Where AI Value Actually Goes (Hint: Not Model Companies)

Transcript

So, net engineers are being hired not fired in aggregate. If you look across the entire ecosystem. So, you're not seeing software engineer in displacement from AI. That's one. Second, the companies that are laying off the most engineers are spending the most money on Nvidia and they're issuing debt to finance their cash flows. That's Oracle and Amazon and Meta. This is all about free cash flow. They got to figure out a way to pay the bills. Uh so, yeah, this is not about AI displacing uh human talent. That's not happening yet.

So, I'll say the part of this that did catch my eye and was super interesting to me is actually what Bezos is up to cuz I know one of the topics Ram you've talked about on here before is where will the value accrual from AI ultimately reside in the market? Does it reside with the model companies or does it reside elsewhere? And so, as I look at what's going on here, if you have a guy who built one of call it the two biggest manufacturing behemoths out there, which was Amazon, right? In terms of operations, logistics, supply chains, moving goods from A to B, maybe the other one recently being Apple, which is fundamentally a hardware company itself,

and he's looking at this saying act the place I think we'll have the value accrual is applying AI to manufacturing companies are stuck previous generation regimes. That's a super interesting signal to me. Because remember his original insight back in the day was wait a minute. Why don't I apply the AI to this fragment or not AI the internet in that case to this fragmented sort of idiosyncratic, not particularly efficient market called books.

Right? And brought Amazon into the world to massively compress margins and improve things in that space. Is he looking to run that playbook again on manufacturing writ large with AI? That becomes a question to me as he's unveiling this sort of thing. So, Chris, you're an investor in early stage companies. Ram, you're an investor in later stage stuff. Is this a sign that the value accrual of AI is not going to go to the model companies, but rather to the people implementing it to improve their own margins?

Yeah, man. I mean, like the the benefits are there. It's the users, the end users. The the venture capital firms are subsidizing the real economy. Financial services firms are benefiting. Uh we had someone on our team use Claude co-work to pre-fill 20 sub-docs with a command the basic command. Like those are the people that we're all benefiting. Claude is losing money. They're handing value to end consumer because they're gaining so much share from OpenAI. People are cord-cutting OpenAI and switching over to Claude. We've done the same. And so, it's all this venture capital money is is subsidizing the real economy. It's not this esoteric innovation like some transformation in biotechnology.

It is basic day-to-day workflow automation, research. We're using it in our investment processes every single day. I'm sure everyone is. It's uh speed of memo generation, investment proposal generation. So, that's where the value captures. You know, in capitalism, the consumer wins. Same thing is happening here. I would say on the on the early stage venture side, you know, we always are looking at the intersection of crypto and AI. And the this this space has a lot of promise.

Um projects like uh BitTensor um and and others um one thing that we're watching is, you know, things like payments and, you know, we had Coinbase launch this XPRL 2 protocol standard. Is Stripe release its equivalent MPP. We haven't seen that take off just yet either either one of those. Um and one of the things crypto has struggled with for a long time is, you know, it's not if you build it, they will come. I think there's some of that going on in AI as well. And so, really looking to see, you know, some of that real product market fit at that intersection. You know, I think it's going to happen. I think agentic payments is going to be big. We haven't seen that evidence yet. Um but that's something I'm I'm I'm really really

focusing on. All right. So, if we're agreeing and Ram, I agree with you that the history of capitalism has been the greatest benefits ultimately get delivered to the end consumer on that front. That like largely a factual statement. And that at least some of these announcements from firms then are in some ways call it presaging other concerns or like avoiding the real reason for the layoffs. Do we think Zuckerberg building an AI agent to skip the org chart is going to be a leading indicator for people in middle management both at Meta and perhaps writ large that your days are numbered because if like Ram, to take

your example, if you're eliminating a lot of the busy work, you're freeing up those employees to call it manage the work that is being done there and then communicate directly about that sort of thing. Are we going to be compressing org charts, eliminating the middle management people and eliminating a lot of the repetitive work and leaving the people who are architecting and designing? Is that the source of these layoffs? Yeah. Yeah, well, you know, one just orient the audience. I thought the headline coming out that Zuckerberg wanted to create an AI executive to help him do his jobs is amazing.

