers, maybe three layers in a in a
14:57 couple places and so information is
15:01 flowing um way more freely. I think that
15:04 then in terms of how we actually build
15:06 on the development side, things have
15:07 changed. I think everyone's probably
15:10 seen, you know, every every CEO out
15:11 there is going on Twitter and showing
15:14 their like green dot on on uh on GitHub.
15:16 Um, but that's real. Like all of our
15:17 designers are are shipping PRs. All of
15:19 our product managers are shipping PRs.
15:21 That's not that interesting anymore. I
15:23 think more interesting is that we have
15:26 uh internal tools that are similar to
15:28 clawed code, but they're like more
15:29 plugged into our infrastructure. So we
15:31 have a tool called Builderbot.
15:33 Builderbot is just autonomously merging
15:36 PRs and actually like building features
15:38 to 100%. We've had some fairly complex
15:40 features that are built to 100%. More
15:41 often than not, it's building them to
15:45 like 85 or 90%. and then a human who who
15:46 has a lot of context and understands
15:49 does like the final the final 10%. So
15:51 that feels really really different. The
15:55 ability to go from um to go from idea to
15:57 like this is in the hands of 100,000 or
15:59 a million customers has been compressed
16:02 massively since uh since December.
16:05 Outside of development, I would say most
16:06 of what we're seeing is like anytime
16:09 there's a deterministic workflow, we're
16:13 we're able to automate that. And so
16:16 generally at a at scale tech company you
16:19 have individuals who are working cues.
16:21 Um a lot of that is just being
16:23 completely automated away like from a
16:25 customer support perspective. This is
16:27 not new but you know our chat bots and
16:29 and AI phone support and and whatnot are
16:31 automating a a majority of inquiries
16:33 that we get. And then it gets into like
16:36 um product operations and risk
16:37 operations and compliance operations and
16:40 any sort of decisioning like generally
16:43 um generally the the the models and the
16:44 agents are going to do a better job than
16:46 humans. Right now I think it's critical
16:48 that we have a human in the loop. Uh
16:51 that's like the key kind of buzzword uh
16:52 when you talk to talk to partners and
16:55 regulators and and what have you. Um,
16:57 but over time it's like pretty obvious
16:58 that these systems are just going to be
17:00 so much better than like having a
17:02 thousand humans who are who are doing
17:04 that work. So that's on the internal
17:09 side. Um, on the on the product side, I
17:10 think that
17:11 >> and maybe just catch people up on kind
17:12 of the shape of the business. Obviously
17:14 you have Square, you have Cash App, you
17:16 you made a big acquisition, Afterpay. >> Sure.
17:16 >> Sure.
17:17 >> What do those businesses look like and
17:19 then yeah, how are they kind of changing with
17:20 with
17:22 >> Sure. So um, so we used to operate in a
17:24 business unit structure. So Square used
17:26 to be kind of its own business unit with
17:28 its own CEO. Cash App was its own
17:30 business unit with its own CEO. Um that
17:31 wasn't leading to the right outcome. So
17:33 about 18 months ago, we functionalized
17:35 the company. Just meaning that all of
17:37 engineering rolls up to our head of
17:39 engineering, all of design to our head
17:41 of design, all of product to me. So we
17:44 have a financial platform team that
17:46 spans the entirety of block. We have a
17:48 business platform team that's doing a
17:50 lot of this automation that spans the
17:52 the entirety of of block. And then
17:54 increasingly we're building features and
17:56 products that actually connect the
17:59 Square side, the Cash App side, and the
18:01 Afterpay side. And so naturally you
18:02 you're building technology and you're
18:04 building infrastructure that is not um
18:07 brand specific. And that's actually like
18:09 kind of central to our our overall
18:13 strategy and and and overall thesis. Um,
18:15 but yeah, I mean, CA Cash App went from
18:17 when I joined Cash App in 2016, uh, we
18:19 had just just started to to figure out
18:21 how to monetize and had our first
18:22 dollars of gross profit. And now I think
18:25 Cash App's probably like I don't know
18:28 60ish% of like overall gross profit at
18:30 the at the company. So overall been been
18:31 growing at a healthy clip over the past
18:34 decade. Um but uh Cash App and Afterpay
18:36 have definitely been growing um more
18:39 quickly, but increasingly we're trying
18:41 to think about things from an ecosystem
18:43 perspective and and that's maybe where
18:46 like goose as a platform comes in which
18:49 is we bu we built goose internally. The
18:52 way to think about goose is um it's a
18:53 nod to a top gun or whatever the
18:55 co-pilot thing but the way to think
18:56 about goose is it's a it's an agent
18:59 harness and it's model agnostic. So I
19:01 can run goose on an anthropic model, on
19:04 a on a on an open AI model, on an open-
19:06 source model. There's probably like 120
19:08 models that we have. And depending on
19:09 what I'm trying to do, I'll kind of swap
19:12 out the swap out the models. And then
19:14 that was useful for a human to use, but
19:16 we've built like the agentic layer on
19:19 top. And so now a lot of the automations
19:21 at at block are actually routing through
19:25 the goose agent harness. And um we've
19:27 been able to leverage this across the
19:28 products that we're building. So,
19:30 Moneybot, which we'd like to think of as
19:32 like a CFO in your pocket, but it's
19:35 essentially like a proactive um a
19:37 proactive uh chatbot that can take
19:39 actions on your behalf within Cash App.
