Tech • AI • Robotics • Game

ENFR
TodayPlay for freeFury BloxLeaderboardShortsTop StoriesFor youTopicsVideosYT channelsArchivesSearchFavorites

Nvidia ousts Jensen Huang as Google launches Gemini 4 Hydrogen

AISaturday, November 14, 2026· 17 videos

Briefing

Audio player

Nvidia loses Jensen Huang

Nvidia was thrown into crisis after longtime chief executive Jensen Huang resigned and left the board amid conduct allegations. Reports centered on relationships including one involving a direct report, creating what appeared to be an unavoidable conflict-of-interest and governance breach. The fallout erased more than $600 billion in market value this week, with some estimates putting the slide from recent highs above $1 trillion. The shock matters beyond one company because Nvidia remains the central hardware supplier to the generative AI industry.

Colette Kress faces succession test

Colette Kress, Nvidia's finance chief since 2013, was installed as interim CEO while the board searches for a permanent successor. Investors are now weighing whether the company can preserve strategy, executive cohesion and customer confidence without its founder-led operating style. Internal unrest has surfaced, including employee calls for Huang's return and fears that rivals could target senior talent. The succession question is especially acute as Nvidia navigates product transitions, geopolitics and hyperscaler demand.

China scrutiny deepens Nvidia turmoil

The leadership scandal has collided with ongoing scrutiny over whether advanced Nvidia chips reached China despite U.S. export controls. Reporting has linked investigations to sales routed through a Singapore-based company, sharpening pressure from lawmakers and regulators. That overlap has turned a personal misconduct story into a broader national-security and compliance problem. It also raises fresh uncertainty over Blackwell shipments, regional policy and the company's China revenue exposure.

Anthropic evidence shifts AGI timeline

A new forecast update pushed expectations for a true AI research breakthrough sharply forward, with contributors to the AI Futures Project now centering 2027 as the most likely date for superhuman AI researchers. The shift followed evidence that Anthropic's internal systems are already materially accelerating frontier AI work. Reported progress on the lab's internal AECI score rose at roughly 29 points per year, up from about 15, or around 1.9 times faster. Forecasters argue the relevant threshold is not generic intelligence but an AI system a top lab would choose over its best all-human research team.

Google unveils Gemini 4 Hydrogen

Google launched Gemini 4 Hydrogen on November 12, positioning it as a fast, lower-cost model for coding and agentic workflows. The company says it runs about 2.5 times faster than Gemini 3.8 Flash and introduces four reasoning modes from low to X high. Benchmarks highlighted unusually strong results for a lightweight model, including gains on coding, workflow automation and agent tasks against rivals such as Claude Sonnet 5.5 and GPT 6.1 Terra. Hydrogen is rolling out across Search, the Gemini app, AI Studio and the Gemini API.

Claude Fable 5.2 wins mixed reviews

Anthropic's Claude Fable 5.2 emerged as one of the week's most talked-about model releases, praised for coding, design and long-running task execution. Users showcased ambitious outputs ranging from games to detailed interactive environments, reinforcing the model's appeal for creative and software projects. But advanced users also flagged five recurring problems, including stalled tasks, silent failures, unwanted edits, runaway parallel agents and weak auditability. A key complaint is that an important persistence fix exists but is not enabled by default, leaving performance heavily dependent on manual configuration.

Anthropic brings Claude into meetings

Anthropic began rolling out Claude Meetings in beta for Team and Enterprise customers using Google Meet. The feature lets Claude join calls, take live notes, save transcripts, generate summaries and create documents or whiteboards during discussion. It also connects meeting context to adjacent workflows through tools like Gmail and other linked apps. The launch signals a broader push to make frontier models ambient workplace software rather than standalone chatbots.

Mozilla credits AI for Firefox defense

Mozilla said an AI-assisted security pipeline helped Firefox avoid compromise at Pwn2Own 2026. Ahead of the contest, the OpenAI cyber team reported several bugs that were fixed quickly, and Mozilla said some registered attacks depended on those exact flaws. The company framed the result as the product of years of investment in fuzzing, bug bounties, sanitizers and red-team infrastructure, now amplified by automated agents. The episode offers a concrete example of AI improving defensive security operations rather than just generating offensive risk.

Videos covered

Previous briefings · AI