Venture's Money Behind the Money Calls AI Boom a 'Yellow Light'

Venture funds that once raised on four-year cycles are now coming back in 12 to 18 months, and the limited partners funding them call the acceleration a yellow light. "It's definitely a yellow light," said Amit Bhatti of TrueBridge, a Chapel Hill firm focused solely on venture. "It gives us a little bit of heartburn."

Bhatti and Beezer Clarkson of LGT Capital Partners, a Zurich-based diversified asset manager, laid out how the investors behind the VCs are navigating a market where nearly everyone has piled into the same two companies — OpenAI and Anthropic — often within the same venture firm, despite the convention against conflicting investments. Nobody's complaining. The LPs expect a bonanza.

Is the concentration in OpenAI and Anthropic a problem?

Bhatti argues it isn't. "Venture has always been a power law business, so I actually view the concentration as a good thing," he said. Funds 10-plus years old have consolidated into a few names — a lot of Databricks, SpaceX and Stripe. In the current generation, he expects a lot of Anthropic and OpenAI. "That's a feature, not a bug. Not every one of these companies is going to work out, but that consolidation is typical. What's different is how quickly it's happened."

Clarkson adds that early-stage portfolios still show diversity, with smaller checks and more varied opportunity sets that only aggregate upward into the same companies later.

How fast is too fast on fund pacing?

The norm was 24 to 36 months between funds, both said. Some managers now return in a year. Several dynamics drive it: rounds have grown enormous, and a fund putting 20% of its capital into an Anthropic-sized raise burns through pace fast. Seed round sizes have roughly doubled, so leading a deal means writing 2x the check. Many funds are also upsizing mid-cycle — unusual, and partly a way to avoid formally raising a new fund.

Clarkson draws a caution from history: "We're seeing funds come back on one-year pacing, which is fast... It could work out, we don't know, but it's typically harder to do a one-year vintage and get it right."

Bhatti tempers the worry with what managers actually bought. The last acceleration cycle bought zero-revenue crypto moonshots. This one is watching companies go from zero to $100 million in revenue — unheard of before.

Can LPs say no to a manager who keeps returning?

Not really, and both explain why. Leaving a manager and coming back is hard; the typical move is downsizing the check, as LPs did in 2021 — offering $5 million instead of $10 million because the original sum was sized for two years. "It's hard to say you're sitting it out altogether, because then you're not reliable capital," Bhatti said. He builds a calendar of expected commitments and needs managers to stay near it.

Clarkson notes some institutions rotate out of venture entirely when capital calls keep running with no distributions — a strain for endowments, hospitals and foundations with operating obligations.

The 2021 vintages, five years in, are still early but argue against one-year funds; the valuations were extreme. Bhatti wouldn't bet on the COVID vintage being his best over the last 15 years, though OpenAI-era and Cursor positions sit in some of those funds.

Where is the liquidity coming from?

Secondaries have become the third leg of liquidity alongside IPOs and M&A, both said. "If you can sell 20% of your position and return your fund, you should probably do it," Bhatti said. The taboo has faded: once a "dirty word" signaling distress, secondaries now happen at multiple levels, with LPs selling GP stakes or portfolio strips.

Real exits are flowing too. SpaceX is the headline, Cerebras was a big exit this year, plus other IPOs and M&A — including $10 billion-plus deals and a $60 billion one involving Cursor. Liquidity, Bhatti said, now goes well beyond IPOs.

On employee tenders, both flag limits. Bhatti endorses them for long-standing companies with clear financials — Stripe, Databricks — but recalls crypto's lesson: employees paid in instantly liquid tokens had no reason to stay when things got hard. Some lead investors are telling companies not to run tenders now, though founder leverage can override that.

Is there room left for emerging managers?

Yes, at smaller sizes. Roughly half the money goes to big funds, but first funds of $25 million to $50 million can still add up to $1 billion in aggregate, Clarkson said. Bhatti points to people raising $5 million or $10 million in San Francisco, formalizing angel investing. A well-known fund can start at $5 million; trying to raise $50 million or $100 million on day one creates "a giant hill to climb."

TrueBridge has backed 22- and 23-year-old GPs with close access to AI-native founders — their accelerator batches at MIT, Harvard and Stanford included Cursor and Mercor. "Success begets success in venture, so we were happy to give them a shot," Bhatti said.

Check minimums: LGT does less than $5 million with difficulty and caps itself near 20% of a fund; TrueBridge writes $3 million to $5 million and will occasionally take half or all of a fund — while telling GPs up front to diversify. One TrueBridge-backed fund where they were half the capital cut them to roughly 20% next time, because it performed.

Both expect OpenAI and Anthropic alumni to spawn a new raft of emerging managers, as Google's did. And neither sees San Francisco losing its density advantage, whatever the housing costs — the talent stays, invests and builds in place.