19:41 That's built on top of Goose. Manager
19:44 bot, which is roughly a similar thing on
19:46 the Square side, that's built on top of
19:47 Goose. So, it's a lot of this
19:49 foundational work on agentic systems and
19:52 then like the the triggers and the
19:54 underlying data and events that you need
19:56 to power them that's working across the
20:00 uh the entirety of the of the company.
20:03 So, on the on the product side, um I
20:04 think that the the biggest shift has
20:06 really been like we're going from a
20:09 world where for the past 10 or 15 years,
20:11 everyone's used to a static UI, a rigid
20:14 UI. you tap through the UI, everyone has
20:16 the same, everyone's Uber or Lyft or
20:17 Cash App or whatever looks the same.
20:19 That's going to fundamentally change in
20:22 the next like six months. Um, generative
20:24 generative UI is is is here. We're
20:26 seeing it with Moneybot. We're seeing it
20:27 with manager bot as the models get better.
20:28 better.
20:29 >> What what is that going to look like
20:30 kind of in practice? I'm curious.
20:32 >> I think I mean in simplest terms, it's
20:35 like your Cash App should look really
20:37 different from mine. And the reason why
20:38 it's like, okay, well, I get my paycheck
20:40 into Cash App and I'm super into
20:42 Bitcoin. Let's say like you don't and
20:44 you use Afterpay all the time. Great.
20:45 When we open up our apps, that should be
20:47 totally different. That you could
20:49 probably achieve that just through
20:50 personalization. That's not that
20:52 interesting. What we're actually seeing
20:54 and Anthropic had some releases this
20:55 week that are that are incredible. We're
20:57 actually seeing is like I can go into
20:58 Moneybot and say, "How have I been
21:00 spending my money?" And it'll show me a
21:02 bunch of charts and uh and visual
21:04 visualizations where it is actually like
21:06 on the fly generating generating that
21:08 visualization. It's not actually in the
21:10 code itself. So that's really cool. It's
21:13 also potentially a nightmare from like a
21:14 QA perspective. And so we need to figure
21:16 out how you're going to QA all these
21:18 like non-deterministic outputs for for
21:21 tens of millions of customers. But um a
21:22 great example on the on the Square side
21:25 is with ManagerBot. Maybe charts aren't
21:26 that impressive to you, but with Manager
21:28 Bot, let's say you're a you're a uh you
21:30 own a multilocation quicks serve
21:32 restaurant. you say like, "Hey, can you
21:35 build me an app where I can uh manage
21:38 scheduling for these two locations and
21:40 like automatically fire off text via,
21:42 you know, WhatsApp or or Signal or
21:45 whatever to my um to my employees. It's
21:47 actually going to like create that app
21:49 for you. And the the way that that app
21:51 looks and feels is not in the source
21:53 code of the actual application that we
21:55 push to the to the app store." And so I
21:57 think it's um it gives folks way more
21:59 control. It's way more
22:02 uh ultimately I think it'll lead to
22:04 higher engagement. Um I think it'll lead
22:08 to better product and and really I think
22:10 the key thing I I don't think that if we
22:14 ask customers to to like prompt these
22:16 tools themselves, they're going to
22:17 necessarily know the right prompts and
22:18 come up with the right answers. So,
22:20 we've invested massively on the
22:22 proactive intelligence side where what
22:24 we've found, especially as it relates to
22:26 money is like we need to be prompting
22:27 our customers with things that we think
22:29 make sense for them and that's where
22:32 we're creating a lot of the the value.
