Thanks, Sam, But We Already Have a Stake in OpenAI

And more from this week…

Sam Altman thinks the federal government should own a slice of the country's leading AI companies. Oren says thanks, but no thanks:

I thought I might take it upon myself, on behalf of the nation’s taxpayers, to respond to OpenAI CEO Sam Altman’s suggestion that the federal government take 5% stakes in the leading AI companies. Thanks, Sam, but no thanks.

As with most such marketing gambits from corporate America, the thin gloss of generosity does little to obscure the self-interested posturing, in this case, to foreclose competition by attaching the public interest to the fate of particular firms. As with most such musings from Mr. Altman, the idea itself betrays a remarkable misunderstanding of the proper relationship between the private and public sectors under democratic capitalism.

The thing is, we don’t need OpenAI’s magnanimous offer of 5% of its equity. We already hold a claim on more than 20% of its profits, in perpetuity, through the corporate income tax. Would that equity become more valuable if OpenAI were to become more successful? Surely. But “successful” here would mean generating much larger profits, of which we would again receive our 20%. And if that 20% stake is insufficient, the people of the United States, through their representatives in Congress, can increase it.

Of course, corporate taxes are an imperfect way for the public to share in the private sector’s gains. But how would a direct equity stake improve matters? Equity delivers no realized value, usable for the public’s benefit, until OpenAI pays a dividend or the government sells shares. Tax payments are more reliable, and less subject to management chicanery, than dividend payments. The proceeds from a sale would be a one-time windfall, not a genuine sharing in the upside of AI.

Perhaps, as the stock appreciates, we could cash in some of the paper gains. But we can cash in those gains anyway, through the capital gains taxes collected from the other shareholders when they sell—again, an imperfect system, to be sure, but, again, a better one than trying to time the market. Enormously wealthy shareholders in companies like OpenAI have ways to shield their gains from taxation. Enormously profitable companies—like, well, definitely not OpenAI, but maybe Google or Microsoft—have ways to shield their profits. But those are not arguments for the government holding direct equity stakes. They are arguments for closing loopholes in corporate and capital gains taxes. Perhaps our Silicon Valley leaders could throw their support behind that effort?

Such a change would not benefit them much. Giving the government a stake in the performance of their stock, by contrast, comes with the tremendous benefit of the government taking an interest in the performance of their stock.

When the public shares in private-sector profits and equity appreciation through taxation, we can all be agnostic about who wins and loses. Maybe OpenAI goes on to dominate the global economy, maybe Anthropic does, maybe some startup yet to launch has the right formula. Maybe it is the frontier AI labs that capture the lion’s share of the profits from the new technology, maybe it is the power generators, maybe it is the firms that find the practical applications, maybe it is the workers who see their productivity rise. So long as we define the tax base well—corporate profits, capital gains, and labor income—we can direct a share of the tangible output from economic progress in every scenario toward public purposes.

Some have pitched the ownership stakes as a form of sovereign wealth fund. But for nations like Norway, Singapore, and the Gulf petrostates that generate enormous wealth from narrowly concentrated economic activity, the premise of such funds is to reinvest their profits as broadly as possible, usually outside their borders. The model is a second-best effort to achieve the broad exposure to economic growth that the United States already has.

Conversely, insofar as our share of AI’s value comes through our holdings of stock in OpenAI and other leading AI labs, our success depends on theirs. Any policy proposal that might not benefit them would be framed as inconsistent with the public interest. Witness how corporate interests already argue against anything that might curb their power and profitability, on the grounds that any harm to the stock market will imperil the retirement savings of millions of Americans. In on the joke, CNBC headlined its coverage of Altman’s pitch: “OpenAI proposes 5% stake to Trump administration to ease Washington pressure.”

What about Intel and other companies in strategic industries such as critical minerals, where the federal government has taken stakes? Unlike the OpenAI gambit, those investments represent an industrial policy aimed at channeling capital toward underfunded sectors and ensuring that the United States has successful companies in them. Likewise, the strongest proposals for a U.S. sovereign wealth fund envision it as a mechanism for supporting needed capital investments, not a way to broaden the public’s exposure to economic growth. There are many important debates to be had about whether and when industrial policy makes sense, and about the utility of equity stakes in particular. But such moves, intended to direct capital where the market would not otherwise send it, and allow the taxpayers taking the risk to share disproportionately in any resulting returns, bear no relationship to exceedingly well-funded tech companies buying favor with the highly valued equity they are already throwing around.

