AI today, 9 November 2026: Anthropic's reported $3.4B generator deal, Nvidia board probes Huang, Anthropic data notice

1. Business · Big

   Anthropic reportedly paid about $3.4B for 1.7 GW of gas generators

   The Financial Times reported on 8 November that Anthropic committed about $3.4 billion in late September to take over orders for 1.7 gigawatts of gas-fired generators from Texas supplier OnePWR Solutions, per CNBC's account of the report. It reportedly took over the orders, not the company. Neither Anthropic nor OnePWR has confirmed it, and Reuters could not immediately verify it.

   Why it matters: Electricity and equipment lead times are becoming a binding limit on AI data-centre growth, and labs are now buying their way up supply queues.

   CNBC

2. Business

   Nvidia's board opens independent inquiry after reports on CEO Jensen Huang

   Nvidia's board said on 8 November that a committee of independent directors, with outside counsel, will investigate reports that Huang was the target of a foreign intelligence effort. It says it has no reason to believe national security or company secrets were compromised, and that Huang will cooperate. CNBC reports that Huang told staff in a memo of a relationship with an executive at Megaspeed whom US intelligence reportedly assessed was recruited by China's Ministry of State Security; the New York Times first reported the underlying memo. Three House Republican committee chairs have asked Huang to attend a closed-door briefing on 18 November.

   Why it matters: Nvidia is the central supplier of AI chips, so questions about its leadership and China export controls reach well beyond one company. Nvidia says Huang's contacts were ordinary customer interactions.

   NVIDIA · CNBC

3. Safety

   Anthropic tells enterprise customers an outside security firm accessed account data

   Anthropic has been emailing enterprise Claude customers (a notice was posted publicly on 7 November) that an outside security research firm accessed internal systems. Exposed data was contact details, account and organisation IDs, plan tier and usage and spend figures; conversations, passwords and payment details were not involved, per the notice. Anthropic says the firm certified in writing that it deleted the data. The notice does not name the firm, and press speculation about its identity is unconfirmed.

   Why it matters: Enterprise customers should check whether they received the notice. Usage and spend metadata can reveal business activity even without chat content.

   MLQ.ai

4. Research

   Alibaba-led benchmark: best coding agent fully completes only 26.4% of real change requests

   Researchers from Alibaba and Asian universities built Co-Design Bench, where agents rework existing web apps from real user change requests. The Decoder reports that the best of ten models, GPT-6 Astra, meets about 70% of requirements on average but is fully correct on 26.4% of tasks, and just under 18% of repository-level tasks. Older benchmarks show near-perfect scores, a gap of more than 35 percentage points.

   Why it matters: Saturated benchmarks overstate how much unsupervised work coding agents can do; expect human clean-up on most real tasks.

   The Decoder

5. Policy

   Presley rights holders and Eminem publisher file new copyright suits against Suno

   Engadget reports two separate suits: one by the group that owns the rights to Elvis Presley's music, another by Eight Mile Style, which owns hundreds of Eminem recordings. They allege copyright infringement, saying Suno generated songs that sound like Presley or echo Eminem's lyrics. The filings are allegations; the report does not name the court.

   Why it matters: Music-generation firms keep facing new claims from estates and publishers, adding to the unresolved question of how copyright applies to AI training and output.

   Engadget

How this was made: compiled by an AI model (Claude) from the linked sources and checked item by item in a separate AI fact-check pass. Corrections: [email protected].