LPs pile into OpenAI and Anthropic as venture fund cycles shrink

The deal

LPs call one-year fund pacing a "yellow light." (substack.com) That's the mild version of a warning. It comes from two limited partners interviewed at a StrictlyVC event and written up by TechCrunch.

Limited partners say nearly everyone has piled into OpenAI and Anthropic, often inside the same venture firm, despite the usual rule against conflicting bets (daily.dev). That rule exists for a reason. Firms avoid backing direct rivals so nobody can accuse them of picking sides. Here, the rule has quietly lapsed.

Who said what

Beezer Clarkson of LGT Capital Partners and Amit Bhatti of TrueBridge discussed how limited partners are navigating a venture market transformed by AI (mezha.net). Clarkson described two to two and a half years as the earlier fundraising interval (tokenpost.com). Bhatti said 24 to 36 months was the former norm and 12 to 18 months is plausible for funds deploying capital especially quickly (tokenpost.com).

"12 to 18 is not crazy"

Amit Bhatti, partner at TrueBridge Capital Partners

That line came with a qualifier: it applies to some of the funds going really quickly (primanews.org). The estimates concern some funds. They are not an industrywide count (tokenpost.com). Nobody has published a median.

The sizes involved

These are the rounds LPs are exposed to.

No single LP wrote those checks. They rode in through venture funds, and that's where the pacing math breaks.

Why now

Start with arithmetic. In Bhatti's example, if Anthropic raises a giant quantum of capital and you put 20% of the fund into it, you're obviously going to be ahead of pace (primanews.org). One check, and a fund's deployment clock jumps.

Valuations outran fund calendars

The Series H nearly tripled Anthropic's valuation from the $380 billion it was worth in February 2026 (valueaddvc.com). That took three months. A traditional fund cycle can't absorb that. A manager who got in early sits on a large mark. A manager who didn't has to pay up or sit out.

OpenAI and Anthropic together took $217 billion in H1, which Crunchbase pegs at 43% of all startup funding (aiweekly.co). Almost half of venture dollars went to two names. If your fund wants exposure to the asset class, you can't avoid them.

Smaller checks got bigger too

Late stage isn't the only place checks swelled. Seed round sizes have also roughly doubled, pushing lead investors to write twice the check size they once did (daily.dev). Bigger checks at every stage mean funds run through capital sooner.

Managers respond predictably. A lot of funds are bumping up their fund sizes mid-cycle, which is definitely unusual (primanews.org). Raise more, deploy faster, raise again. Each step makes sense on its own.

The IPO window

Public markets come next. Anthropic filed its public S-1 on October 26, 2026, following a confidential draft filing on June 1, 2026 (tastytrade.com). OpenAI also moved. LPs can see liquidity on the horizon. That makes returning to raise again feel safe.

What the money signals

Concentration alone doesn't alarm these LPs. Both said concentration in a small number of highly valued companies is not necessarily a warning in itself (mezha.net). Bhatti described venture as a power-law business, where a few investments account for much of a fund's returns.

Prior generations did the same

He said older funds are consolidating around Databricks, SpaceX and Stripe, and said he hopes newer funds will hold significant stakes in OpenAI and Anthropic (mezha.net). Fair. The difference is speed. The speed of this concentration is unusually fast (daily.dev).

Two labs, one bet

One market newsletter read it bluntly. Fund vintages are becoming single-name bets on two labs, so LP capital for non-lab deep-tech managers may get tighter and pricier (substack.com). That's a commentator's inference, not the LPs' own claim. It's also the part I'd watch.

Three signals in the data

• Sixteen companies raised billion-dollar rounds in Q2, and seven of those were frontier labs (aiweekly.co).
• Mega-rounds now function as capital-markets events anchored by Amazon, NVIDIA, and Microsoft rather than traditional venture syndicates (aiweekly.co).
• OpenAI shares have fallen out of favor in secondary trading while Anthropic demand strengthens (aiweekly.co).

Not identical bets

The labs aren't priced the same way. Anthropic's preliminary Q2 2026 revenue was more than $11.5 billion, ahead of OpenAI's $6.7 billion (explodingtopics.com). Pricing for new money keeps climbing. Bloomberg reported on September 29 that OpenAI was aiming to raise at least $30 billion at a valuation of around $1.4 trillion (useluminix.com). Being in both names doesn't mean being in them at the same price.

