UPDATED · News · 24 May 2026 · Hannah Foster Since this story, we have covered Anthropic ASL-3 protections for Claude in our latest reporting.
Anthropic is on track to post its first ever operating profit in Q2 2026 on projected revenue of £9 (about $10.9) billion, according to figures the company shared with investors and confirmed by multiple outlets last week. CNBC reported the numbers on 20 May 2026 as Anthropic continues a funding round that could push its valuation past OpenAI.
- Anthropic projects £9 (about $10.9) billion in Q2 2026 revenue, more than double Q1’s £4 (about $4.8) billion.
- Expected operating profit of roughly £440 (about $559) million in Q2 2026.
- The Anthropic profit milestone is the first in the company’s six-year history.
- Anthropic does not expect to stay profitable in following quarters as compute spending ramps.
Why the Anthropic profit milestone matters in 2026
The Anthropic profit story is the cleanest evidence yet that one of the two leading frontier AI labs has reached a sustainable business model on the enterprise side. Anthropic’s Q1 2026 revenue of £4 (about $4.8) billion was already a record for the company; the projected £9 (about $10.9) billion in Q2 represents 127 percent quarter-on-quarter growth. That is unheard of at this revenue scale for an enterprise software company, and it pulls Anthropic well clear of the assumption that it would always trail OpenAI in commercial traction.
The headline competing number is OpenAI’s £5 (about $6) billion Q1 revenue, reported the same week. On a Q1 comparison, Anthropic is still smaller. On the Q2 projection – £9 (about $10.9) billion for Anthropic versus the £9 (about $11) to £10 (about $13) billion range analysts now expect from OpenAI – the two companies are within a quarter of each other. That is the convergence that has been hidden in 2024 and 2025 narratives, and it changes how every enterprise CIO should think about lock-in.
Where the Anthropic profit is coming from
Three lines do most of the work. Enterprise Claude API subscriptions are the biggest. Claude Code, the developer subscription that lets coding agents work continuously on a codebase, has been adopted by an estimated 80 percent of Fortune 100 engineering organisations according to recent Anthropic disclosures. And Claude Security, which entered public beta on 15 May 2026 and we covered in our Claude Security Opus 4.7 public beta piece, has already begun generating per-seat revenue from financial services and healthcare.
What is conspicuously not driving the Anthropic profit number: consumer subscriptions. Anthropic does not break out Claude.ai consumer revenue, but Goldman Sachs estimated in March that consumer Claude tops out at roughly 12 percent of total revenue, versus 35 to 40 percent for ChatGPT Plus on OpenAI’s side. The Anthropic strategy has been deliberately enterprise-first, and it has worked.
The compute bill that ends the Anthropic profit streak
Anthropic has been explicit that it does not expect Q3 or Q4 2026 to stay profitable. The reason is compute spending. In May 2026 alone the company committed to a £1 (about $1.8) billion cloud computing deal with Akamai Technologies and a fresh contract for access to xAI’s Colossus 1 data centre. Both deals are aimed at scaling Claude Opus 5 inference and training for the rumoured Q4 successor. Compute, not headcount, is now the main cost line at every frontier lab.
| Lab | 2026 quarterly revenue trajectory | MTW read |
|---|---|---|
| Anthropic | Q1 £4 (about $4.8)B → Q2 £9 (about $10.9)B projected | Fastest enterprise-side growth in software, full stop. |
| OpenAI | Q1 £5 (about $6)B → Q2 ~£9 (about $11)-13B (analyst range) | Bigger absolute, slower percentage. Consumer-heavy mix. |
| Google DeepMind / Gemini | Not broken out, bundled in Google Cloud | Probably top of the heap. Disclosure problem, not a revenue problem. |
What the Anthropic profit news means for enterprise buyers
If you are signing a multi-year frontier AI contract in 2026, the Anthropic profit projection should change your calculus in two ways. First, Anthropic now has the cash flow to honour long contractual terms without funding-round risk; the company is no longer a credit story. Second, the rough revenue parity with OpenAI means you can negotiate harder on price – the supplier-of-last-resort argument that OpenAI used through 2024 no longer holds.
Customers should still hedge. Our take in the Gemini vs GPT-5.5 piece argued the same thing from a model-quality angle: multi-vendor is the default 2026 posture for serious enterprises. The Anthropic profit news does not change that; it just means you have a healthier, more stable supplier on the Claude side of the contract.
IPO implications of the Anthropic profit number
The Anthropic profit disclosure landed during a funding round that valued the company in the high £235 (about $300) billion range, just above OpenAI’s last private mark. That valuation makes a 2027 IPO look credible: profitable companies with £32 (about $40) billion plus annualised revenue and 100 percent growth do not get blocked from public markets. Anthropic’s CFO Krishna Rao has not committed to an IPO timeline, but the disclosures look like a public-company prep exercise.
For competitors, the message is hard: the AI lab race has converged on enterprise revenue as the differentiator, and Anthropic just demonstrated it can win that race on capital efficiency. The bigger story across 2026 will be how Google, OpenAI and the second-tier labs (Mistral, xAI, DeepSeek) respond to a frontier lab that is no longer constrained by funding-round timing.
MTW verdict
The Anthropic profit projection is the most important enterprise AI story of May 2026. It changes Anthropic from a credit risk to a credit anchor and forces every CIO to take the Claude product line seriously as a long-term supplier. Multi-vendor is still the answer; Anthropic just made itself the obvious second vendor for almost everyone.
Buyer action
Where to buy or check next
Use this as the final check before ordering a phone, changing network or trusting a headline monthly price.
Reader discussion
Leave a comment
Comments are moderated. Keep it useful, accurate, and on topic.