Two crypto investors looked at the same market this summer and drew opposite conclusions, and the gap between them is the whole story.
One is Framework Ventures general partner Rajiv Patel-O'Connor, who says his firm's move into AI, robotics, and energy is "a natural extension of our founding strategy, not a departure from it." The other is Strobe Ventures managing partner Thomas Klocanas, who says the quiet part: "When you are managing a couple of billion dollars, the current crypto deal count that meets the bar cannot absorb it."
One frames the pivot as conviction. The other admits it's plumbing. Only one of them is telling you where the money actually went and why.
One frames the pivot as conviction. The other admits it's plumbing. Only one of them is telling you where the money actually went and why.
Start with the fund that triggered the debate. Framework Ventures, founded in 2019 by two former Netflix and Snapchat employees, is a crypto-native firm — one of the largest holders of Chainlink and Synthetix outside their founding teams. Its earlier funds backed DeFi, layer-1 blockchains, and crypto infrastructure.
Last week it closed an oversubscribed $400 million fourth fund, led by Joshua Kushner's Thrive Capital.
The fund targets AI, robotics, and energy.
That is the narrative violation, stated plainly: a firm built entirely on crypto raised more money than it asked for and pointed it away from crypto. Everyone is calling this conviction in AI. It is more accurately an admission that the asset class these firms were built on can no longer hold the capital they've raised.
The tell is in the numbers Framework isn't citing. Per CryptoRank, the number of unique investors participating in crypto funding fell to 651 in the second quarter of 2026, down from a 2022 record of 2,564. Monthly participation bottomed at 222 in June — the lowest level in the period measured, and comparable to 2020, before the last cycle even began.
Deal flow contracted alongside it. Galaxy Research counted roughly $4 billion across 355 crypto deals in the first quarter of 2026, a 50% drop in invested capital from the prior quarter.
So here is the arithmetic Klocanas is describing. A fund raised in the 2021-2022 cycle sized itself for a market with 2,564 active investors and abundant late-stage rounds. That market has shrunk to a quarter of its investor base and half its deal dollars.
A couple of billion dollars cannot be deployed into a deal count that small at a defensible valuation. The capital is trapped. It has to go somewhere, and "somewhere" is wherever the deals are — which today means AI, robotics, and energy.
This is not a bet on AI's future. It is capital fleeing a market that can no longer absorb it, wearing conviction as a costume.
The face-saving version of this — the one making the rounds — is that the framework transfers even if the asset class doesn't. Crypto VCs, the story goes, learned asymmetric return-hunting and patient frontier-tech investing, and are now applying those lessons to a different frontier.
It's a tidy narrative. It also happens to be exactly the story you'd construct if your real problem were that you raised too much money for a market that died, and you needed your LPs to read a pivot as strategy rather than distress.
Notice who isn't buying it. L1D's Ray Hindi, whose firm is a fund of funds that stayed crypto-focused, called crypto VCs' chances of competing with generalist firms outside crypto "very unlikely" and warned that "most will fail." His LPs — pension funds — still want dedicated crypto managers. It's the high-net-worth individuals and family offices, he notes, who are "often more interested in hot trends and are easier to onboard with that shift in mind."
Translation: the pivot is being sold hardest to the LPs most susceptible to a fashionable story.
Now widen the lens, because crypto's collapse into AI is just one visible edge of a gravity well operating across all of venture at once.
Institutional Investor put the number on it in November. AI startups attracted nearly $193 billion — 53 percent — of the roughly $367 billion in U.S. venture deals through last September. Brad Conger of Hirtle Callaghan described "almost every dollar going into venture" flowing into AI.
His sharper observation is the one this piece is built on: even companies not explicitly labeled AI now "anchor on some AI angle." Conger suspects under-reporting, with GPs "not wanting to appear to be casting all their bets on the same theme."
That's the same gravitational effect on the founder side of the table. When capital concentrates this hard, non-AI founders don't stop raising — they cross-dress.
Consider Lucra Sports, an eSports company that is not an AI company. Its CEO, Dylan Robbins, led his pitch with AI anyway, framing his business as a hedge: if AI succeeds, people have more leisure time to game; if it fails, you've diversified into something non-AI.
Heads, AI wins. Tails, AI wins. He walked out with $20 million from ARK Invest.
That pitch is a perfect artifact of a monoculture. The winning move is no longer to have the best non-AI business — it's to explain your non-AI business through AI, in either direction, until the trend-following capital finds a reason to say yes.
The same distortion shows up wherever you look. Proptech's recovery — Xpanner's $18 million construction-robotics round, Rebar's $14 million for AI HVAC quoting, Ownwell's AI property tax appeals — reads less like proptech reviving than like real estate learning to speak the only language capital rewards.
Even a crypto seed round now needs the label: Canopy Network raised $8.5 million as an "AI-native blockchain development framework."
And here is the structural cost, the part that outlasts any single vintage. Capital this concentrated doesn't just starve viable non-AI startups — it collapses the diversity of the VC industry itself.
That collapse is now reshaping how firms are organized. In Milan, P101 acquired fellow seed investor PranaVentures in what it calls the first such VC consolidation in Italy — and its founder says more mergers are coming, because there simply isn't room for everyone.
Consolidation among managers and monomania among LPs are the same phenomenon seen from two ends. When one theme absorbs every dollar, the firms that don't specialize in it either broaden their mandates past the point of competence or get bought.
It's worth naming what this pressure-tests. Even the most sophisticated framing of the AI wave — the picks-and-shovels, robotics-as-moat thesis that treats energy and hardware as the bottleneck AI must pay to cross — assumes AI's gravity is a given. That assumption is exactly what a market this concentrated stops examining.
The people running the money aren't blind to the risk. Paul O'Brien of the $12 billion Wyoming Retirement System says he hasn't seen "the concentration of economic risk and opportunity in a single investment theme" since railroads — and railroads predate VC entirely.
His advice to allocators is not optimistic. "Get some hedges," he said. "There will be a bust at some point, and you'll need to cover your liabilities."
That is the honest terminus of a gravity well. Capital concentrates until the industry is a monoculture, and monocultures don't correct gradually. They need a bust to reset — to reintroduce the diversity that concentration destroyed, and to remind everyone that "the framework transfers, the asset class doesn't" was always a story about where the money couldn't go, not where it wanted to.
Klocanas said it cleanly. The deal count cannot absorb it. Everything else is narration.