AI might be a great investment --- but not for the government

Article Tools

Contact The Editor Articles By This Author

These are dark days for free-market economists when one of the few areas of bipartisan consensus is for a terrible idea: Both Vice President JD Vance and Senator Bernie Sanders want the federal government to take an explicit stake in AI firms.

Both would like to create a sovereign wealth fund made up of shares of AI-related companies. At first glance it may seem reasonable. This is the fastest growing sector of the economy and will make some people very rich. Why not give the government a piece, which it can then redistribute to the less fortunate?

The Sanders bill is more extreme. It involves taking a 50% share of AI companies and putting the government ownership in a sovereign fund. By comparison, Vance's proposal is relatively tame and vague; it involves the government buying some shares in AI companies, and perhaps putting them in a fund, as Trump has expressed interest in a the US having a sovereign wealth fund.

I'd like to believe that a proposal for the government to seize half a private company is too absurd to be taken seriously, but I know better: All sorts of crazy economic ideas are being taken seriously lately. Still, for the time being, and leaving aside recent socialist victories in various urban areas, something like the Vance plan may be likely. Maybe the government is limited to owning less than 50% of the shares. Maybe it is required to pay market rates. Maybe it takes the shares in lieu of taxes.

Again, it doesn't seem like a bad idea - if you don't think about it too deeply. After all, AI will probably exacerbate inequality in the coming decades, creating more Elon Musks and turning office parks full of white-collar workers into ghost towns. If the government has an explicit stake in these companies, then it can use that money to fund social programs and redistribute the riches.

Allow me to provide a reality check. First, the government already gets upside from AI companies - it collects taxes, both from the corporations themselves and from the income of their workers, as well as on the capital gains of shareholders. Besides which, if AI companies do what they say they will and increase productivity throughout the economy, it will make companies in other industries more productive, bringing in even more government revenue.

The US government is already long AI because it is long GDP growth. A sovereign wealth fund invested in AI doubles down on this bet - and then dials it up, because the country is running a deficit. We would essentially be making a wager with borrowed money.

This is neither a smart investment strategy nor prudent risk management. If the government bets on anything, it should be something negatively correlated with AI, like call centers or organic farms. True, Norway's sovereign wealth fund has been successful and has generated great returns for its population. Norway also runs a budget surplus and finances its fund with revenue from government-owned natural resources, not debt.

The argument for a sovereign fund also assumes the government's "investment" does not affect the outcomes for the AI companies themselves. But the government would not be a passive investor. It is a very large stakeholder that also decides regulatory policy. Its stake will make the AI industry less competitive: Not only because the large investment distorts the price of capital, which provides important signals about risk and value, but also because the government would be in the business of picking winners, creating a bias against new upstarts who will face higher costs of capital and a more difficult regulatory environment. Historically speaking, a system in which the government picks winners has not gone well.

All these conflicts risk reducing expected returns and making America's AI industry less competitive. If US capital markets were inadequate to fund AI, then maybe a public subsidy could be justified. But they are not - they remain the largest and most liquid on the planet, and they tend to be pretty good at funding innovation.

None of this is to argue that AI isn't risky. It will almost certainly introduce more volatility to markets and the economy; it may well bring big dislocations to the labor force; it could turn evil and kill us all. With all these unknowns, doubling down on one idea - starting an AI sovereign wealth fund - is not just a bad betting strategy, it's bad public policy.

(COMMENT, BELOW)

Allison Schrager, a Bloomberg columnist, is a senior fellow at the Manhattan Institute and a contributing editor of City Journal.