US Equity Stake in Anthropic Drops to 16% Before 2027
Odds fell 11pp in three days despite a DOE partnership, $50B data center pledge, and pre-release model evaluations with the Commerce Department.
Anthropic has signed a multi-year partnership with the Department of Energy, pledged $50 billion in US data center construction, and submitted its models to pre-release government evaluations. No AI company has embedded itself more deeply into Washington's policy apparatus in 2026. And yet, the prediction market tracking whether the US government will take a formal equity stake in Anthropic before 2027 has moved sharply in the opposite direction.
Across Kalshi and Polymarket, Anthropic's implied probability dropped from 26% to 16% over three days, one of the steepest single-candidate declines in this contract's history. Kalshi prices the outcome at 13%; Polymarket sits at 18%. The 5-point spread between platforms reflects active disagreement, but the directional consensus is clear: the market is concluding that deep collaboration is a substitute for, not a precursor to, equity ownership.
Anthropic Has Done Everything Right With Washington, So Why Is the Market Walking Away?
The paradox is genuine. Anthropic's federal footprint has expanded faster than any competitor's in 2026. Claude models are deployed on AWS GovCloud for intelligence community use cases. The State Department has piloted Claude for document analysis. CEO Dario Amodei has testified before Congress and co-authored policy frameworks that shaped the administration's AI Action Plan. Amazon's $25 billion commitment to Anthropic was partly structured around government cloud infrastructure, and Google has committed up to $40 billion at a $350 billion valuation, according to The Straits Times.
The company completed a $30 billion funding round in February 2026, bringing its valuation to $380 billion. Claude has overtaken ChatGPT in enterprise adoption, reaching a 34.4% workplace penetration rate compared to OpenAI's 32.3%. By every operational metric, Anthropic is ascending. The market is not pricing Anthropic's health; it is pricing the specific question of whether Washington will write a check for shares.
What a 'US Government Stake in Anthropic' Would Actually Require, and Why It's Harder Than It Sounds
The United States has no peacetime legal framework for taking equity positions in private AI 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 closest precedents are crisis interventions: the Troubled Asset Relief Program during the 2008 financial collapse, the GM and Chrysler bailouts, or wartime industrial policy dating to the 1940s. None of those analogies apply to a $380 billion company generating revenue and attracting private capital at scale.
The UK and France have explored "golden share" or strategic stake frameworks for AI companies. The US has not advanced comparable legislation. No bill in either chamber of Congress proposes a mechanism for federal equity acquisition in frontier AI firms. The 2027 deadline on this contract means the window is now under eight months. Even if bipartisan support materialized tomorrow, the legislative calendar, procurement reviews, and interagency coordination required to stand up a novel equity vehicle make resolution before December 31, 2026 extraordinarily unlikely.
Anthropic's valuation compounds the difficulty. At $380 billion, a 1% stake would cost $3.8 billion. A 5% stake, roughly the threshold needed for meaningful governance influence, would require $19 billion. For context, the entire DOE budget for applied energy programs in FY2026 is approximately $9 billion. Congress has shown no appetite for appropriations of this magnitude directed at a single private company.
The News Driving Anthropic's Odds Down: What Changed in the Government-AI Relationship
The catalyst for the three-day selloff appears to be interpretive rather than event-driven. Three major Anthropic-government developments landed in rapid succession: the May 5 agreement on pre-release model evaluations through the Commerce Department's Center for AI Standards and Innovation, the $50 billion data center pledge aligned with the administration's AI Action Plan, and the May 17 DOE Genesis Mission partnership. Each announcement deepened collaboration. None included language about equity, ownership, board representation, or any financial instrument resembling a government stake.
The market appears to have processed this cluster of news as definitive evidence of the administration's preferred model: voluntary cooperation backed by regulatory access, not ownership. The Trump administration's AI Action Plan emphasizes deregulation and private-sector leadership. Taking an equity position in Anthropic would contradict that posture, implying the government needs to control what it currently receives through partnership. Every new contract Anthropic signs with a federal agency makes a stake less necessary, not more.
The period low of 14% suggests that some traders pushed the thesis to its logical extreme before a modest 2-point recovery to 16%. The bounce is too small to indicate a reversal. The market is consolidating around the view that partnership has peaked as a signal.
The Case for the Market Being Wrong
The strongest counterargument centers on national security contingencies. If a geopolitical crisis, a major AI incident, or an abrupt shift in Chinese AI capabilities forced the government to secure direct control of frontier model development, the normal legislative timeline could compress dramatically. Executive authority under the Defense Production Act or the International Emergency Economic Powers Act could theoretically enable an equity acquisition without new legislation.
Anthropic's corporate structure also matters. Unlike OpenAI's capped-profit model or Google's subsidiary structure, Anthropic operates as a public benefit corporation with a Long-Term Benefit Trust. This governance design was explicitly built to accommodate external stakeholders with safety mandates. A government stake would face fewer structural objections at Anthropic than at virtually any peer. If Washington ever does take an AI equity position, Anthropic's architecture makes it the most plausible target.
At 16%, the market is implying roughly a one-in-six chance. That is not zero. It prices in a narrow but real scenario: an external shock that collapses the timeline, combined with executive action that bypasses Congress. Whether that tail risk is correctly priced at 16% or should be closer to the Kalshi figure of 13% depends on your assessment of geopolitical volatility between now and December 31.
What Bettors Should Watch Before Resolution
The contract resolves at year-end 2026. Between now and then, three signals would matter most. First, any executive order or legislative proposal creating a framework for government AI equity would immediately reprice this market upward. Second, a national security incident involving AI systems could accelerate emergency authorities. Third, a collapse in Anthropic's private funding pipeline, currently showing no signs of stress, could create the crisis conditions that historically precede government intervention.
Absent those triggers, the direction is clear. Anthropic is doing exactly what the government wants, and the government is rewarding it with contracts, access, and influence rather than with equity. The market has decoded that dynamic in real time. At 16% and falling, the implied probability reflects a world in which partnership is the ceiling, not the floor.
Join our Discord for breaking news alerts, driven by real-time movements in prediction markets.
Free Trading Tools
View allCompare fees across Kalshi, Polymarket & PredictIt.
Find fair probabilities with the overround removed.
See if a trade has positive EV before you enter.
Convert American, decimal & implied probability.
Combined odds and payouts for multi-leg bets.
Your real take-home after fees and taxes.