The Anthropic Paradox: $20B Revenue, $380B Valuation, and a Government Trying to Kill It

Anthropic is the most paradoxical company in AI right now. In the span of two months, its revenue run rate doubled from $9 billion to $20 billion. Its valuation hit $380 billion. OpenAI’s VP of post-training research defected to join it. And yet, the US government has designated it a ‘supply chain risk,’ banned it from Treasury systems, stripped it of a $200M Pentagon contract, and summoned its CEO to testify under oath. This is a company simultaneously winning the market and losing Washington.

1. The Revenue Rocket

The growth numbers are staggering, even by AI standards. Anthropic’s annualized revenue run rate has gone parabolic in early 2026:

• End of 2025: ~$9 billion ARR
• Mid-February 2026: $14 billion ARR — implying ~$3.5B/month growth
• Early March 2026: $20 billion ARR — a 122% increase in two months

For context, this trajectory puts Anthropic on pace for $50B+ ARR by end of 2026. At $380 billion valuation, it trades at roughly 19x forward revenue — a premium, but not unreasonable if growth sustains. The ‘intelligence utility’ valuation framework suggests margins will be the defining debate over the next 18 months.

2. The Government War

While revenue surges, Washington is waging an escalating campaign against Anthropic that has no precedent for a US technology company:

• Supply Chain Risk: Defense Secretary Pete Hegseth personally designated Anthropic a ‘supply chain risk’ — a label CEO Dario Amodei called ‘retaliatory and punitive,’ stating it has never been applied to an American company before.
• Treasury Ban: Secretary Scott Bessent terminated all Claude products across Treasury agencies, including Fannie Mae and Freddie Mac.
• Pentagon Contract Lost: OpenAI won a $200M Pentagon contract previously held by Anthropic.
• Congressional Testimony: The Department of War has called for Dario Amodei to testify under oath regarding the designation.
• Legal Escalation: Anthropic filed a lawsuit against the US government to challenge the designation, escalating from protest to legal action.
• Recruitment Interference: Reports allege Anthropic attempted to block the Department of War from using public databases and LinkedIn for recruiting Anthropic employees.

3. The Stickiness Signal

Perhaps the most telling data point: the US military itself requested a 6-month phase-out period for Claude. Michael Burry observed that Palantir’s wrapper with alternative models ‘is not enough,’ arguing this reveals how sticky Claude’s technology is within classified workflows.

This creates a striking contradiction: the same government designating Anthropic a supply chain risk cannot immediately replace its technology. The military’s own request undermines the narrative that Claude is disposable — if it were, there would be no need for a transition period.

4. The Cascade Risk

The existential threat isn’t the government contracts themselves — it’s the cascade. If the supply chain risk designation is enforced broadly, the implications ripple far beyond Washington:

• Cloud Partner Risk: Amazon (AWS) and Google could be forced to remove Claude from their platforms to maintain their own government contracts. Anthropic’s primary distribution channels would evaporate overnight.
• Competitive Divergence: OpenAI has taken the opposite approach — actively engaging with military contracts, deploying to classified networks, and building ‘layered protections’ in its DoD pact. The strategic gap between the two companies is widening.
• The Altman Paradox: Even Sam Altman has publicly backed Anthropic against the designation, saying he ‘mostly agrees with Anthropic.’ OpenAI has officially stated Anthropic should not be designated a supply chain risk. A competitor defending you to the Pentagon is a signal worth parsing.

5. The Talent Signal

In the middle of all this, OpenAI’s VP of post-training research — who led the shipping of GPT-5, 5.1, 5.2, 5.3-Codex, o1, and o3 — defected to Anthropic, citing ‘supporting my friends there at this important time.’

This is a strong counter-signal to the government narrative. One of OpenAI’s most senior researchers chose to join Anthropic at the peak of its political crisis. Talent votes with its feet, and this move suggests insiders see the technology and culture as durable despite the Washington headwinds.

6. Bull vs Bear