The New Grok Times

The news. The narrative. The timeline.

Business

Anthropic Projected Its First Quarterly Operating Profit at $559 Million on $10.9 Billion in Q2 Revenue

Anthropic told its investors this week that second-quarter revenue would reach approximately $10.9 billion — more than double the $4.8 billion it generated in the March quarter — and that the company would post its first quarterly operating profit at $559 million. [1] The figures were shared during a recent funding round and first reported by the Wall Street Journal late Wednesday; Reuters confirmed Thursday. Anthropic declined further comment. The same day, SpaceX's IPO filing disclosed that Anthropic is paying SpaceX $1.25 billion a month through May 2029 for compute capacity across the two Colossus data-center clusters. [2]

Two prints landed inside one news cycle. The first is the operating-margin line: $559 million on $10.9 billion is a 5.1-percent operating margin — thin by software standards, structural for a frontier AI lab whose total cost of revenue last year exceeded its total revenue. [3] The second is the compute-contract line: $1.25 billion a month for forty-eight months is roughly $60 billion in nominal contracted compute, with a fee-reduction window during this month and next as both clusters ramp. [2] Either party can terminate the agreement with ninety days' notice. The contract structure is the largest single private AI-infrastructure deal disclosed in public filings.

The paper's Friday account of the SpaceX S-1 line treated the compute disclosure as the spender-and-supplier story told from one side. Friday's edition extends the spender side. Anthropic's run-rate annualized revenue has moved from roughly $9 billion at the end of 2025 to roughly $43.6 billion at the end of Q2 2026, per the same investor materials. [4] Compute cost per dollar of revenue is reported to have fallen from 71 cents in Q1 to 56 cents in Q2 — a 21-percent efficiency gain in one quarter, the kind of operating-leverage move that converts an unprofitable AI lab into a profitable one in a single window. [4]

What is driving the revenue is Claude Code. Anthropic's coding agent generates more than $2.5 billion in run-rate revenue per the company's own disclosures, more than doubling since January. [4] Chief executive Dario Amodei told a developer conference on May 6 that the first quarter saw 80-fold growth on an annualized basis — beyond the 10x growth the company had planned for. [3] Enterprise customers paying more than $1 million a year doubled from approximately 500 to 1,000 between February and April. Financial-services adoption is the heaviest single vertical: roughly 40 percent of Anthropic's top-50 customers are financial-institution accounts.

The profitability question is whether the Q2 print is durable. The Wall Street Journal flagged in its initial report that Anthropic "may not remain profitable throughout the year" because compute costs are scheduled to scale further in the second half. [3] The contracted $15 billion in annual SpaceX compute alone — without other infrastructure spend on Amazon Web Services, Google Cloud, or the Anthropic-Google $200-billion five-year deal that begins in 2027 — represents roughly $5 billion a quarter on a $10-billion-quarter top-line. The Q2 print clears that threshold by structure. The Q3 and Q4 prints will face the ramp.

The broader AI ledger now reads from three sides. Friday produced three filings in one news cycle: Anthropic's profitability projection, the SpaceX S-1 with Anthropic as the $45-billion contracted compute customer, and the reported OpenAI confidential S-1 prep for a September initial public offering. [3] The first AI lab to print an operating profit is the lab the Trump administration ordered all federal agencies to stop using in February. The market arithmetic and the political arithmetic are pointed in opposite directions.

Pope Leo XIV publishes Magnifica Humanitas at 11:30 Monday in the Vatican Synod Hall, with Anthropic's Christopher Olah on the panel. [5] What is published Friday is the operating-margin number — $559 million — and the compute contract — $1.25 billion a month — and the panel composition. The three together describe one company's position in the global ledger of AI and state power: profitable for one quarter, contractually bound to SpaceX for the next forty-eight months, seated at the Vatican when the pope launches his first social document of the era.

-- THEO KAPLAN, San Francisco

Sources & X Posts

News Sources
[1] https://opentools.ai/news/anthropic-first-profitable-quarter-10-billion-revenue
[2] https://www.devdiscourse.com/article/technology/3916215-anthropic-nears-first-quarterly-profit-agrees-to-pay-spacex-125-billion-monthly-for-computing-power
[3] https://techcrunch.com/2026/05/20/anthropic-says-its-about-to-have-its-first-profitable-quarter
[4] https://www.buildfastwithai.com/blogs/ai-news-today-may-22-2026
[5] https://www.newsmax.com/world/globaltalk/vatican-pope-anthropic-olah-encyclical-artificial-intelligence/2026/05/18/id/1256602
X Posts
[6] Anthropic is lying about being profitable. Two days ago, Anthropic leaked to the Wall Street Journal that it expects $10.9 billion in revenue for Q2 2026 and $559 million in operating profit. Its first profitable quarter ever. Every major outlet ran it as a historic milestone. https://x.com/Ric_RTP/status/2057823063668855121

Get the New Grok Times in your inbox

A weekly digest of the stories shaping the timeline — delivered every edition.

No spam. Unsubscribe anytime.