Cybersecurity & Tech

The Invisible Hand Wasn’t Meant to Be Held by the State

Lisa Klaassen
Thursday, November 19, 2026, 1:00 PM

Washington now owns equity stakes in 40 companies. Unlike past bailouts, these aren’t rescues—they are bids for control, without an exit strategy. 

A volatile financial chart. (Pexels, https://www.pexels.com/photo/dynamic-financial-chart-showing-bullish-trend-30268013/; Public Domain)

For all the tumult of TACO (Trump Always Chickens Out), there’s a more reliable item on the menu: GUAC (Government Usually Asks for a Cut). Spoiler alert: It usually costs extra. In a little over a year, the administration has bought its way into 40 companies, giving Washington a foothold in the industries it considers strategic, from chips and critical minerals to quantum computers and nuclear energy.

It started with Intel. In August 2025, the federal government bought 433 million shares in the company that once defined American silicon, a 9.9 percent stake that made the U.S. government the largest single shareholder. Much of the $8.9 billion transaction was conjured from unpaid grants that Intel had already been promised under the CHIPS Act. The U.S., the president proudly announced on Truth Social, now “owned” 10 percent of a “Great American Company.”

As far as domestic policy goes, the intervention was anything but ordinary. Intel was a limping stag of a firm, losing the initial artificial intelligence (AI) boom to Nvidia and bleeding tens of billions of dollars as it tried to rebuild its factories and catch up to rivals like TSMC. But it was not quite knocking on death’s door. Every prior occasion that the federal government felt compelled to take a controlling stake in an American company happened under conditions of acute emergency.

During World War II, President Franklin D. Roosevelt used his executive powers to seize more than 70 factories, railroads, and coal mines in an attempt to alleviate economic pressure and prevent labor strikes from crippling the war effort. When Chrysler teetered on the edge of bankruptcy in 1979, Congress extended $1.5 billion in loan guarantees to save hundreds of thousands of jobs (and military tanks), taking warrants in return. In 1984, when Continental Illinois—then the nation’s seventh-largest bank—buckled under bad energy loans, regulators took an 80 percent equity stake, giving birth to the phrase “too big to fail.” And in 2008 and 2009, the George W. Bush and Obama administrations injected billions of taxpayer funds into AIG and General Motors in exchange for a majority equity stake, in a last-ditch effort to keep the global financial system from going down with them.

Each one of these interventions was a rescue operation performed under severe duress. Washington stepped in and unwound its stakes as fast as political decency would allow. The state assumed the role of a reluctant custodian, bailing water from a sinking ship it had no desire to captain into the future.

The AI and Intel stakes are a different beast. Rather than representing acts of crisis management, they were calculated bids for leverage over strategic industries worth billions of dollars. Nor did the stakes come alone. Intel arrived alongside a “golden share” in U.S. Steel, handing the president a veto over the firm’s major decisions, and a 15 percent royalty cut of every advanced semiconductor chip sold by Nvidia and AMD to China. 

The same logic was extended to rare earth elements, 72 percent of which the United States imports from China. In June 2025, the Pentagon became the largest shareholder in the rare-earth miner MP Materials after buying a $400 million stake in the company. Additionally, it injected $1.6 billion into Oklahoma-based USA Rare Earth—with warrants to buy an additional 17.6 million shares—a 10 percent stake in Korea Zinc, and a $50 million equity stake in Vulcan Elements.

Nuclear energy, which Washington views as essential for powering the AI build-out, received a similar treatment. The government partnered with Brookfield Asset Management and Cameco Corporation on a $80 billion deal to build a new fleet of Westinghouse Electric Company AP1000 nuclear reactors by 2030. In addition, it purchased minority equity stakes in nine quantum computing companies as part of a $2 billion federal funding initiative. Rather than handing out traditional grants, the Department of Commerce used funds from the CHIPS and Science Act to take ownership shares in companies like IBM and GlobalFoundries.

The next industry facing equity stakes will likely be AI-related. In June, President Trump mused that the government should take an ownership stake in the country’s largest AI companies, for the benefit of “the American people.” “There’s something very interesting about it, where it almost becomes a partnership with the American public,” he told reporters aboard Air Force One. In October, Trump told Time magazine that the administration would “maybe” buy a piece of OpenAI and Anthropic, modeled on its existing position in Intel. Earlier in July, OpenAI reportedly discussed giving the government a 5 percent equity stake, framing it as a way for citizens to share in the upside of AI.