Zuck wants to create an AI robot to run Meta. We already have a robot running Meta. So, this is more evidence that we live in a simulation. Uh but yeah, look, I think uh you're going to see the more of the menial tasks go away. That's always the history of this stuff, right? The the toll booth collector gets disrupted. The the switchboard operator gets disrupted. They move on to a higher value added, more meaningful, more satisfying uh work life. You know, there's someone in intensive care unit who every 30 minutes must rotate a body. That person's going to get a higher quality life and it's going to lower the cost of care and delivery. That's a menial, thankless chore that's essential.

Uh so, uh I I I don't think you're going to see this massive disruption. Like who whose road map is going to suddenly evaporate? Have you ever seen technology come be a road map say, "Okay, we're good now. We wrapped up the road map. We shipped our product. We are good to go." There's always new innovation. Uh so, you know, you will still have humans playing the role of setting vision, AI accountability, AI supervision, uh quality assurance, uh listening to the customer. Those types of roles will and work functions will take up more of the time.

Chris, what do you think? Is this bad news for middle management at Meta? Uh you know, honestly, people are resilient and I I don't don't bet against people. They tend to adapt very thoughtfully to innovation and technology. When I was growing up, we didn't have like influencers. Um now that's that's a thing. Um so, I think Ram's right. You you'll have menial tasks, you know, maybe travel agents. But then it all comes back, right? And it comes back in a different form. We're super high touch. It you know, the people will want that human connection. They want that super high touch person-to-person feeling for certain types of things. Once you get rid of it, um I I I'm not fussed, but people are super adaptive. The economy

will shift. Um birth rate's going down anyway, so it's going to reach a natural stasis. Um and I think it's going to be very complimentary towards civilization going forward. Yeah, what's interesting to me is the skill sets that it pushes forward because I have a friend for instance who's a founder in the AI space um working on a product that does signals actually for supply chain disruption. So, wow has that been timely lately by the way? Echo Labs if you're in that space. But we were just talking today and she was telling me, "I don't think I'm the best engineer in the world, but the job now is becoming more about architecture, more about high touch customer service,

more about understanding how to communicate and deliver things for your end clients." And she was saying, "This favors me. If it was just a fair game of engineering skills, I'd have gotten my ass beat by some 12-year-old in China." Right? To paraphrase. And I think the one thing we're not going to be able to replicate with AI is human interaction. Right? Like even if you have a robot doing your menial tasks, can it replicate the actual act of interacting with another human being? Or definitely, how soon will the three of us be replaced by AI to do this podcast?

Well, Yeah, and then what do you study, right? You study computer science like the last generation, you know, needed to do? Do you study a humanities now to be more effective in the workforce? You know, I think the easy answer is, "Oh, now you got to be a humanities major because they know how to think." I think that's I think you just have to be a critical thinker in whatever discipline that you want and that will help you inform and how to best partner in your discipline with the AI. Uh I'm not too concerned about particular discipline losing out. I do get worried about, you know, this decline in population and just replacing more and more people with robots.

It's easier to have more conflict and war when you've got robots that are soldiers. There's no real cost. And uh I you know, I I I hope we vector towards a society where you continue to have humans in the loop. But if if you're having more of this automated lethality lethality, then the cost of conflict declines. You know, so that that's my concern. Like when you have population growth, you have more people to have a stake in the system.

You know, people in China don't want to see their only child go to war, for example. But Lindsey Graham, who's a boomer, loves seemingly lots of different wars. Uh you know, he's not someone with kids that he's going to see presumably con- conduct those kinds of wars. So, you know, that's the downside from from all of this is you know, you you lose the you lose the cost of the side. You know, there's a there's a period where Mark Zuckerberg he made sure to harvest, hunt, or kill all of his food. He made a direct visceral connection with everything he consu- consumed, right? So, if he was going to have chicken, he took the chicken and he he killed the chicken. I'm not saying

people should do that, but you're going to see more mechanistic societies like that evolve, where why wouldn't you just con- con- continue conflict if the cost of it is measured in $8,000 robots at scale, and wouldn't that increase the likelihood of actors like China then investing in robots uh engaging in conflict also? So, you know, I think that's the that's a risk, a real risk that we need to consider.