ED @equableinst (advancing public pension resilience); CEO @ Verdara (reclaiming the 401k); YIMBY; Hayekian-technocrat; speramus meliora, resurget cineribus
New York, NY
Joined December 2008
- Tweets2.6K
- Following549
- Followers1.6K
- Likes935
AI holdings as a share of total assets, June 30, 2026:
Colorado PERA 18.7%
Minnesota SBI 17.4%
NYC Pension Funds 14.8%
NYSTRS 14.6%
CalPERS has the most in dollars, at $65.2B (10.4% of assets).
Full table in the briefhttps://equable.org/report/brief-pension-funds-ai-investments/
Equable's estimated AI-related exposure to public pension plans could be a lot more precise if state retirement systems were more transparent about their investments. Every fund should follow the LA County Employees' Retirement Association and release an AI exposure study. Our $578 billion to $707 billion is a lower bound. It counts public equities only at the lower end of the estimate, and just 2% of private holdings and corporate debt as tied to AI companies to reach the higher end of the estimate. See more in our brief. equable.org/report/brief-pen…
As of June 30, 2026, 35.1% of all reported public equity holdings by state and local pension funds were in a basket of 52 AI-related companies. The basket covers chips, chip equipment, software, data center infrastructure, and power. Read more in our new brief. equable.org/report/brief-pen…
What AI-related companies do public pension funds hold? The largest AI positions across the public pension funds we reviewed are NVIDIA ($47.3B), Apple ($42.5B), and Microsoft ($28.2B). After SpaceX’s Q2 IPO, the funds reported $4.1B in SpaceX shares in their first post-IPO 13F filings. That figure is likely to grow. See more in our new brief. equable.org/report/brief-pen…
Thank you to DCIIA and P&I for recognizing Verdara's team as a Bright Idea winner in this year's Plan Sponsor Workplace Excellence Awards.
There's a case for state pension funds to invest in AI-companies — capture investment returns directly + drive economic growth that supports other investments.
But there is also a bright orange "systemic risk" flag waving right now.
@tonyromm reports: nytimes.com/2026/10/27/busin…
Public pension fund holdings of AI-related companies continue to grow.
In a new Equable brief, we estimate that state and local pension funds had between $578 billion and $707 billion invested in AI-related companies as of Q2 2026.
That’s 9%–11% of public pension assets tied to both the upside and the downside of concentration in AI companies.
We built this estimate by starting with 13F filings from public pension funds that report their holdings, extrapolating to funds that don’t (because they use external asset managers), and then adding estimated AI investments held through private equity and corporate debt (which collectively lack transparency).
Pension funds could provide much more transparency about their AI investments.
equable.org/report/brief-pen…
Anthony Randazzo retweeted
Repost for a chance to win a game-ready ball from JV’s 2011 no-hitter, a bobblehead & a team-issued jersey!
Added external evaluators to monitor safety compliance is sensible generally. But keep in mind how the financial industry captured rating agencies during the housing bubble—there is also energy needed to watch the watchers and affirm the quality of 3rd-party evaluators.
We Must Pace the Frontier: I’ve written a new essay on why the AI industry should slow down, with a three-part plan for doing so.
Anthropic is unilaterally committing to the first of these steps. We’ll provide third-party evaluators with permanent, employee-level access to our systems, so that they can verify adherence to our safety measures, report on incidents, and assess models’ alignment during training.
You can read the full post here: darioamodei.com/post/we-must…
New article at Equable.org — From investment portfolios to government revenues to public worker longevity, the spread of AI in the coming years will touch nearly every variable that determines the health of state and local pension plans. equable.org/resource/the-man…
Anthony Randazzo retweeted
32.2% of public pension assets are now in alternatives — ~$1.9T, quadrupled since 2008.
More reward-seeking, less liquid, harder to price → equable.org/report/state-of-…
#PublicPensions #PrivateEquity
Why public pension funds face greater valuation risk than ever, even as funding improves pionline.com/institutional-i…
Even I didn't see how dependent public pensions have become on A.I. until we took a close look the last few months... and I've been studying these plans for more than a decade.
I don't think most public pension fund leaders realize what's going on.
nitter.cf/anthonyrandazzo/status…
Colorado PERA holds 14.64% of its assets in AI companies. Minnesota SBI, 14.08%. NY State Teachers', 12.07%. Every large fund that transparently reports its public equity is heavy in AI — and the ones showing low numbers aren't the exceptions. They're the ones disclosing less.
We ranked the 25 largest public pension funds against a basket of 51 AI companies, using Q1 2026 SEC filings. The pattern in the transparent funds is remarkably consistent: 6% to 15% of assets, clustered. Maryland shows just 1.68% — but less than 10% of its assets are internally managed and disclosed, so that's a measurement floor, not a smaller bet. Two funds disclose nothing at all.
Here's the real story. What we can see is uniformly high. What we can't see — externally managed equities, and every fund's private equity positions in AI — only hides exposure that exists. None of the blind spots point downward. So the true industry number sits somewhere above what any of us can currently measure, and without better disclosure, not even the funds can state it with confidence.
Full ranking of all 25: equable.org/report/state-of-…
Here is what we got wrong and where we were correct in last year's State of Pensions 2025 report.
Every State of Pensions report includes forecasts. Most public plans end their fiscal years in June, but some don't wrap their annual measurement periods until December. So in January, we used liability forecasting and asset-class benchmark returns to predict returns and funded status.
→ We predicted the 2025 average investment return would be 9.5%. | The actual average was 10.4%.
Markets performed stronger in the second half of 2025 than the first. Plans with fiscal years ending in September and December beat June-ending plans across almost every asset class.
→ National funded ratio — Predicted: 82.5%| Actual: 81.2%
→ Unfunded liabilities — Predicted: $1.27T | Actual: $1.37T
Public payrolls grew slower than projected. States paid over 100% of required contributions as a percentage of payroll — but slower payroll growth meant fewer actual dollars toward debt paydown.
Generally, our projection methods are directionally close, but we can't always see the full picture given some transparency limits. Each year we refine our methodology to factor in evolving norms — like slower growing payroll-growth — so that anywhere we are off is instructive.
Our initial 2026 projections are now in — 85.0% funded, $1.13T shortfall — and we'll update those in October and January, and then grade those next year too: equable.org/report/state-of-…
27.1% of public pension assets are priced by valuation models, not markets — triple the pre-2008 share.
Funded ratios are only as accurate as those models.
We call this "valuation risk." It's the highest it's ever been for public plans. Deep dive on this over the coming weeks
Roughly $600 billion in public workers' retirement money is riding on AI companies — and that's likely an undercount.
From SEC 13F filings of the 25 largest public pension funds: 8–10% of assets sit in a basket of 51 AI-related companies. That's $513B–$642B.
External managers don't disclose holdings, so the transparent number understates it.
We spend all of next week on what this means.
