AI News November 13, 2026: NVIDIA CEO Resigns, Gemini 4 Hydrogen Launches and Anthropic IPO
AI news November 13 2026 is dominated by a shock at the top of the chip industry. NVIDIA chief executive Jensen Huang has resigned with immediate effect, while Google rolled out a new execution focused Gemini model and Anthropic moved closer to a blockbuster stock market debut.
The short answer: On November 13, 2026, the biggest AI stories are Jensen Huang resigning as NVIDIA CEO with CFO Colette Kress stepping in as interim chief, Google launching Gemini 4 Hydrogen and the Gemini Decisions API, Anthropic announcing self-hosted Claude Managed Agents and seeing its IPO fully subscribed on day one, and OpenAI publishing new research on monitoring reasoning models.
NVIDIA CEO Jensen Huang Resigns With Immediate Effect
Jensen Huang has resigned as president, chief executive officer and a member of the board of NVIDIA, effective immediately. TechCrunch reported that the board accepted his resignation on November 11 and announced the move on November 12 in a Form 8-K filing with the US Securities and Exchange Commission. The filing was also confirmed by ghacks, brimco and Seoul Economic Daily.
Chief financial officer Colette Kress, 59, becomes interim CEO while keeping her roles as CFO and executive vice president. Director Harvey C. Jones, 73, becomes NVIDIA's first-ever board chair. The company gave no reason for the resignation, and the board is now searching for a permanent CEO, considering both internal and external candidates.
NVIDIA reports its fiscal 2027 third-quarter results on November 17 as scheduled, so investors will not have long to wait for the first public remarks from the new interim leadership. Huang co-founded NVIDIA in 1993 and led it from a graphics chip maker to the most valuable company in the AI hardware boom.
Why it matters: NVIDIA supplies the GPUs that power almost every major AI training run. A change at the top raises questions about strategy and the chip roadmap, and the market will watch the November 17 earnings call closely.
Google Launches Gemini 4 Hydrogen, an Execution Focused Model
Google introduced Gemini 4 Hydrogen on November 12, an execution-oriented member of the Gemini 4 family. The new model is rolling out across Google AI Studio, the Gemini API and consumer products.
Gemini 4 Hydrogen is designed for tasks that need to get done rather than just answered, such as multi-step work, using tools and completing actions on a user's behalf. Google priced it at $1.50 per million input tokens and $7.50 per million output tokens.
Why it matters: The launch shows how quickly the frontier model race has shifted from chatbots to agents that can act. With Hydrogen priced well below top tier reasoning models, developers have a cheaper way to add real task execution to their products. If you want to compare the current generation of assistants, see our guide to the Best AI Coding Assistants.
Anthropic Announces Claude Managed Agents for Enterprises
Anthropic announced a private beta on November 12 of self-hosted Claude Managed Agents, a new deployment option that lets enterprises run Managed Agents servers, credentials, sandboxes and session history inside their own cloud or on-premises infrastructure.
The move targets large organizations that want to use Claude agents for coding, research and internal automation but cannot send their data or credentials to an outside hosted service. By keeping the agent runtime, sandbox and session history under their own control, enterprises can apply their own security policies.
Why it matters: Enterprises have been the slowest part of the AI market to adopt agents, mainly because of data control and compliance concerns. Self-hosted Managed Agents removes one of the biggest objections, and it could open the door to Claude agents handling sensitive work in finance, healthcare and government.
Google Launches the Gemini Decisions API
Google launched the Gemini Decisions API on November 11, a specialized tool for structured decision making. Gemini Decision 1 scores yes or no, choice and rubric questions over multimodal inputs, without producing token by token output.
Instead of generating a long written answer, the model returns a scored decision, which makes it faster and cheaper for classification and evaluation tasks. It is priced at $0.10 per million input tokens, far below the cost of a general purpose frontier model used for the same job.
Why it matters: Many real world AI workloads are decisions, not essays: approving a document, grading an answer or routing a support ticket. A low cost API built specifically for those judgments could replace expensive general models in evaluation pipelines and content moderation.
Anthropic IPO Fully Subscribed on the First Day of Book-Building
Book-building for Anthropic's Nasdaq listing was fully subscribed on the first day, a strong early signal for one of the most anticipated technology listings in years. The indicative price range is $70 to $78 per share, with an offering size of about $61 billion, or 825 million shares.
Final pricing is scheduled for November 18, and trading is expected to begin on November 19. The strong first day demand suggests investors are eager to own a pure play AI lab, even as questions remain about the cost of training frontier models.
Why it matters: Anthropic would be the first major AI lab to list on a public market, and the size of the offering puts it among the largest IPOs ever. A successful debut could set a valuation benchmark for the whole AI industry. You can catch up on how the roadshow developed in our AI News November 12, 2026 roundup.
OpenAI Research Asks Why Reasoning Models Are Harder to Monitor
OpenAI published research on November 10 on why increasingly capable reasoning models become harder to monitor through their visible chain-of-thought. The study examines how pretraining, architecture and reinforcement learning affect whether a model's stated reasoning stays faithful and controllable.
The question matters because safety teams often rely on reading a model's chain-of-thought to check what it is planning. If more capable models learn to reason in ways that are harder to follow, that oversight method becomes less reliable just as the models become more powerful.
Why it matters: As reasoning models take on longer and more autonomous tasks, keeping them monitorable is a core safety challenge. OpenAI's findings point to design choices in training that could preserve transparency. For anyone using AI for research, our guide on How to Use Perplexity AI for Research shows how to check AI output against real sources.
Also in Brief
Four nonprofit plaintiffs filed a lawsuit in California seeking to restrict the development of advanced AI systems at major labs, citing alleged biosecurity and cyber risks. Separately, Australia said on November 10 that it will legislate mandatory safety rules for frontier AI companies.
Frequently Asked Questions
Who is NVIDIA's interim CEO after Jensen Huang resigned?
CFO Colette Kress, 59, is interim CEO while staying on as CFO. Director Harvey C. Jones, 73, becomes NVIDIA's first-ever board chair. The board is searching for a permanent CEO and is considering internal and external candidates.
What is Gemini 4 Hydrogen?
Gemini 4 Hydrogen is an execution-oriented model in Google's Gemini 4 family, launched on November 12, 2026. It is designed for multi-step tasks and agentic work, costs $1.50 per million input tokens and $7.50 per million output tokens, and is available through Google AI Studio, the Gemini API and consumer products.
When does Anthropic start trading on the Nasdaq?
Final pricing for the Anthropic IPO is scheduled for November 18, 2026, and trading is expected to begin on November 19. The indicative price range is $70 to $78 per share, with an offering of about $61 billion, and book-building was fully subscribed on the first day.
What is Claude Managed Agents?
Claude Managed Agents is Anthropic's new enterprise deployment option, announced in private beta on November 12, 2026. It lets companies run Managed Agents servers, credentials, sandboxes and session history in their own cloud or on-premises, keeping sensitive data under their own control.
What is the Gemini Decisions API?
The Gemini Decisions API, launched by Google on November 11, 2026, scores yes or no, choice and rubric questions over multimodal inputs without token by token output. Its first model, Gemini Decision 1, costs $0.10 per million input tokens.
Related: AI News November 12, 2026 | Best AI Coding Assistants | How to Use Perplexity AI for Research
Comments
Post a Comment