“Anybody who pretends to be certain right now is probably missing the point.” That’s how Nick Shalek, Partner at Ribbit Capital, opened a recent conversation on The Distribution. Ribbit’s portfolio spans Coinbase, Robinhood, Stripe, Revolut, Mercado Libre, and Juniper Square—firms that each built from first principles in highly regulated industries when incumbents couldn’t or wouldn’t move. Shalek sees the same pattern repeating now, at a faster pace, in nearly every sector of the economy. With AI VC investment reaching $258.7B in 2025—or 61% of all global venture capital—there is no denying that capital is following conviction.
Not disruption—acceleration
Shalek’s framing of AI is simple enough. “It’ll accelerate what’s been happening the past 20 years and just add fuel to the fire,” he said.
Three forces are converging at once. The first is the growing power of small teams: what once required tens of millions of dollars and years of development can now be prototyped in an afternoon. Shalek doesn’t think it’s far off before a single person builds and runs a billion-dollar company.
The second force is the primacy of data. “People who have proprietary data or figure out how to create proprietary data through interactions with users are going to be the companies that become the biggest companies in the world,” he said.
The third force is software as interface: intelligence fused into every business, not as a product, but as the medium through which every function operates.
The data moat has no substitute
The SaaSpocalypse framing isn’t wrong, Shalek said, but it’s incomplete. If a generic productivity application can be replicated with AI code tools, that part of the moat has evaporated. But the firms that pair software with a hard-to-replicate data asset—datasets about deals, LPs, and portfolio companies—have something different. The question is whether those organizations are structured to use them.
Shalek used an industrial analogy: when electric motors replaced steam engines in factories, the first-generation use was a direct substitution—one power source for another. The productivity gain was real but modest. The transformational gain came a generation later, when factories realized that motors could be distributed throughout the building, meaning the building itself could be completely redesigned. Assembly lines, distributed workstations, and entirely new layouts became possible only after the shift in assumptions.
The venture capitalist vantage point matters here. Ribbit’s track record was built backing founders who started with a different set of assumptions than incumbents. “The truth was, when you had a big business with hundreds of thousands of branches and a lot of people in a certain way, it’s hard to bake in a new set of assumptions,” Shalek said, drawing a direct parallel to what’s happening now with AI-native founders and legacy financial institutions.
Shalek is adamant that treating AI as a drop-in replacement for an existing process, capturing incremental gains but leaving the underlying structure untouched, is a fundamental mistake. "I would venture to say you can’t be extreme enough. Every organization needs to figure out how to centralize and organize all their context and data, how to give access to agents to do it, how to permission those agents, which people should oversee them, and what that means for the way they organize functions."
The data is there. The technology is falling in cost. The question is whether the organization is willing to be rebuilt around both, not just augmented by them. That distinction—between augmentation and reconstruction—is where Shalek believes the next generation of durable firms will be.
Tokenization: making assets legible to machines
Tokenization, in Shalek’s framing, is the third leg of this transformation. The practical implication for the private markets is significant. Access to investment opportunities is deeply uneven today: Tokenization is how that changes.
Brazil’s PIX and India’s UPI are the proof-of-concept at the infrastructure layer: real-time payment networks mandated across the banking system, resulting in payment volumes that dwarf those of traditional electronic networks. According to S&P Global research on tokenized private credit, the private credit market is approaching $1.7T globally—but tokenized private credit represents only around $500M of that total. The gap between the asset base and its programmable representation is enormous.
“Every asset will be tokenized,” Shalek predicted.
In conclusion
Shalek said that winning firms will be small, focused teams that sit on proprietary data and use software as the primary interface for how their business runs. That pattern maps directly onto the structure of a private markets GP. The consolidation is already happening. The firms that choose an operations partner will sit on top of the data and workflows that the next five years of AI will depend on. The firms that wait will inherit whichever stack they end up with.
Shalek noted that Juniper Square is “uniquely positioned to lead the private markets into the AI era” when Ribbit led the firm’s $130 million Series D.