Investors looking to reduce their exposure to artificial intelligence (AI) may need to rethink their pension exposure. According to an Equable Institute analysis released Wednesday, state and local pension funds hold a lower-bound estimate of $578.1 billion to $706.6 billion in AI-related companies.
Meanwhile, a Wall Street Journal report found that across 13 major public funds, a basket of domestic AI-linked stocks drove $138 billion in paper gains between the end of 2022 and June.
Those gains helped pensions climb out of the hole from the 2008 financial crisis – yet heavily exposed them to one sector with 3 risks building at once: heavy concentration, political backlash against data centers, and warnings of “bubble-like” financing.
A Handful of Giants
As of June 30, 2026, AI-related companies made up 35.1% of all reported public equity investments by state and local pension funds. For funds reporting in the second quarter, the expanded AI basket reached 38.5% of public equity holdings.
The Magnificent 7 account for 63.3% of public AI-related holdings. The largest positions are NVIDIA Corp. at $47.3 billion, Apple Inc. at $42.5 billion, and Microsoft Corp. at $28.2 billion. Funds also reported $4.1 billion in newly listed Space Exploration Technologies Corp. shares.
Overall, at least 9% to 11% of the $6.42 trillion in public pension assets is tied to AI companies. Some funds are far more exposed: Colorado PERA has 18.7% of its total assets in AI, New York City’s pension funds 14.8% and the New York State Teachers’ Retirement System 14.6%.
“A positive economic outcome from the emergence of AI could meaningfully stabilize public pension plans; but a negative event like an AI-bubble bursting could cause very meaningful losses that stakeholders should be aware of,” said Anthony Randazzo, Equable’s executive director.
“Every public pension fund should be doing an internal audit of their own AI exposure and making public the findings.”
See More:
Top Momentum Stocks
The Infrastructure Contradiction
The WSJ’s analysis points to a clear contradiction. Pensions rely on AI returns, while local politicians keep fighting the infrastructure AI needs.
In July, New York put a moratorium on many new data center projects while it studies regulation, even as its pension funds’ AI holdings surged. “It’s so contradictory,” said Joseph Hernandez, who ran as the Republican candidate for New York state comptroller. “We’re on both sides of the bet.”
In Pennsylvania and California, officials and union representatives are pushing back over power prices, land use, water consumption and threats to communities.
Sen. Katie Muth [D-PA], who serves on the Public School Employees’ Retirement System board, said many former public employees don’t know their retirements rely partly on AI. She said even she struggles to work out the fund’s exposure: “It’s like being Nancy Drew.”
‘Bubble-Like’ Exuberance
According to David Colla, global head of credit investments at Canada Pension Plan Investment Board (CPPIB), there’s “bubble-like” exuberance in AI financing.
“You cannot wake up one day and just say, ‘Well, we just had this theme that AI was great, and we woke up, and it was 25% of our portfolio,” Colla said for Bloomberg.
Circular financing adds to the risk. Hyperscalers are bankrolling AI upstarts that rely on their products. If smaller players fail, the odds of a domino effect rippling through the ecosystem are significant.
Equable’s $578 billion floor also doesn’t fully capture private equity or corporate debt. Colla said CPPIB is examining off-balance-sheet structures in which shell companies take on the debt needed to build data centers. He said lenders should look even harder at privately held AI developers, whose finances are less transparent.
“There will be winners and losers along the way,” Colla said. “And there’s going to be a lot of volatility.”
Image via Shutterstock
