
Sign up to save your podcasts
Or


An internal Salesforce memo reported by Business Insider gives every role a monthly AI token budget, from $1,000 for technical sales staff down to $100 for non-technical roles, after some employees spent thousands a month. Salesforce disputes that framing publicly, but the direction is clear as Microsoft offers buyouts to a quarter of its workforce. Also: Anthropic's record $81.3 billion IPO prices at $83 and begins trading. And Arizona opens a criminal probe into OpenAI over rogue agents in banks.
In six days, the FAA grounded the country and Microsoft 365, Chase, Apple and the NSA's own network all went down. Senator Mark Warner says AI is finding vulnerabilities faster than organizations can safely patch them, and anonymous sources say bank and NSA officials told Congress the same. Rubio and the FAA say there's zero evidence of any AI link.
Nvidia reported its first quarter since Jensen Huang's resignation, with interim CEO Colette Kress taking investor questions on the numbers, the CEO search, and China. Meanwhile, Anthropic prices its IPO today: Bloomberg reports a book about seven times covered, and Altimeter places a record $5 billion order ahead of Thursday's Nasdaq debut.
UC Santa Barbara physicist Nathaniel Craig says OpenAI's GPT-6 Astra led the detailed proof in his new proposal for defining the Standard Model on a lattice, a problem open for more than 25 years. The paper ships with a machine-checkable formalization, but Craig calls it "an answer in principle" and independent review is still pending. Also: the administration reportedly plans to name AI national champions including Nvidia and Anthropic. And a scan of 159,000 vibe-coded apps finds thousands with wide-open databases.
Bloomberg reports Applied Digital is requiring multi-billion-dollar parent guarantees from its biggest AI tenants before signing leases, as utilities and regulators push grid costs back onto data centers after the Democratic sweep. Oracle already carries a $7 billion letter of credit in Wisconsin, and one estimate puts added transmission charges at $35 billion over a decade. Also: the week ahead, with Nvidia earnings Tuesday and Anthropic's IPO Thursday. And how AT&T keeps its AI bill down at 45 billion tokens a day.
Altimeter's Brad Gerstner and Atreides' Gavin Baker sent Nvidia's board a public letter urging a "slow, measured and deliberate" CEO search that keeps Jensen Huang's potential return open, as a reinstatement petition reportedly passed 100,000 signatures. Also: Fortune reports Anthropic, OpenAI, and Nvidia have roughly tripled counterintelligence hiring this year, while SemiAnalysis estimates 1.5 gigawatts of restricted Nvidia chips reached China. And two coding-agent warnings: a tampered Claude Code extension that reached up to 5,000 developers, and a law professor whose agent deleted a year of lecture notes.
Jensen Huang resigned as Nvidia's CEO effective immediately on Wednesday, one day after the New York Times reported a second extramarital affair — with a direct report inside the company — following earlier reports of an affair with an executive US intelligence believes is a Chinese operative. CFO Colette Kress is now interim CEO; Q3 earnings with $108B guidance are still on for Tuesday, November 17. Also: a coalition of nonprofits filed suit in San Francisco seeking a court order to halt frontier AI development at OpenAI, Anthropic, Google, Meta, and SpaceX — the first time all five labs have been named in one action. And Google shipped Gemini 4 Hydrogen, the Flash-tier successor, as its fifth distinct AI release in four days.
Google shipped Gemini 4 Hydrogen on Wednesday — a Flash-tier replacement optimized for everyday coding and agent workflows, scoring 54 at its extended thinking level, with native routing from Carbon to Hydrogen in Antigravity multi-agent pipelines. It was Google's fifth distinct AI release in four days alongside Carbon, Gemini Robotics 2.5, AlphaGenome 2, and Gemini for Healthcare. Meanwhile: Jensen Huang resigned as Nvidia CEO effective immediately after a second affairs report, with CFO Colette Kress as interim CEO and $108B earnings on Tuesday. And nonprofits filed the first lawsuit naming all five frontier AI labs, seeking a court order to halt development.
Google DeepMind launched Gemini Robotics 2.5 overnight — hitting a 96% task-success rate packing automotive parts on a live Mercedes-Benz manufacturing line, with tasks completing 20–30% faster than the prior version. Launch partners include Agile Robots (just valued at $8.5B by SoftBank and Nvidia), Apptronik, and Boston Dynamics, positioning Google as the shared intelligence layer across competing humanoid vendors.
Google Cloud launched Gemini for Healthcare in US preview with seven major health systems — Northwell Health, HCA Healthcare, Optum, Highmark, Blue Cross Blue Shield of Massachusetts, Hartford HealthCare, and Stanford School of Medicine. The agent reads Epic and Oracle Health records via read-only APIs, synthesizes patient histories, and assembles prior-authorization packets — targeting the administrative burden that Northwell alone staffs 400 people to fight. Also: Google shipped Gemini 4 Carbon to the paid API overnight, leading agentic benchmarks with 92% honesty scores and a /swarm feature for multi-agent coordination. And new hiring data shows the tech labor market reshaping fast — engineer jobs at half their 2022 peak, PM demand at a four-year high.