AI Newsletter

Microsoft will no longer pay OpenAI revenue sharing and is ending the exclusivity of the partnership. This is a fundamental break in the most important strategic alliance in the AI industry and signals that Microsoft is betting on its own models and a multi-vendor strategy. For OpenAI, which is already missing revenue targets and preparing for an IPO, the loss of a major financier is existentially threatening.

The U.S. Department of Defense has signed contracts with seven leading AI companies to use their technology in classified Pentagon networks, including Microsoft, Amazon, and Google. This marks the deepest institutional embedding of commercial AI in national security infrastructure to date. The move heightens geopolitical tensions and is likely to increase pressure on European and Asian governments to build their own sovereign AI capabilities.

DeepSeek V4 has cut the price per million tokens to $2.50, putting massive price pressure on Western providers. The community response is clear: Closed Labs like OpenAI and Anthropic are caught in a cost squeeze between falling prices and exploding infrastructure expenses. This accelerates consolidation in the LLM market and fundamentally strengthens China's position in the global AI competition.

Meta has acquired Assured Robot Intelligence, a startup that develops AI models specifically for robots, entering into direct competition with Google DeepMind and OpenAI in the embodied AI segment. In parallel, KKR secured over $10 billion for building AI infrastructure, showing that Private Equity has identified the hardware layer as the next major investment field. The race for physical AI systems is entering a new, capital-intensive phase.

Anthropic has simultaneously launched 9 new integration connectors and accidentally revealed its complete strategy for the creative industry – Claude is meant to be embedded deeply as an intelligence layer in existing tools like Photoshop, rather than building native applications. This uncovers the critical strategic fork in the industry: platform integration versus standalone AI products. This approach differs clearly from OpenAI's strategy of consumer-facing products and could give Anthropic a quiet but lasting distribution advantage.