OpenAI vs Anthropic: The $8B Memo Drama That Shook AI

by

OpenAI, that perpetual showman of the tech world, circulating a memo this week that reads like a quarrel between two very expensive universities, has targeted its rival Anthropic with the kind of swagger you only see in corporate hallways when someone has money to burn and opinions to spare. The charge? Anthropic has allegedly inflated its $30 billion revenue figure by about $8 billion as Claude’s rather persuasive charm over enterprise AI becomes increasingly hard to ignore.

Summary

  • OpenAI chief revenue officer Denise Dresser fired off a four-page memo to staff accusing Anthropic of overstating its run rate through gross accounting on cloud deals with Google and Amazon.
  • The memo describes Anthropic’s strategy as built on “fear, restriction, and the idea that a small group of elites should control AI,” and calls its compute position a “strategic misstep.”
  • Anthropic’s annualized revenue has supposedly topped $30 billion by its own figures, up from $9 billion at the end of 2025, as Claude has become “a religion” among enterprise users at a major AI conference.

OpenAI’s chief revenue officer Denise Dresser sent a four-page internal memo to employees this past Sunday blasting rival Anthropic, alleging it inflated the widely reported $30 billion run-rate by roughly $8 billion. The memo, reported by CNBC and The Verge, claims Anthropic “grosses up” revenue sharing from its cloud partnerships with Amazon and Google rather than reporting net figures, which OpenAI does with its Microsoft arrangement.

The accusation puts the real Anthropic figure closer to $22 billion, which would place it behind OpenAI’s reported $24 billion run rate. Both companies are eyeing potential IPOs and are competing aggressively for enterprise contracts and investor positioning.

What the Memo Says

Dresser goes well beyond accounting in the note. She describes Anthropic’s strategy as built on “fear, restriction, and the idea that a small group of elites should control AI,” contrasting it with what she frames as OpenAI’s more “positive message.” She also calls Anthropic’s compute strategy a “strategic misstep,” noting that OpenAI is targeting 30 gigawatts of compute by 2030 while projecting Anthropic will have only 7 to 8 gigawatts by end-2027.

Anthropic announced a deal with Google and Broadcom earlier this month for “multiple gigawatts” of compute. OpenAI itself is also in the middle of a pivot, turning to Amazon after acknowledging that its Microsoft partnership has “limited our ability” to reach enterprise clients on rival cloud platforms.

Claude Mania and the Enterprise War

The sharpness of the memo reflects a real competitive problem for OpenAI. At the HumanX conference in San Francisco last week, enterprise sentiment was overwhelmingly in Anthropic’s favor. Arvind Jain, CEO of enterprise AI startup Glean, described the phenomenon plainly. “It has become a religion, that’s the level of that mania,” he said of Claude’s penetration into corporate workflows.

Anthropic’s momentum has come primarily from Claude Mythos and its coding tools, which have driven the revenue surge from $9 billion to $30 billion in under a year. The two labs are also racing to build competing AI cybersecurity products, with OpenAI finalizing a security tool for limited partner release while Anthropic runs its tightly controlled Project Glasswing initiative.

What It Means for the AI Race

OpenAI is valued at over $850 billion following a March fundraise. Anthropic was valued at $380 billion in its most recent round. Both companies are heading into IPO windows with very different stories to tell investors about their enterprise position.

The memo is notable precisely because confident market leaders do not typically challenge a rival’s accounting in writing. It signals that Anthropic’s gains are being felt inside OpenAI in a way that a memo to employees alone cannot solve.

Read More

2026-04-16 00:44