PwC spots a major shift in retirement money

I would never have imagined a situation where a teacher checking a pension statement would, in some way, tie into headlines celebrating another multibillion-dollar AI data center.

Those two stories seem poles apart. Increasingly, they aren’t.

The money that’s being pumped into building AI’s physical backbone isn’t just coming from Microsoft, Amazon, or Nvidia. An eyebrow-raising sum is coming from investment pools backing public pensions, as well as sovereign wealth funds investing on behalf of governments. 

Those institutions are doing more than just scooping up shares. They’re helping to own the data centers, power systems, and companies that make the technology work.

A PwC study puts the gigantic scale in perspective, and for investors, it might look like an opportunity to participate in an infrastructure buildout. 

On top of that, it raises the possibility that different parts of your retirement portfolio might be riding the same AI bet, leaving your savings exposed to the technology’s fortunes more than one might realize.

Pension funds are taking a bigger role in building AI

PwC’s Nov. 12 report shows how the investment business is changing.

It underscores how the largest public investors are organizing deals and owning physical infrastructure, instead of simply giving money to outside fund managers. That report also puts that scale in perspective.

More Social Security:

• Social Security just changed for anyone nearing retirement
• Social Security rewards waiting till one age with a bigger safety net
• Social Security loses billions to OBBBA tax breaks

PwC projects $36.3 trillion in global public pension assets and $22.4 trillion in sovereign wealth assets by 2030. Collectively, they total a whopping $58.7 trillion.

PwC’s Eric Janson says they’re “originating transactions themselves and taking controlling positions with operational roles.”

That said, the evidence backing it up is much closer to home than a futuristic AI forecast.

In July 2026, CPP Investments committed $1.75 billion alongside investment firm EQT to support EdgeConneX’s development and operation of AI-ready data centers. That’s capital from the investment manager of Canada’s national pension plan going into facilities built to develop cloud computing and grow AI demand.

The U.S. is also a major destination for this capital.

PwC found that $66 billion in global state-investor money flowed into AI and digitization in 2025, with the United States attracting the largest share.

That also includes sovereign investors along with pension funds. Additionally, retirement funds are increasingly positioned to earn strong returns from the infrastructure that supports AI, as well as from traditional investments in tech stocks.

Americans already have substantial exposure to AI through retirement savings

A new U.S. analysis makes that connection concrete.

On Nov. 4, the Equable Institute estimated that state and local public pensions held between $578.1 billion and $706.6 billion in AI-related businesses as of June 30. 

That’s nearly 9% to 11% of their $6.42 trillion in assets.

Among the public stockholdings reported by those funds, a basket of 51 AI-related companies accounted for 35.1%. The Magnificent 7 accounted for around two-thirds of the reported AI-related stock holdings.

And it’s already influencing oversight.

On Nov. 9, CalPERS, which serves 2.4 million members, announced plans to present a responsible-AI investment framework to its board. The nation’s largest public pension said AI featured in more than 100 company engagements during the 2026 proxy season, with concerns centering on the energy, water, and community impacts of AI projects.

For someone like a retired teacher, a defined-benefit pension generally provides payments under plan rules, so day-to-day swings in AI stocks might not directly affect that monthly check.

However, investment performance impacts a plan’s funding over time, and contributions from public employers ultimately matter to taxpayers.

For a 401(k) saver, changes in stock prices are likely to reach the account balance more directly.

The same AI bet can hide in several investments

Vanguard’s 2026 research found that 84% of participants in its defined-contribution plans invest in target-date funds, which typically hold broad stock indexes.

Those funds spread investments across many securities, but market-cap weighting means the largest companies receive the largest allocations.

At the end of September, the 10 largest S&P 500 holdings accounted for 37.3% of the index, S&P Dow Jones Indices indicated. Nvidia, Apple, and Microsoft were among the leading names.

I see a genuine two-sided opportunity.

AI growth could potentially lift tech stocks while pension-backed data centers generate income from the infrastructure those companies need. More long-term institutional capital might also help finance expansion, benefiting developers, equipment suppliers, and other businesses serving AI customers.

Yet an investor holding both may be exposed to the same spending cycle in two different ways.

If AI customers expand quickly, chipmakers, data-center owners, and power providers can all benefit. If customers scale back their plans, the earnings expectations supporting several investments could change together.

This concern is strong enough that Britain’s Pensions Regulator issued guidance on Oct. 22, asking pension trustees to examine AI exposure across technology, infrastructure, energy, and other holdings, rather than treating each asset class separately.

Related: AST SpaceMobile sends strong signal to SpaceX investors