For ordinary Americans to share meaningfully in the gains from AI, the technology will have to deliver meaningful gains for ordinary Americans—not through convoluted schemes of redistribution, but through the development and diffusion of useful tools and technologies that improve their lives and jobs. That is where policymakers, and PR-hungry AI pitchmen, should focus their attention.

If a substantial share of newly created economic value is going to come from AI, and we want a disproportionate share of that value to support public goods, we should establish an AI-centric tax base and raise revenue from it, just as we do from profits, investment gains, and wages. And we should be clear that the goal is to direct some of AI’s value toward public services that benefit the nation, not to fund some new handout to compensate for a reality in which the technology is causing more harm than good.

This is the intuition behind a “token tax,” which would capture some share of the value from every computation performed. That may not be quite the right construct—for instance, the chips themselves or the energy usage might be better tax bases. Economists can help to refine the best mechanism, a much better use of their time than boldly collecting signatures for a comically vague statement that “We Must Act Now.” And technologists, instead of play-acting roles from The Wolf of Wall Street and calling up Washington with an amazing offer on a stock, but we have to move quickly, should focus on making their technology useful. — Oren

GOOD READS FOR YOUR WEEKEND

In Compact, Gregory Conti makes the humanist case against the machine in “The AI Apocalypse Is Already Here,” arguing that generative AI is not another industrial revolution but a moral and cultural cataclysm already underway, one that corrodes the work ethic, common culture, and human distinctiveness that conservatives claim to prize.

In “Young Adults are Poor Despite Every Metric Which Suggests Otherwise,” Johann Kurtz writes: “My argument is that previous generations received an enormous stock of social capital: trusted neighbors, functional public schools, a productive courtship culture, predictable career arcs, and a public square in which children could roam and adults could be relied upon. That stock, once given for free, has now been substantially liquidated. Instead, the young must now buy back, item by item and at retail prices, what their grandparents received as a bounty of prior civilizational investments.”

In “Mom, Dad, I Want to Be a Welder,” the New York Times reports that “Gen Z is increasingly turning to trade schools in hopes of future-proofing their careers against AI. But getting their parents and peers on board can be a challenge.”

And one Bad Read: For a window into the poverty of the Deirdre McCloskey school of “economics,” read the Cato Institute scholar’s conversation with Yascha Mounk on why China is “more free economically” than the United States, which she uses as part of an argument for why the United States should drop its trade barriers on Chinese autos and allow for their importation (little mention, however, of the state subsidies the Chinese government provides to its auto sector, and the beggar-thy-neighbor policies it’s used to build and protect it).

IMMIGRATION AND GROWTH

Last week, Daniel covered a new paper from Daron Acemoglu, David Autor, and coauthors, which found that across more than a hundred countries and hundreds of U.S. communities since 1950, lower birth rates predicted higher output per worker and higher wages, with no fall in aggregate output. The causal mechanism is induced innovation: when young workers grow scarce, employers must invest in technology and process improvements, and that scramble to do without cheap labor is what drives the gains. Daniel argued that the finding ought to inform our immigration policy: nothing holds so-called “labor shortages” at bay and wages down as effectively as importing foreign workers into the U.S. labor market.

This week, the Wall Street Journal covered the same paper, alongside recent work from University of Minnesota demographer Steven Ruggles, to make a contrarian case under the headline “Why AI Might Actually Help Solve the Next Labor Crisis.” Ruggles projects net entry into the labor force to turn negative in the 2030s, an unprecedented shortage he expects to lift young workers’ wages, give organized labor a rare opening, and produce the first cohort in half a century to out-earn its parents. A labor shortage, Ruggles argues, is “a huge incentive to adopt labor-saving devices, like AI.” Labor scarcity, in other words, is the forcing function that makes the machine worth building. Ruggles offers one caveat that doubles as the point: a rebound in immigration would blunt the shortage, and the wage gains with it, making every proposal to import labor a proposal to foreclose those gains before they arrive.

