OpenAI's Leaked Finances Reveal $21 Billion Loss in 2025

A leaked financial statement shows OpenAI lost $21 billion in 2025 on $13 billion in revenue, ahead of its anticipated IPO. Separately, Google DeepMind's upcoming Gemini 3.5 Pro was detailed with a 2 million token context window and Deep Think reasoning, launching July 17. California also announced a deal to deploy Anthropic's Claude to 230,000 state workers, and Alibaba banned its employees from using Anthropic tools after a distillation attack accusation.

Written by AI from the sources it cites; it can be wrong — follow any citation to confirm.

4 sources1 min read

OpenAI's leaked 2025 financials show $21B loss on $13B revenue The documents, obtained by Ed Zitron and the Financial Times, reveal steepening losses as R&D costs surged ahead of an expected IPO. The figures highlight the immense capital required to compete in frontier AI. 1

Google Gemini 3.5 Pro leaks with 2M context window and Deep Think reasoning The model, launching July 17, represents a complete architectural redesign and aims to rival GPT-5.6 with advanced multi-step reasoning and SVG generation capabilities. 2

California signs deal to put Anthropic's Claude in front of 230,000 state workers The first-of-its-kind agreement gives state and local entities a 50% discount on Claude, with early deployments at the DMV and health agencies. Governor Newsom framed it as a controlled expansion of generative AI in government. 3

Alibaba bans employees from using Anthropic's Claude after distillation accusation Anthropic had accused the Chinese tech giant of carrying out the largest known distillation attack on its models, violating its terms of service. Alibaba placed Claude Code on a high-risk software list effective July 10. 4