AI
🛡️Pentagon Signs AI Deals With Six Giants, Bars Anthropic Over Military Guardrail Dispute
The Rundown: The U.S. Department of Defense awarded classified AI network access to SpaceX, OpenAI, Google, NVIDIA, Microsoft, and AWS, while explicitly excluding Anthropic as a supply-chain risk.
The details:
- ●Anthropic was banned from the Pentagon deal after a dispute over military guardrails — the company reportedly refused to allow its models to be used without certain safety restrictions that the DoD considered operationally limiting.
- ●The six awarded companies will gain access to classified government networks, representing a significant commercial and reputational milestone in the race to win federal AI contracts.
- ●Meta separately acquired robotics startup Assured Robot Intelligence to build the AI intelligence layer for humanoid robots, putting it in direct competition with Google DeepMind's Gemini Robotics strategy.
- ●xAI launched Grok 4.3 with a 1M-token context window, voice cloning from 120-second audio clips, and pricing of $1.25/$2.50 per million tokens — 40–60% cheaper than Grok 4.2.
Why it matters: For founders and investors, the Pentagon decision is a masterclass in how enterprise AI deals are being won and lost on policy positioning, not just capability. Anthropic's exclusion — despite being considered a frontier safety leader — shows that inflexibility on use-case guardrails can cost you the biggest procurement contracts in the world. The Grok 4.3 pricing drop also signals an accelerating commoditization of large-context models, compressing margins across the board and forcing every AI infrastructure play to compete harder on cost.
📰 Source: Techpresso