Chris walked into the end-of-year superlatives question — "If you finish this sentence, 2025 was the year of what?" — and didn't hesitate. AI. He said it like he almost didn't want to, like the word had been worn thin by people who don't actually know what they're talking about. But then he kept going, and the distinction he drew is the one I keep coming back to: not just AI as terminology, not AI as hype cycle, but AI as use cases finally meeting comprehension. The moment, as he put it, when people stopped hearing the word and started understanding the opportunity behind it.
That's a different claim than "AI is big." Everyone knows AI is big. The claim is that 2025 was the year it crossed from abstraction into behavior — the same inflection point Chris compared to 2009, when Twitter and Facebook stopped being things people heard about and became things people built habits around. That shift, when it happens, is worth paying attention to. Not because it tells you to invest in AI — that ship has sailed and the bidding wars are vicious — but because it tells you what comes next, and what comes next is almost never the thing everyone is already fighting over.
The room nobody is fighting over yet
I stopped him when he got to construction. Not because it surprised me — I've been thinking about this for a while — but because it's the cleanest version of an argument I've been trying to make to people around me and not quite landing.
AI needs infrastructure. Not metaphorically. Physically. Data centers, electrical capacity, cooling systems, fiber runs — all of it gets built by humans with tools and permits and general contractors. The conversation around AI investment is almost entirely about software: which model wins, which API gets embedded into which product, which company raises the billion-dollar round. Nobody is fighting over who pours the concrete.
That's the gap. And gaps like that don't stay open forever.
I've been working toward my GC license partly because of this exact logic. The construction industry is depressed right now — interest rates, slow development, the perception that it's an "old economy" play in a moment when everyone wants to be in tech. But that perception is exactly what creates the opportunity. When AI infrastructure spending starts converting into physical building at scale, the contractors, electricians, and project managers who are positioned and capitalized are going to be extremely hard to find and extremely well-compensated. The glamour is in the software. The durable money is in the room the software runs in.
I passed on a real estate infrastructure play a few years back — not because the thesis was wrong, but because I was distracted by what felt more interesting at the time. The opportunity cost on that decision is a number I don't love saying out loud. This construction argument feels similar. The boring version is usually the right version, and the boring version is usually the one you have to convince people to believe in.
What the social media comparison actually means
Chris's comparison to 2009 social media is worth slowing down on, because most people hear it as hype reinforcement — AI is the next social media! — and miss the actual point.
The 2009 comparison isn't about scale. It's about behavior change. In 2009, the question stopped being "will people use social media" and became "how fast can we build on top of the fact that they will?" The infrastructure play at that moment wasn't Facebook — it was everything Facebook needed to work. Servers, CDNs, ad-tech plumbing, analytics tools. Most of those companies weren't glamorous. Most of them got acquired quietly for good money by people who were paying attention to the layer below the thing everyone was looking at.
2025 is the same structure. The question is no longer "will people use AI." ChatGPT has a hundred million users. OpenAI became, as Chris put it, one of the strongest consumer brands in tech almost without anyone noticing it becoming a brand at all — not through ads alone but through utility, which is the only way a brand actually sticks. When people say "let me ChatGPT that" the same way they say "let me Google that," the consumerization question is closed. The infrastructure question is wide open.
PULL QUOTE: "It was really the year where AI met culture in a real way." — Chris, on what made 2025 different from the hype cycles before it
The runner-up that's going to embarrass people
Stable coins is where Chris went for the second place answer, and I want to spend a minute on why that's a more interesting pick than it sounds.
The reason stable coins haven't had their mainstream moment yet is the same reason they're worth paying attention to now: they're not volatile. The thing that made crypto exciting to most retail investors — the price swings, the 10x potential, the story — is absent from stable coins by design. They're pegged. They're boring. They are, as Chris pointed out, the complete opposite of speculation.
Which is exactly what financial institutions want. Stripe is integrating them. Shopify is integrating them. Circle went public this year on the back of a stable coin infrastructure thesis. The use case is cross-border payments, micro-transactions, eliminating the fees that sit between every international financial interaction — not as a moonshot, but as a plumbing upgrade to a system that everyone knows is outdated.
The irony is sharp: the crypto asset class that got the most serious institutional traction in 2025 is the one that promised the least excitement. People got rich watching Bitcoin go from 30K to 100K and felt like they were in the future. Stable coins are going to quietly reshape how money moves across borders and almost nobody making the big gains on them will have a story worth telling at a dinner party. That's usually how the real infrastructure plays work.
I've been in the crypto conversation long enough to remember when NFTs were the thing everyone was certain about. The same energy — everyone hears about it, everyone has an opinion, the use cases feel real and also slightly made up at the same time. Stable coins feel different to me, the way the construction argument feels different: less exciting, more structural, which is usually the tell.
The one I should've gotten
Two to three years ago I had a seat at the table for a WNBA expansion group. Fifty million dollar valuation. I was committed — ten toes down, as I said on the show — and then some things moved and I let it drift and I didn't stay on it the way I should have.
That same team is now expanding at $150 to $200 million.
That's not a 2x. That's a 3x to 4x in two or three years on a sports asset, which is the kind of return that makes venture investors jealous. And I walked away from it not because the thesis was wrong — I believed the thesis — but because I wasn't detail-oriented enough about the follow-through. I said the word salty on the show and I meant it.
Women's sports is the real business now. Caitlin Clark is not a marketing phenomenon bolted onto an unprofitable league — she's a revenue driver for a real business that also has franchise valuations moving. The NWSL numbers are moving. What Alexis Ohanian is doing with Athlos, turning track into a media and entertainment property, is the same playbook: take an undervalued asset with real underlying quality, build a brand around it, let the business catch up to the brand. It works. We've seen it work. The question is whether you're in the room before the valuation corrects or after.
I was almost in the room. That's the version that stings.
What I'd actually do with this conversation
Three things, for anyone who's sitting on capital or attention and trying to figure out where to aim it:
- Stop competing for the obvious layer and find the necessary one. If you're trying to get into AI companies right now, you're competing with the best-capitalized investors in the world for access to a relatively small number of deals. That's a losing game unless you already have the relationships and the check size. The better question — the one I asked myself after this conversation — is what has to exist for AI to keep scaling, that nobody is currently fighting over. Data center construction. Electrical capacity. Skilled trades at a scale the industry doesn't currently have. That's where the access gap is, and access gaps are where money actually gets made.
- Take the boring version of the thesis seriously before it becomes the exciting version. Stable coins are boring right now. Construction is boring right now. Women's sports franchises are still undervalued relative to where they're going. The pattern is the same in all three cases: real underlying demand, structural tailwinds, and a perception gap that makes the asset accessible. The moment the perception catches up, the access closes. You don't have to be first. But you have to be before the bidding war.
- When you're in on a thesis, stay on it. This is the one that's personal. I believed the WNBA was going where it's going. I had the seat. I let the details slide and the moment passed. The lesson isn't to be more optimistic or more aggressive — it's to be more disciplined about follow-through on the things you've already decided are true. The analysis is usually not the failure point. The execution on the analysis is. Write it down. Set the meeting. Do not assume the opportunity will still be there in six months at the same terms, because it almost never is.
Chris closed the year-of-AI argument by saying the real test is what comes in the next six months, the next year — how fast the use cases compound. I think that's right. But the more important version of that observation, for anyone making bets right now, is that the infrastructure question runs about eighteen months ahead of the use-case question. By the time the use cases are obvious, the infrastructure is already priced in. By the time the AI demand is showing up as visible revenue, someone already owns the building.
Figure out who's going to build the building.