Today in AI — 28 February 2026

Two stories dominated the AI news cycle today, and they're connected in ways that matter for product builders. Anthropic got banned from the US government while OpenAI closed the largest private funding round in history. The power dynamics in AI are consolidating fast, and the gap between the biggest players and everyone else just got wider.

Washington picks winners

The Trump administration ordered all federal agencies to stop using Anthropic's technology after the company refused to remove safety guardrails for Pentagon use. Hours later, OpenAI announced a deal to deploy models on classified military networks. The signal to every AI company: compliance opens doors, principles close them.

• Trump orders federal agencies to stop using Anthropic products — SiliconAngle
• OpenAI reaches deal to deploy AI models on US classified network — Reuters

OpenAI's $110 billion war chest

OpenAI closed a $110 billion funding round at a $730 billion valuation. Amazon led with $50 billion, making AWS the exclusive cloud for OpenAI's enterprise platform. SoftBank and NVIDIA each contributed $30 billion. The round reshapes cloud AI infrastructure in ways that affect every builder on the platform.

• OpenAI raises $110B in one of the largest private funding rounds in history — TechCrunch

The infrastructure arms race

Meta signed a deal for 6 gigawatts of AMD GPUs worth up to $60 billion over five years, diversifying away from pure NVIDIA dependency. NVIDIA, meanwhile, detailed its Vera Rubin platform with 10x lower cost per token than Blackwell, purpose-built for agentic AI workloads.

• Meta to use 6GW of AMD GPUs, days after expanded Nvidia AI chip deal — CNBC
• Inside the NVIDIA Rubin Platform: six new chips, one AI supercomputer — NVIDIA

Building better agents

Anthropic acquired Vercept, a Seattle startup building desktop automation, to strengthen Claude's computer use capabilities. Microsoft launched Copilot Tasks, moving from chat to autonomous action across apps. And Zhipu AI released GLM-5 under MIT licence, a 745B parameter model trained entirely on Chinese hardware that competes with proprietary systems on coding benchmarks.

• Anthropic acquires Vercept to enhance Claude's computer use features — TechCrunch
• Copilot Tasks: from answers to actions — Microsoft
• MIT-licensed GLM-5 brings open source parity to closed AI giants — Open Source For You

Media fights back

The BBC, Financial Times, Guardian, Sky News, and Telegraph launched SPUR, a coalition to establish licensing standards for AI companies using journalistic content.

• UK media coalition launches SPUR initiative to demand AI training payments — BBC

The companies with the deepest pockets are consolidating power across cloud, compute, and government contracts. Builders should be watching the platform lock-in implications more closely than the headline valuation numbers.

---