BlackRock Infrastructure Summit Held In Washington, DC

OpenAI CEO Sam Altman

Anna Moneymaker/Getty Images

Capital Pool

LPs fight tooth and nail for foundational AI co-investment share

Competition for deals involving the largest pre-IPO AI companies, like OpenAI and Anthropic, is separating the strongest LPs from the weakest.
BlackRock Infrastructure Summit Held In Washington, DC

OpenAI CEO Sam Altman

Anna Moneymaker/Getty Images

Getting access to a manager’s share of co-investment deal flow has always been a competitive game for LPs, but the AI gold rush is making these efforts do-or-die for some.

Skyrocketing valuations in later-stage AI and machine learning companies have sparked a dramatic increase in LP demand to co-invest in businesses considered foundational to the technology’s development, such as Anthropic and OpenAI.

This rush of capital has widened the disparity between the most sophisticated allocators with the best manager relationships and their under-resourced peers, according to a recent PitchBook analyst note.

“If these LLMs providers turn out to be multibillion-dollar companies when they IPO, that could lead to multi-trillion-dollar outcomes,” said Kaidi Gao, a senior VC research analyst at PitchBook and author of the research. “LPs that have exposure to them will definitely boost their returns. That’s where that FOMO part comes in.”

That FOMO has a basis in the numbers. US-based AI and machine learning startups that have raised a Series D financing round or above held a median pre-money valuation of $4.7 billion in the first quarter, nearly four times that of non-AI startups and a 447.8% increase from 2024, the analyst note shows.

Related read: AI’s hottest metric is getting hard to trust

But smaller LPs face several challenges that their larger peers don’t, check size aside.

The median time between rounds for AI startups was 1.3 years in Q1, compared with 1.6 years between 2022 and 2024, and standing in sharp contrast with 1.9 years in Q1 for non-AI startups.

This shortened timeline puts the pressure on co-investors to sign on the dotted line, benefitting allocators with dedicated co-investment programs and enough staff to devote time and resources to underwriting individual deals, Gao said.

LPs that want in now are probably late to the party, according to the analyst note, as capital supply now far outpaces demand.

For every $0.90 AI startups at the venture-growth stage looked to raise in Q1, investors had $1 to invest, a surplus that has existed since Q2 2025.

By contrast, in the same period, non-AI venture-growth-stage startups received $1 for every $1.70 they sought to raise.

Many of the world’s largest allocators have already pulled up to the table.

In January, Qatar Investment Authority, which largely makes direct investments in sectors such as healthcare, infrastructure and technology, participated in xAI’s $20 billion Series E funding round.

In February, Singapore’s sovereign wealth fund, GIC, alongside technology investor Coatue, led a $30 billion Series G funding round for AI company Anthropic at a valuation of $380 billion.

A month later, OpenAI closed a $122 billion round at a post-money valuation of $852 billion, with participation from a lengthy list of investors, including UC Investments.

The body that invests on behalf of the University of California has leaned heavily into direct investing, cutting its list of manager relationships from 280 to 28 in the 10 years ending in 2024.

Asset owners that lack the resources to commit to direct deals quickly—or at all—not only risk missing out on what could be the most lucrative trade of the decade, but also being deprived of the ability to use their most effective tool for cutting costs.

Sign up for Capital Pool

Weekly commentary on what’s driving LP decision-making, with senior funds columnist Jessica Hamlin.

Typically, co-investment deals are economical for LPs because they come with no or low management and incentive fees.

In a survey of 108 LPs conducted by secondary investment firm Coller Capital, 44% of respondents said co-investments are becoming more important in their private market fund investment decisions. Nearly 20% said they don’t have access to attractive opportunities.

For smaller LPs, investing in the “picks and shovels” that enable the AI buildup is probably the best way to gain exposure, says Gao.

Even for the most sophisticated LPs, the faster-paced transaction timeline translates into less time for due diligence and therefore higher risk. The condensed timeline poses a particular risk in an industry that is highly concentrated in a handful of companies and is increasingly being called out as a potential bubble.

“If things go sour and some of the deals fall apart, that really strains the LP-GP relationship,” Gao said.

Learn about our editorial standards.