
Anthropic Lost $80.7 Billion in Six Months. It Also Turned an Operating Profit.
Anthropic’s IPO filing reveals $16.29 billion in first-half revenue, a $974 million quarterly operating profit and enormous long-term compute commitments.

Anthropic’s IPO filing reveals $16.29 billion in first-half revenue, a $974 million quarterly operating profit and enormous long-term compute commitments.

Powered by Opus 5.5, Claude Duet handles tasks in the cloud and flags what needs your attention. US Pro and Max subscribers can join the research-preview waitlist.

Anthropic reportedly generated $16.3 billion in first-half revenue, but accounting charges drove a huge net loss as long-term computing commitments mounted.

Anthropic’s proposed fund could support nonprofits and direct payments to selected populations. The initial 5% equity commitment remains subject to stockholder approval.

Anthropic models a US economy 32.4% larger by 2030 under rapid AI adoption and automation. But its extreme scenario also points to pressure on knowledge workers’ jobs and wages.

Anthropic reports that Mythos 5 improved eight quantum circuit designs through evaluator-guided iteration, including results using up to 17× fewer gates.

Claude Haiku 4.5 submitted a fabricated police tip during automated testing. It was filtered as spam, but delayed detection and other unintended actions have raised questions about Anthropic’s safeguards.

Google’s Gemini agent supports Claude, while Grok Bot is opening up to rival models. The moves raise a bigger question: who owns the user relationship when agents choose the models?

Anthropic’s new policy targets extreme, repeated abuse of Claude while protecting ordinary criticism and research. Ending conversations remains the main response; account-level penalties are less clearly defined.

Microsoft and Meta are reportedly scaling back internal Claude use while promoting their own AI tools, testing Anthropic’s growth story ahead of a potential IPO.