The Many Ways A.I. Will Reshape Pensions

Earlier this year, investment staff at the Los Angeles County Employees Retirement Association realized that their $94 billion pension fund had unintentionally become concentrated in A.I.-related investments. They looked at all of their public equity holdings and private equity investments and classified companies into categories, such as those that provide A.I. services or A.I. infrastructure. The investment analysts concluded in a report presented in July that the local pension fund has between 8% and 19% exposure to A.I.

This LA County pension fund is not alone. 

Nationally, at least 8% to 10% of public pension plan assets are invested in a single basket of 50 publicly traded A.I.-related companies. Little of this was an explicit choice to bet on artificial intelligence. It accumulated over the last several years as the financial market generally has come to be dominated by A.I.-related companies. The result is that as of spring 2026, at least $500 billion to $600 billion of public pension assets are concentrated in A.I. investments — and that’s before factoring in private equity A.I. investments or non-disclosed public equity holdings. 

Investment exposure is not the only channel through which artificial intelligence is influencing public pension plans. It might be the most measurable channel, but there are potentially larger effects elsewhere.  

A.I. influences the tax bases that are used to fund pension contributions, the headcount and payroll growth that unfunded liability amortization schedules are built against, and the life expectancy that determines how many years of benefits a plan owes. Some of these effects compound one another. Others offset.

This article maps where A.I. will influence public pensions in America. The intention is not to provide a forecast but to frame up the size of effects if big A.I.-related changes do materialize. In order to assess claims about A.I. in the public pension sphere, it is important to have a clear mental model of the four channels of influence — investments, government revenue and demographics, the measured value of promised benefits, and system operations. 

Investment Effects

Artificial intelligence will influence financial markets in numerous ways, potentially positive or negative depending on the time horizon measured or specific companies considered. The exact scope of investment exposure that public pension funds have to A.I. cannot be completely known without transparency from public plans — for both public equities and private equities down to the portfolio level. However, there is quite a bit we do know to help think through the ways in which A.I.-investments could influence public pension assets. 

Direct Exposure Is Already Material

Public pension funds are more exposed to A.I.-driven equity valuations than most trustees and beneficiaries realize. As noted in the introduction, first quarter 2026 SEC filings show that the 25 largest U.S. public pension funds — which together represent roughly 67.6% of all public plan assets — have 8.6% of their assets in a basket of about 50 A.I.-related public sector companies.^[1] 

That is a meaningful concentration in a narrow set of companies — and it also certainly understates the actual level of exposure because it does not capture:

• public equities managed for pension funds by external asset managers (a common practice),
• exposure to private equity or private debt investments related to A.I., or
• further concentration of public equity indexing into A.I. investments given the recent initial public offering of SpaceX (which owns xAI) and the likely forthcoming IPOs for Anthropic and OpenAI.

Possible Upside: Growth Scenario A.I. Powers the World

There is a theoretical bull case for pension fund A.I. investments:

• Pension funds that held early positions in the best-performing A.I.-related companies have already generated significant returns. 
• If the future business performance of A.I. companies exceeds current expectations, then recent investments (e.g., intentional strategies to put money into A.I. firms, or incidental exposure to A.I. simply by investing in the largest public companies) could be lucrative in the long run. 
• Plus, to the extent that A.I. investment is driving GDP growth broadly, that should generate positive returns across pension portfolios — not just in the A.I. names themselves.

Possible Downside: The Spectre of an A.I. Bubble

There is also a clear risk that concentrated exposure in A.I. investments could lead to concentrated negative returns: 

• Clearly, the bursting of an A.I. bubble with, or without, a broad recession would break financial losses across the spectrum. 
• But even a narrow, sector-specific “repricing” of A.I. companies triggered by revenue growth failing to meet valuation assumptions, a competing technology disrupting the current paradigm, or a shift in interest rates — would hit pension fund balance sheets in a way that standard stress tests may not adequately model. 

The “Losing” Companies in the A.I. Revolution

Whether or not the next decade continues an A.I. boom or bust, there will be non-A.I. companies whose values decline because they are outcompeted by A.I.-driven disruption. Consider that A.I.-driven disruption is changing the profitability of industries across the economy — logistics, financial services, legal services, healthcare administration, media. Pension funds hold large positions in many of these sectors. Even in a scenario where A.I.-related stocks continue to perform, pension funds might face meaningful underperformance from their investments in A.I.-disrupted incumbents.

Government and Demographic Effects

Artificial intelligence will have a broad influence on the economy and labor markets. The degree and scope of that influence is not fully known, but there are at least three key dimensions of government and demographic A.I. effects for public plans that should be front of mind for public pension stakeholders.

Revenue Effects, Positive or Negative

Pension systems are sensitive to macroeconomic conditions, and A.I. technological, financial, and demographic influences on the broader economy will flow through to state and local government revenues, which are the ultimate backstop for pension debt payments. If A.I. drives sustained GDP growth, state and local tax bases could expand, giving governments more fiscal capacity to make required contributions and pay down unfunded liabilities. If A.I. creates an economic reset — through job displacement that outpaces reemployment or through a financial market correction that depresses income and capital gains tax collections — the fiscal headroom for pension contributions shrinks at exactly the moment it may be most needed.

Workforce Effects

A.I. effects will extend beyond the macroeconomic and to the actual functioning of government, how workforces are developed, and strategies for compensation packages. This is meaningful for public pension plans because they are sensitive to payroll size, headcount, and salary growth experience. But consider that state governments that deploy A.I. productivity tools will, at some point, hire fewer people to do the same work. That means slower growth in the active member base, which means the payroll base against which pension contributions are calculated grows more slowly than actuarial projections assume. Lower total payroll relative to projected growth means future pension debt contributions, expressed as a percentage of payroll, will need to be recalibrated. Plans running tight amortization schedules are more vulnerable to this kind of headcount pressure than those further along toward full funding.