BONUS LINK: Read John Carney’s take on Acemoglu, Autor, and Ruggles: How Immigration Robbed Gen X and Millennials of Their Future

Meanwhile, the Wall Street Journal’s opinion pages—long a redoubt for the case that immigration answers almost any economic ailment—show a crack. In an op-ed arguing against wealth taxes as a fix for the nation’s fiscal woes, the author concedes what immigration boosters rarely will: that importing workers is no cure for our demographic problems. “Immigration simply defers the problem. Immigrants also age and retire, and their birthrates converge with the national average within a generation or two.”

ON CAPITOL HILL

Russia Sanctions Bill Hits 60 Co-Sponsors (Axios). A Russia sanctions bill championed by the late Sen. Lindsey Graham (R-SC) might soon hit the Senate floor. The legislation would impose new sanctions on Russian defense, energy, and financial targets, as well as on the shadow fleet Moscow uses to evade current restrictions. It would also authorize President Trump to impose tariffs of up to 100% on the five countries that buy the most Russian oil and gas, China and India among them, for as long as they keep buying, and on the countries and entities that help Russia evade the energy sanctions. This would mark the first time Congress has authorized tariffs as an explicit geopolitical weapon (that is, unless you believe the Supreme Court erred in its Learning Resources, Inc. v. Trump ruling…), and give the administration a new tariff authority on strong statutory footing for the duration of the Russian-Ukrainian war. It would also amount to yet another instance of the merging of the economic and national-security spheres, which Oren described at the onset of the administration’s second-term trade agenda, citing former chair of the Council of Economic Advisers Stephen Miran’s prediction that tariffs would be wielded in ways deeply intertwined with national-security concerns.

Connected Vehicle Security Act scheduled for markup. Next week, the Senate Commerce Committee will mark up the Connected Vehicle Security Act, bipartisan, bicameral legislation from Sens. Bernie Moreno (R-OH) and Elissa Slotkin (D-MI), which American Compass endorsed upon its introduction. The legislation would codify a Biden-era rule prohibiting Chinese-linked connected-vehicle software in the U.S. market and expand the prohibition to hardware. Unsurprisingly, certain industry actors are lobbying behind the scenes to soften some of the bill’s prohibitions. The Commerce Committee should resist them and report the strongest version it can, especially in light of expectations that the version reported out of committee will be folded into this year’s National Defense Authorization Act (NDAA).

Speaking of the NDAA: the current version of the must-pass legislation circulating in the Senate, as currently drafted, includes the MATCH Act, the AI Overwatch Act, and the Chip Security Act, a trio of measures that would curtail China’s access to AI semiconductors and semiconductor manufacturing equipment. As American Compass wrote in Stop Selling the Rope, these measures and others like them are vital for U.S. economic competitiveness and national security.

REINDUSTRIALIZATION AND DEINDUSTRIALIZATION

And finally, some reindustrialization (and deindustrialization) headlines to end your week.

In the Wall Street Journal, “The White House Made Fixing Intel Its Pet Project. It’s Working.” The administration took a 10% stake in Intel and has since twisted the arms of Apple, Nvidia, and SpaceX into signing on as customers. The plan is starting to pay off.

In the New York Times, “TSMC Adds $100 Billion to Its U.S. Spending Plan.” The pledge brings the Taiwanese giant’s total Arizona commitment to $265 billion, which the Times concedes is “partly in response to increasing political pressure on foreign semiconductor companies to locate factories in the United States.”

And also in the Journal, “The U.S. Is Trampling Allies in the Global Hunt for Rare Earths.” The United States has outspent the European Union on critical minerals by roughly eight-to-one over five years, buying up non-Chinese supply faster than Europe can convene a meeting about it. The complaint frames American decisiveness as the culprit (refreshing!). But the deeper lesson runs the other way: while Europe deliberates its way to dependence on Beijing, the United States is building the alternative. As one European executive quoted in the piece put it, “while we’re talking, the next value chain is bought by the Americans.”

Recent headlines confirm as much: EU Accuses China of Seeking to Reshape Global Order in Stark New Strategy Paper (SCMP), and Europe Has to Take on China to Save Free Trade. Novel insights and strategies for—checks calendar—July 2026. The EU better act soon: Volkswagen Warns It May Need to Cut 50,000 More Jobs (WSJ).

Enjoy the weekend!