← Back to feed

Anthropic has received investor offers valuing the company at approximately $800 billion, according to Bloomberg, a figure that would place the Claude developer in direct competition with OpenAI’s $852 billion valuation — and make it the most valuable AI-focused company outside Redmond. The company has so far declined the offers and has not confirmed plans for a new round, though a public listing has been discussed internally for as early as October.

The proposed valuation more than doubles the $380 billion pre-money figure from Anthropic’s $30 billion fundraise closed in February. What makes the number plausible is the revenue trajectory behind it: Anthropic ended 2024 at roughly $1 billion in annualised revenue. By December 2025, that figure had reached $9 billion. It was $14 billion in February, $19–20 billion in March, and crossed $30 billion in the first week of April — approximately 1,400% year-over-year growth. No private technology company has compounded revenue at that rate at this scale.

The catalyst is largely coding. Demand for Claude in software development workflows — from Cursor integrations to the newly-launched Claude Managed Agents platform — has driven enterprise adoption across the same customer base OpenAI is trying to capture with Codex and ChatGPT for Work.

The Accounting Dispute

Into that context arrived a leaked four-page internal memo from OpenAI chief revenue officer Denise Dresser, sent to sales staff and first reported by CNBC and The Verge on April 13. The memo accuses Anthropic of inflating its $30 billion run rate by roughly $8 billion by recording cloud marketplace revenue on a gross basis.

When a customer purchases Claude through AWS or Google Cloud, Anthropic books the full contracted amount as revenue and reports the cloud provider’s cut as a marketing expense. OpenAI, which routes enterprise business through Microsoft Azure, records only its net share after Microsoft’s commission. Both approaches comply with US accounting standards. The net effect, according to the memo, is that Anthropic’s disclosed ARR is $8 billion higher than it would be under OpenAI’s methodology.

Anthropic has not responded publicly to the accounting characterisation. Its infrastructure disclosures complicate OpenAI’s framing: a Broadcom SEC filing confirmed a long-term TPU deal with Google and Broadcom covering 3.5 gigawatts of compute from 2027, and a recently disclosed Google-financed $5 billion Texas data center adds further credibility to its stated growth trajectory.

The memo also outlines OpenAI’s five-priority enterprise plan: ChatGPT for Work, Codex, a new agent platform, deeper Amazon integration, and accelerating deployment of its most capable reasoning models. Dresser describes Anthropic’s approach as built on “fear and restriction” — language that has since circulated outside OpenAI.

The Investor Context

Accel, which backed Anthropic at a $183 billion Series G valuation, announced a new $5 billion fund on April 15. Its pitch to limited partners leans heavily on returns from two investments: its Anthropic stake, now worth more than four times its entry price, and Cursor, which Accel backed at $9.9 billion and is now reportedly valued near $50 billion. The fund — comprising a $4 billion Leaders Fund V and a $650 million sidecar — targets 20–25 late-stage AI investments at approximately $200 million per cheque.

The fundraise lands in a venture market that deployed $297 billion globally in Q1 2026, more than double Q4 2025 and the highest three-month total on record.

Both Anthropic and OpenAI are preparing for public listings. The revenue accounting dispute — whatever its resolution — will matter when auditors finalise the S-1s. For now, Anthropic is not raising; it’s letting the offers sit while the revenue compounds.