22:33 >> So, I mean, I think we're all incredibly
22:36 bullish on on kind of the impact of AI,
22:37 you know, in kind of in the way that all
22:39 these businesses run and the products
22:41 you can create. How does that flow back
22:44 to your stock price? you know the the
22:46 business is the stock has been roughly
22:48 flat for I don't know six or seven
22:50 years. interesting sort of reminding me
22:53 >> but the b the business has grown a lot
22:55 you know to your point the gross profit
22:56 per employee has grown you know
22:57 massively like how do you sort of
22:59 reconcile the that that dimension
23:01 >> yeah I think um so so I think you know
23:03 markets are markets are cyclical and
23:04 there's all sorts of things that are
23:09 happening I remember uh in 2021 when our
23:11 stock price was like I don't know 260
23:12 bucks and I was like that was a little
23:15 bit irrational um you can take a a kind
23:17 of longer term mature view and say you
23:19 So markets are voting machines in the
23:21 near term, but they're weighing machines
23:22 in the long term. Just like focus on building,
23:23 building,
23:24 >> you know, David and Jonathan earlier
23:25 talked a bit about kind of
23:27 defensibility. How do you think about
23:30 your own moes at Square? I mean, at
23:31 Block, excuse me. You, you know, you
23:32 talked a bit about the ecosystem. You
23:34 guys obviously have, you know,
23:37 regulatory infrastructure. Um, you know,
23:38 how do you think about, you know, that
23:39 the business overall in that context?
23:41 Yeah, I think in the I think in the near
23:44 term and the medium term, um there's a
23:46 bunch of there's a bunch of modes that
23:48 exist for for block and and we can talk
23:50 about the industry more broadly. I think
23:51 I think distribution and network effects
23:54 are are one of them. I I agree on the
23:56 the catrini piece and uh and Door Dash.
23:57 I don't think anyone's vibe coding Door
24:00 Dash in the next couple of weeks here.
24:02 Uh I like to say like any of us can can
24:04 create a peer-to-peer app in probably a
24:07 week. uh no one's going to vibe code you
24:09 know 50 or 60 million monthly activives
24:11 who are actually using that. So I think
24:14 that that's true. Uh I think um you know
24:18 licenses and and regulatory posture um
24:19 definitely exists. I think hardware
24:22 right now it's like harder to imagine
24:24 how some of the AI tools flow through to
24:25 the to the hardware side. Like you can't
24:29 vibe code a piece of square hardware. Um
24:33 but I I think longer term if we continue
24:34 like if we look at the rate of the
24:36 change and and the change in the change
24:40 I think longer term the key thing that's
24:43 going to make uh a company defensible is
24:46 um the extent to which the company
24:49 understands something that is pretty
24:52 hard for other companies to understand.
24:56 And so we're increasingly building
24:58 toward a world and talking about block
25:02 as an intelligent system itself.
25:05 >> So basic like the the the the way that I
25:08 see this going if we can if you
25:09 extrapolate forward the past several
25:12 months is that ultimately a company is
25:14 sitting on top of some sort of signal,
25:18 some sort of like rich data and and and
25:21 deep insight. Um for us it's like how
25:23 sellers and buyers participate in the
25:26 economy. Um and and most companies I
25:27 think have this thing that they
25:29 understand deeply. And then the question
25:31 is going to be how quickly can you
25:34 iterate to improve that understanding
25:37 over time. And so we're building world
25:41 models internally and externally of like
25:43 understanding who our customers are but
25:45 then also understanding how block
25:47 operates. It's like you can imagine you
25:49 can imagine for any company just like a
25:51 markdown file of like who you are and
25:53 then you need the feedback loop with two
25:55 things. You need a feedback loop with
25:58 the signal which is like what do you
25:59 what do you deeply understand that's
26:01 hard for others to understand and then
26:03 you need a tool like builderbot or
26:04 clawed code or what have you. And then
26:06 you can just iterate through that loop
26:07 over and over and again. It's like this
26:09 is this is what I'm seeing. This is
26:10 what's happening. Great. This is our
26:12 markdown file for for block. These are
26:14 our values. this is the metrics we're
26:16 trying to optimize for. Um, this is what
26:18 we care about. This is what we don't
26:19 care about. And then you have a gentic
26:21 system, so you can just build stuff. And
26:23 right now, you basically you've taken
26:25 that humans used to do that and it used
26:27 to take a couple months to build a
26:30 feature. Um, now it takes maybe a week
26:32 or two and there's still humans
26:34 involved. Pretty clear that in the
26:36 future you'll be able to run that loop
26:38 like I don't know hundreds, thousands of
26:39 times a day and maybe there's some
26:41 humans involved. Maybe not. Maybe the
26:44 humans are more like editors. And so I
26:46 think the the biggest moat is going to
26:48 be like which companies understand
26:50 something that's super hard for other
26:52 people to understand. And if your answer
26:55 to that is is um I don't know, then uh
26:58 then you maybe could get vibe coded away.
26:58 away.
26:59 >> This has been an amazing conversation.
27:01 Thank you. Uh thank you so much for for
27:02 joining us.
27:03 >> Appreciate it. Thanks so much. >> Awesome.