Whether any of this is legal remains an open question. The CHIPS Act authorizes government grants, loans, and tax credits, but makes no mention of corporate shares. The Nvidia deal is uncomfortably close to violating the Constitution’s Export Clause, which forbids any “tax or duty” to be laid on articles exported from a state. As Vivian Dong argues, Trump’s demand that Nvidia pay a cut of its China sales in exchange for export licenses is illegal—not just because the Export Control Reform Act expressly prohibits the Bureau of Industry and Security from charging a “fee” in connection with export licenses, but because the executive cannot impose new taxes without the explicit authorization of Congress.

It is hardly surprising that none of these arrangements have been tested in court. When it comes to equity stakes, the White House and private companies are comfortable bedfellows, and neither entity has an immediate interest in unwinding the arrangement after the ink has dried. For the government, the stakes offer a direct avenue to profit-sharing in addition to political alignment. For companies, having the government act as a shareholder can secure subsidies, favorable regulation, and federal contracts—though the line between patronage and pressure is a thin one.

Why Now?

You might be wondering why a government that long preached the gospel of the free market is entering into unorthodox, not to mention expensive, corporate arrangements it has rarely felt the need or desire to enter into before. There are two plausible explanations, which are not mutually exclusive. 

The first rationale is strategic, and is the one the administration likes to broadcast through a weekly publication called the Executive Order. AI competition with China, it asserts, is too important to leave to market forces alone. This argument is not without merit. Government stakes in companies like MP Materials reduce reliance on foreign suppliers—especially Chinese ones—in critical industries like rare earths, while allowing the United States to maintain competitiveness in global markets. 

Equity stakes also act like a blue light for private investors by announcing that a firm or industry is a strategic asset to the nation, which, in turn, can create a virtuous cycle of private investment. Since Congress passed the CHIPS and Science Act in 2022, companies have committed more than $450 billion to rebuilding domestic semiconductor manufacturing. Under the Biden administration, the Department of Commerce signed multibillion-dollar agreements with chip giants like Intel, TSMC, and Samsung. Upon entering his second term, President Trump disparaged the CHIPS Act, calling subsidies a waste of taxpayer money, and instead deploying aggressive tariffs—like the Section 232 action invoked in January 2026—to force foreign companies to build factories on U.S. soil.

This tariff regime has since been split in two. Tariffs on chips, aluminium, and autos have survived through narrowed carve-outs, while broader global tariffs have been blocked by the courts. Yet rather than reject the CHIPS Act outright, the Trump administration escalated prior arrangements with domestic companies, converting billions of unpaid grants into a 10 percent stake in Intel. Within weeks, Nvidia and SoftBank committed a combined investment of $7 billion to the company. The invisible hand, it turns out, enjoys being held by the state sometimes.

The second explanation is less flattering. The country, it may be argued, is being run like a trillion-dollar real estate portfolio. The symptoms are hardly subtle, ranging from the president’s Board of Peace—a semi-governmental organization run like an exclusive member’s club—to the presidential commitment to “personally direct” $1.5 million in pledged investments from major U.S. trading partners, and a military invasion of Venezuela many observers saw as an unapologetic bid to control the country’s oil reserves. Read in this fluorescent light, the equity stakes look less like industrial policy and more like acquisitions of a Mar-a-Lago conglomerate hungry for money and power.

The trouble with running a country like a company is that this particular company also writes the rules. Washington has purchased stakes in firms it regulates before, but went to great lengths to separate the two roles. When the Federal Reserve took a controlling stake of AIG in 2008, it housed the shares in an independent trust. Similarly, when the Treasury bought the majority of General Motors, it vowed to be a “reluctant shareholder” and not interfere with management. Today’s stakes carry no such buffers. The president holds a personal veto over the management decisions of U.S. Steel, and regulatory approval has become part of the package: Nvidia and AMD were granted export licenses only once they had given the government a portion of their China sales. Any claim to neutrality has been gleefully abandoned. And with stakes in 40 companies—often rivals in the same industry—the arrangement invites cronyism, the age-old practice of politicians rewarding friendly businesses and sharing the spoils.

Favors come in many flavors: subsidized loans, tax credits, and equity stakes bought on terms no private investor would typically enjoy. But once “too big to fail” becomes entrenched as government policy, it introduces market distortions that cannot be easily reversed. Buying stakes in a small number of incumbents the government can’t afford to lose distorts competition and stifles innovation in the long term. These are hazards that one equity stake, let alone 40, will multiply.

Like any investment, there’s also the risk that there is no reward. The government might be the most powerful shareholder in America, but that doesn’t mean it always holds the winning hand. Intel offers a live case. On paper, Washington is seeing handsome returns: By September, the $8.9 billion stake, purchased at $20.47 a share, had quadrupled. But stocks are a fickle yardstick for commercial success. Less than a month later, one bad earnings call wiped more than $10 billion from the company in a single day. And while the federal stake may have improved the company’s market position, it still trails legions behind TSMC, the Taiwanese semiconductor giant that controls roughly 72 percent of the world’s chipmaking capacity. An equity stake can signal confidence and buy attention from private investors. It cannot, however, buy customers or ensure long-term success.