The people

Two voices carry this story, and both sit on the allocator side.

Beezer Clarkson

Clarkson, a managing director at LGT Capital Partners, said some venture funds are returning to raise after about a year (tokenpost.com). That pace may strain LPs if distributions lag. She points at a cash-flow gap, not a collapse.

She also looks beyond today's marks. She said it's entirely possible there'll be a whole new raft of emerging managers backed by people coming out of OpenAI and Anthropic (primanews.org). She compared it to what happened after the Google era (mezha.net).

Amit Bhatti

Bhatti is a partner at TrueBridge Capital Partners. He brings the pacing numbers and the power-law framing. He treats 12 to 18 months as acceptable for fast movers and doesn't call it a new standard.

The outside voices

Analysts are less relaxed than the LPs. He said future growth and profit are being priced into that kind of valuation (cnbc.com).

Commentator Kevin Roose points to one figure in Anthropic's S-1. He cites the 900% net revenue retention in the prospectus and argues the valuation is effectively a bet that AI's exponential trajectory continues (substack.com).

The interviewer's own read of the room: the LPs seem unbothered by this big gamble, given the potential returns, and are expecting a bonanza (primanews.org). That's an impression, not a survey result.

The risk

Liquidity is the weak point. If distributions lag, a faster raise pace may strain LPs (tokenpost.com). An LP who commits to a new fund every year needs cash coming back from the old one. Paper gains don't pay capital calls.

Where cash comes back

Some does. The interview says liquidity has started flowing (primanews.org). OpenAI also ran a secondary. It completed a roughly $7 billion secondary share sale at its $852 billion valuation, letting current and former employees cash out (siliconvalleyinvestclub.com). That pays employees. It doesn't pay fund LPs.

Valuation risk

Anthropic's entry price is large. Anthropic closed its last funding round in May, raising $65 billion at a $965 billion valuation (cnbc.com). Forbes framed the tension around compute. The pattern works as long as enterprise demand keeps growing faster than compute can be built (forbes.com). The moment that ratio inverts, the structure prices very differently very quickly.

Early backers stepped aside

Forbes also reported that some of Anthropic's earliest backers are reportedly skipping this round entirely (forbes.com). That's reported, not confirmed by the investors. Still, the people who got in cheapest weren't buying at the top.

Gaps in the evidence

The source interview names no holdings. The investors described LP exposure as potentially lucrative, but no specific holdings or payouts were identified (tokenpost.com). So this is a mood piece with numbers around it. Nobody has shown LP-level returns yet.

What this means

If you build or sell AI infrastructure, your funding climate runs through two cap tables. Funds that put a fifth of their capital into one lab round will be back in market early. That means more checks for some and fewer for others.

For founders outside the labs

A fund that returns after 12 to 18 months needs wins to show LPs. Expect it to chase application-layer companies that ride the labs, since those have the clearest marks. Deep-tech and hardware teams without a lab narrative may find their LP-backed funds smaller. That last point rests on a market newsletter's inference, so treat it as a hypothesis.

For engineers at the labs

Clarkson's point matters most here. The investors said employees from successful startups may become limited or general partners themselves, or start companies and investment firms (mezha.net). Expect a wave of operator-led funds. They'll know the product better than most investors do.

For anyone holding private-market exposure

Read this part twice if you hold private markets through a pension or fund-of-funds. Concentration worked for Databricks, SpaceX and Stripe. It could work here. The new variable is speed. A fund that raised again after a year hasn't yet seen its first cohort of exits.

My take: the LPs are probably right that lab stakes are valuable. They're also describing a system where everyone needs the same two assets to keep working. One bad Anthropic quarter now hits a lot of funds at once.

What to watch

• Anthropic's IPO could take place as soon as mid-November 2026, at a valuation of around $2 trillion, according to Bloomberg (explodingtopics.com).
• OpenAI's next priced round, and whether it lands at or above its private mark.
• Whether any LP publishes actual distributions from funds that raised after a year.

If both labs price at or above their marks, the 12 to 18 month cycle holds. If either prices below, LPs will ask harder questions of every fund that sped up.