Hide table of contents

That's the question millions of people may soon be asking. 

Anthropic (ANT) began trading on Nasdaq today. With its valuation, it's on track to join the Nasdaq-100 and the Russell 1000 before the end of the year. Funds that track those indexes have to buy it. Meaning, ANT is about to show up in a lot of people's portfolios.

However, this IPO mostly piles onto something already in the works.

(Note: Most of this post is about the US, since that’s where the main labs get regulated. More on the rest of the world below.)

Most retirement accounts already bet on AI

The exposure is there already. (Whether or not people realize that is another thing.)

In early October, JPMorgan's collection of 41 AI-linked companies made up 49% of the S&P 500. The 10 biggest stocks in the S&P make up around 40% of the index (higher than the .com peak).

If you have a 401(k), you probably already hold a big, concentrated AI position. But most people don't know that—*I think that's the part that could change. 

(*Editor's note: You can't take my em dashes away from me.)

Does owning stock change how people vote?

Some evidence points to yes, but it's light. In Finland, when customer-owned companies went public, tens of thousands of people became shareholders without choosing to. Afterward, areas with more new shareholders voted more right-of-center (Kaustia, Knüpfer & Torstila, 2016).

When you own a stake in something though, you do pay attention to different things.

I mean, look at NIMBY-ism. William Fischel’s Homevoter Hypothesis is nothing new, and we've seen similar implications play out time and time again in local politics. Homeowners vote to protect their home values. One of the reasons housing reform is so hard, even when the majority of people say they want cheaper housing.

Ok, but what does this actually look like?

Let me pause and say I don't expect voters to start marching for accelerationism. 

The more likely problem will trickle up: Lawmakers and regulators who know a safety rule could visibly hit people's retirement accounts might look at an AI slowdown as crashing the market. And no one wants their name on that. All of a sudden, the "too big to fail" logic can be applied to frontier labs instead of banks.

Problematic.

There's pressure on the companies too. Forrester argues public-market pressure is going to affect how Anthropic behaves. Anthropic's own S-1 spends about 80/261 pages on risk factors, including the commercial risk of slowing down while competitors don't. (Race 101, public trading edition.)

Why this might not happen

I can think of a few solid reasons I could be, and hope I am, wrong:

  • Index fund owners are diversified, and mostly passive. Fischel’s whole point is that homeowners get active because they can’t diversify their biggest asset (their home). Someone with a target-date fund might never even look at what’s in it.
  • Public opinion leans cautious. In an October AP-NORC poll, 64% of Americans said AI is developing too fast, while only 8% say too slow. And in a September Quinnipiac poll, 86% of Americans surveyed supported requiring AI companies to meet independent safety standards, even if it slowed down development.
  • Anthropic's structure. With the founders keeping majority voting control, and the Long-Term Benefit Trust electing 4/7 directors, the company is somewhat shielded from shareholders pushing for speed. (Although, vice versa, they're also somewhat shielded from shareholders pushing for caution.)
  • A crash could cut the other way. If AI stocks fall hard, people who lost retirement money could turn on the industry. À la 2008 public sentiment turning on banks.

Beyond the US

The exposure is, of course, global. 

As of August, the US made up 70.6% of MSCI World, the default global index, so a European with a global tracker owns a concentrated bet on US tech whether they chose it or not. Many Australian super funds hold 20–30% in international shares.

However, most people with exposure abroad can't vote on it. The voting implications mostly come down to Americans, as the US regulates the main labs. Aka Americans hold the votes, but people everywhere hold the stock. 

Governments are an interesting case. Norway's sovereign fund owns about 1.5% of all listed stocks globally, Nvidia is its biggest holding, and withdrawals fund about a quarter of Norway's budget. They modeled an AI crash and the results were...pretty brutal, with the fund dropping ~35% of total value.

Could exposure like that make governments slower to back international AI agreements? I sure hope not.

What this means for how we talk about AI safety

Surprise! I'm not actually an economist. (As you could probably guess—comments very open for corrections and debate.) My background is comms, public relations, and psychology. 

My worry is that if the public sees AI safety as a threat to their savings, AI safety loses. As a gut hypothesis (and maybe common sense), I'd avoid any message that puts the two on opposite sides. Instead, I'd test ones that put them on the same side: An AI disaster could put your retirement at risk. AI safety and responsible regulations lower the risk of disaster.

Which does fit existing message research. Concrete, personal harms (jobs, kids) move people more than extinction framing (Social Change Lab, 2026; "From Catastrophic to Concrete," 2025). 

AFAIK, no one has tested a retirement framing yet. Might be worthwhile—asking whether people who know they hold AI stock are less supportive of a slowdown seems like an easy enough question to add to an existing survey. (If you run AI opinion polling, I’d love to hear future results.)

And EA isn't immune here—as others have pointed out, a lot of this community's donor money is tied up in AI stock, and fluctuations can have big implications. 

If people’s honest answer to “could a slowdown cost me my retirement?” becomes “yes,” AI safety gets a lot harder to sell. Seems worth getting ahead of.

If you’ve got data on any of this, or think I’m way off, tell me in the comments 🍿

3

1
0

Reactions

1
0

More posts like this

Comments
No comments on this post yet.
Be the first to respond.
Curated and popular this week
Relevant opportunities