The Other Side of the Coin

If the government can buy its way into strategic industries, it can also threaten to cut off their lifeline. In March, Defense Secretary Pete Hegseth ordered Anthropic to give the Pentagon unfettered access to its AI models “for all lawful purposes.” In practice, this meant waiving the lab’s restrictions on using Claude to build autonomous lethal weapons and for the mass surveillance of American citizens. After Anthropic CEO Dario Amodei refused, Trump ordered all federal agencies to stop using Claude, and the Pentagon designated the company a “supply chain risk to national security.”

This was the first time Washington directed that designation, built for throttling the likes of Huawei, at a company founded and headquartered on American soil. Weeks later, a federal judge blocked the designation, calling it an “Orwellian” punishment for disagreeing with the government. The administration appealed, and the federal appeals court later upheld the Pentagon’s blacklisting of Anthropic. The AI company then filed a petition for an en banc rehearing, arguing that the case “presents questions of exceptional importance” on the interpretation of the Federal Acquisition Supply Chain Security Act—a request that is now being decided by the U.S. Court of Appeals for the D.C. Circuit.

The courts haven’t always been so obliging. In 1952, against the backdrop of the Korean War and a looming steel strike, President Truman seized the nation’s steel mills, U.S. Steel among them. The Supreme Court ruled against the president in Youngstown Sheet & Tube Co. v. Sawyer, inspiring Justice Robert Jackson to set out the limits of presidential power—a canonical text that has since been required reading for generations of law students. Seventy-odd years later, the president holds a golden share in U.S. Steel, and the courts have let the Pentagon blacklist an American company for refusing to give up its safety redlines. Truman would be forgiven for feeling a little cheated.

Anthropic explicitly cited Youngstown in its March 2026 lawsuit against the president’s directive that “EVERY Federal Agency in the United States Government [] IMMEDIATELY CEASE all use of Anthropic’s technology.” When a presidential order does not stem from powers granted through “an act of Congress or from the Constitution itself,” Anthropic stated, the order violates the separation of powers doctrine, which states that the president and executive branch lack the statutory authority to issue sweeping blacklists against a company without explicit congressional authorization.

There is precedent for America doing equity stakes well. From 1932 to 1957, the Reconstruction Finance Corporation (RFC)—a federal government agency established to provide financial support during the Great Depression—purchased stock in thousands of banks, and financed the factories that provided the firepower for World War II. When Japan cut off the country’s supply of natural rubber, it built a synthetic rubber industry from scratch. After the war, the government sold its aluminium plants, breaking Alcoa’s monopoly over the industry and incentivizing market competition.

The success of the RFC in assembling and dismantling equity stakes relied on its ability to retain institutional independence and a postwar exit strategy that prioritized market competition above the financial or political interests of the state. In its early and most prosperous decades, the agency operated as a semi-independent, nonpolitical corporate entity. This degree of independent oversight guaranteed both the RFC and the industry’s success and prevented the squandering of public funds.

The lesson Washington can learn from the RFC is not that it should never take a slice of the private pie, but that how it does so is equally important as whether it does in the first place. The RFC operated under a statutory mandate, reported to Congress, and was responsibly unraveled when its work was concluded. Today’s equity stakes carry none of these guardrails. Deals between the government and companies are struck in quick succession and announced on social media without anything resembling an exit strategy. Paired with the power to blacklist firms that refuse to fall in line, the government is flexing its federal muscle while playing a hazardous game of monopoly. 

While China has spent decades building and controlling its national champions, America’s edge has come from fostering market competition with minimal intervention. The power of the invisible hand is that no one is holding it. “The statesman,” Adam Smith cautioned, who should try to direct private capital, would “assume an authority which could safely be trusted, not only to no single person, but to no council or senate whatever.” Nowhere would this be so dangerous, he warned, “as in the hands of a man who had folly and presumption enough to fancy himself fit to exercise it.”

Trading that hand for a portfolio of equity stakes and golden shares does not merely mean playing by China’s rulebook. It means crowding out competition and stifling the innovation that made America great in the first place. In a long game, that’s the weaker hand to play.


Lisa Klaassen is a Tarbell Fellow at Lawfare, a PhD candidate at the Oxford Internet Institute, and a research affiliate at the Oxford Martin AI Governance Initiative.

Subscribe to Lawfare