At the same time, governments could wind up hiring more expensive workers or take on additional functions, either of which could put upward pressure on payrolls. A.I.-driven productivity gains tend to accrue disproportionately to high-skill workers. If A.I. adoption raises wages for technical and professional staff while reducing headcount in administrative roles, the composition of the active workforce shifts. Higher average salaries with fewer members could mean the value of promised benefits grows faster than the contribution base.

These dynamics will compound with states, cities, or school districts that wind up with declining populations or student enrollment. For example, school district enrollment declines flow into revenue declines (since K-12 funding is largely per-pupil), which flows into headcount pressure, which flows into slower payroll growth. Adding A.I.-driven administrative efficiency to that picture accelerates the effects of demographic changes.

The Pension Debt Payment Problem

States with declining projected enrollment, low pension funded ratios, high required contributions, or negative cash flow are the places where macroeconomic, workforce, and second-order demographic effects will be felt most acutely.

The practical implication of workforce contraction for pension finance is underappreciated. Most amortization schedules are built on payroll growth assumptions that may no longer hold. When actual payroll growth underperforms assumptions, the contribution rate required to stay on the amortization schedule has to increase — or the amortization period extends. Plans that have already reformed their funding policies to be more disciplined are exposed to this dynamic even when they are doing everything right on the governance side.

The Value of Promised Benefits Effects

Artificial intelligence will have some influence on the health outcomes of Americans. It is not entirely clear whether these will be positive or negative and with what scope or speed any changes would spread through the country. There is a reasonable case that different populations within the U.S. will experience A.I.-health outcome effects in different ways. However, there are three key dimensions of A.I. effects on the liabilities of public plans that can be studied now to gauge how seriously they might influence future funded status levels.

Measuring Changes in Life Expectancy

A.I. applications in healthcare — accelerated drug discovery, earlier disease detection, personalized treatment — could very easily extend average life expectancy in America. The unknowns are how fast these changes might come, which people will have access to the most dynamic changes, and what the scope of improvement in longevity will be. For pension funds, longer lives mean more years of benefit payments, which means higher liabilities. Traditional actuarial approaches do not consider this a near-term concern; mortality tables are regularly updated to incorporate longevity improvements over time. But the pace of A.I.-driven healthcare improvement could outrun the lag in actuarial assumption updates, creating a period where plans are systematically underestimating their true liabilities.

Disability and Early Retirement Claims

A corollary to longevity improvement is a potential reduction in disability claims and early retirement driven by health deterioration. If A.I.-assisted medicine, diagnosis, and treatments reduce the incidence of conditions that currently drive early pension claims, liabilities trends might change — whether they go up or down depends on whether individuals who would have otherwise claimed an early pension wind up working more years to earn larger total pensions.

Liability and Mortality Modeling Itself

Actuarial science itself could be improved by A.I. technology — better mortality modeling, more granular demographic projections, improved tools for tracking experience against assumptions. All of this could lead to assumption changes that alter funded status calculations (and they could cause liabilities to increase or decrease). Plans using conservative mortality assumptions may find that A.I.-assisted modeling confirms those assumptions or that they need to make adjustments. Either way, better tools for measuring the liability improve the long-term credibility of public pension finance.

Retirement System Operation Effects

Beyond structural effects, artificial intelligence could make pension fund operations themselves more efficient — better investment management, lower administrative costs, more effective participant communication.

This final category is the smallest in scope but could have outsized effects on overall public plan health in certain states. Smaller and rural pension funds have historically struggled to compete for investment and actuarial talent.^[2] A plan covering municipal workers in a mid-size city cannot easily hire the same caliber of investment staff as CalPERS, the largest U.S. pension fund. A.I. productivity tools could create a geographic leveling effect — giving smaller plans access to analytical capabilities that were previously only available at scale. That would improve governance quality across the tail of the distribution, which is where the worst pension management outcomes tend to occur.

For participant-facing operations, A.I.-powered tools can translate complex plan structures into language that participants can act on. Tools that close the knowledge gap about how retirement plans fit into an individual’s overall financial picture can improve retirement outcomes if they successfully help people improve their decision making.

Conclusion: What It All Adds Up To

The structural effects of artificial intelligence on public pension systems will not happen all at once. The investment exposure is a present-tense issue. And operations improvements from deploying A.I. can improve public plan processes and governance now. The workforce and demographic effects are bubbling up now and will compound over the next decade. The longevity effects are a longer-horizon concern — but may require proactive near-term changes.

Pension trustees, actuaries, and policymakers who engage with this analysis now will be better positioned than those who wait for every detail in the A.I.-effects story to be clear. By the time certain A.I. influences are clear, amortization schedules will already have been set, workforce assumptions will already have been embedded, and investment exposures will already have been taken. 

The time to think through these scenarios and determine what proactive changes to be adopted is now.

Notes

1. Equable Institute analysis of Q1 2026 13F filings for 23 of the 25 largest public retirement funds. See details published with State of Pensions 2026.
2. See AI-CIO, October 2025: “For instance, some outside of financial hubs like New York may be struggling to recruit investment professionals. However, he said some funds have taken steps to address that recruitment gap, such as by setting up satellite offices in financial hubs.” And Governing, December 2023: “public plans historically have not been able to compete with private-sector salaries, especially for senior leaders and crucial in-house investment positions.”