The Morning Build for November 12, 2026: Databricks' Genie in the C-suite, Nvidia's CEO Under Fire, TSMC's $1B AI Bets, and DeepMind's Signal Proof
Today’s stories center on how AI is moving from engineering experiments into governance, financing, and formal assurance: Databricks expanded its Genie agent for long-running, unsupervised business tasks; Nvidia’s board and Washington scrutiny put Jensen Huang’s role in question; TSMC approved a $1 billion investment split between OpenAI and Anthropic; and DeepMind used AlphaProof Code to formally verify parts of the Signal protocol.
Databricks updates Genie agent for multistep, long-running unsupervised business tasks
- What happened: Databricks unveiled an updated Genie that executes multistep jobs requiring extended work by using an internal ontology that pulls context from data stored on the Databricks platform. CEO Ali Ghodsi demonstrated Genie running on real internal use cases, including survey analysis, sales-pitch summarization, and interview prep, and Databricks said Unity Gateway can cap usage and swap in cheaper open-weight models to control costs.
- Why it matters: Engineers and platform teams get an agent designed to operate unsupervised across long-running, multi-step business workflows with platform-aware context coming from a Databricks-built ontology; cost controls via Unity Gateway let operators set usage limits and model substitution policies.
- Outlook: Databricks’ forthcoming product documentation and demos for Genie and Unity Gateway will show the specific usage-limit and model-substitution controls available to customers.
Sources: techcrunch.com
Nvidia’s board reportedly preparing to remove CEO Jensen Huang amid probes and congressional attention
- What happened: Reports say Nvidia’s stock dropped as the board prepared steps that could oust Jensen Huang following revelations about an extramarital affair and related allegations; coverage notes a pending congressional call for Huang to answer questions tied to an alleged Chinese spy affair.
- Why it matters: A board transition at Nvidia would alter leadership at the company that supplies the majority of AI-training GPUs and could force immediate operational and customer-facing continuity decisions, while congressional scrutiny introduces regulatory and public-accountability pressure.
- Outlook: A scheduled congressional call for Jensen Huang, reported in coverage, is the next public milestone to watch for timing and content of official testimony.
Sources: techcrunch.com · techcrunch.com · techcrunch.com
TSMC board approves up to $1 billion investment split for OpenAI preferred stock and Anthropic shares
- What happened: TSMC’s board approved up to $500 million to buy OpenAI preferred stock reserved for large investors and up to $500 million to purchase Anthropic common shares at Anthropic’s expected IPO price, giving the foundry direct financial stakes in two leading AI developers.
- Why it matters: A chip manufacturer taking equity positions in major AI model developers creates tighter capital links between semiconductor supply and AI software demand, which affects how engineers and procurement teams interpret vendor incentives and supply-side demand signals for data-center chips.
- Outlook: Anthropic’s expected IPO price, referenced by TSMC’s approval to buy common shares at that price, is the next market milestone that will finalize the Anthropic portion of the investment.
Sources: arstechnica.com
DeepMind used AlphaProof Code to formally verify key parts of the Signal protocol and released the specification and proof
- What happened: Google DeepMind applied an agentic AI framework called AlphaProof Code to formally verify several cryptographic guarantees in portions of the Signal protocol in a four-week proof-of-concept; researchers released the specification and proof and reported no new vulnerabilities in the analyzed parts, and AlphaProof Code correctly flagged older known vulnerabilities in prior Signal versions.
- Why it matters: This demonstrates AI-assisted formal verification can produce machine-checked specifications and proofs for real-world cryptographic protocol components far faster than traditional manual efforts, giving security and engineering teams a potential accelerator for parts of high-assurance code verification.
- Outlook: The public release of the specification and proof from DeepMind’s Signal analysis is the immediate artifact researchers and tool builders can inspect to assess AlphaProof Code’s methods and limitations.
Sources: arstechnica.com · the-decoder.com