Anthropic Drops to 26% for US Equity Stake Despite Deep Federal Ties

The Trump administration is actively defending its classification of Anthropic as a national security risk in federal court, even as Claude runs on AWS GovCloud for intelligence community workloads, even as Anthropic co-authors federal AI policy frameworks, and even as the company has pledged $50 billion in domestic data center construction. That contradiction is now fully visible in the prediction market for whether the US government will take an equity stake in Anthropic before the end of 2026.

The implied probability has fallen to 26%, down 15 percentage points in just three days, according to aggregated pricing across Kalshi and Polymarket. This is the sharpest single-candidate decline in the contract's history and lands at the period low. Kalshi prices the outcome at 12%; Polymarket sits at 40%. The spread between platforms is wide enough to question whether the two markets are even pricing the same event, but the directional consensus is unambiguous: traders are exiting.

---

Anthropic Has Done Everything Right, So Why Is the Market Selling Off?

The catalyst appears to be the May 19 appeals court hearing, where federal judges appeared divided over the Pentagon's national security designation of Anthropic. The same day, Axios reported that the Trump administration doubled down on its defense of the blacklisting, arguing the designation was lawful and within executive authority. No timeline was given for a ruling.

For a market already skeptical of the government's willingness to become Anthropic's shareholder, this was confirmation of a worse scenario: the government is not just failing to invest, it is actively restricting Anthropic's access to new defense contracts. The implication is binary. A government that classifies a company as a national security risk does not simultaneously negotiate to purchase its equity.

---

Anthropic's Washington Footprint Is Larger Than Any AI Company's

The depth of Anthropic's federal integration makes the selloff harder to explain on the merits. Claude models are deployed on GovCloud. The State Department has piloted Claude for document analysis. CEO Dario Amodei has testified before Congress and contributed to the policy frameworks that shaped the administration's AI Action Plan. Anthropic submits its models to pre-release government evaluations through the Commerce Department, a level of transparency no competitor has matched.

The financial commitments are equally outsized. Anthropic completed a $30 billion Series G in February 2026 at a $380 billion post-money valuation. It is paying SpaceX $15 billion per year for computing resources. Its annual revenue run rate has passed $30 billion, up from $9 billion at the end of 2025, with over 1,000 enterprise customers each spending more than $1 million annually.

By any operational measure, Anthropic is the AI company most deeply embedded in Washington's policy and procurement apparatus. The market is not pricing that relationship. It is pricing something else entirely.

---

The Legal Trap: How Does the US Government Actually Buy Shares in Anthropic?

Here is the structural problem the market is now absorbing: the United States has no peacetime legal framework for taking equity positions in private companies. The mechanisms that exist, including DARPA contracts, In-Q-Tel seed investments, and SBA loan programs, do not constitute equity ownership in any sense relevant to this market's resolution criteria.

The only precedents are crisis interventions. TARP authorized Treasury purchases of equity in financial institutions during the 2008 collapse. The auto bailout structured GM and Chrysler stakes through emergency legislation. AIG's government takeover required the Federal Reserve to invoke Section 13(3) authority. Every case involved a public company or a firm undergoing restructuring, and every case required either new legislation or emergency regulatory powers triggered by systemic financial risk.

Anthropic is a Public Benefit Corporation with a complex cap table that includes Amazon, Google, and sovereign wealth funds. Its $380 billion valuation makes any equity purchase astronomically expensive. CFIUS, the committee most people associate with government intervention in corporate ownership, operates in the opposite direction: it blocks foreign stakes rather than creating domestic ones. No executive order, proposed bill, or agency rulemaking currently in progress would grant the federal government authority to acquire shares in a private AI firm.

The December 31, 2026, resolution deadline compresses the window for novel legislation to roughly seven months. Congress would need to draft, debate, and pass a bill creating an entirely new investment authority, then the executive branch would need to negotiate terms with Anthropic's existing investors, all while the Trump administration is simultaneously litigating against Anthropic in federal court over its national security designation.

---

The Case for the Market Being Wrong

The strongest counterargument is that the legal framework problem can be solved with a single executive order or a rider attached to a defense appropriations bill. Washington moves slowly until it doesn't. If an AI-related national security crisis emerges before year-end, or if the administration decides that strategic control over frontier AI models requires an ownership stake rather than contractual access, the legal obstacles collapse.

Anthropic's secondary market valuation reportedly reached $1 trillion in April 2026, making it the most valuable private company in the world. A government stake at that valuation would be a statement of strategic intent on par with the creation of NASA or the establishment of the Strategic Petroleum Reserve. Some traders on Polymarket, where the price remains at 40%, appear to be pricing exactly this scenario: a low-probability, high-conviction bet that the administration will move faster than the legal consensus expects.

But the appeals court dynamics work against this theory. The Pentagon's national security designation of Anthropic suggests the current administration views the company with suspicion, not partnership. A government that is blacklisting Anthropic from new defense contracts is not positioning itself to become a shareholder. The 26% implied probability may still be generous, given the structural barriers. The market is not mispricing Anthropic's value to Washington. It is correctly identifying that value and investment are different questions, and only one of them has a legal pathway before 2027.