Google yesterday released a preview version of Gemini 3 Pro, with the new AI model immediately topping performance benchmarks and powering a range of Google experiences including AI Mode in search.

Why it matters: The rollout builds on Google's recent momentum, including reports that a future custom model could power Apple's Siri.

Driving the news: Google says Gemini 3 Pro is capable of more nuanced and complex answers and can generate interactive graphics on the fly. Google also says the new model is less likely than others to say what it thinks users want to hear — a known issue with some of ChatGPT's recent releases.

• Gemini is available to all users in the Gemini app, though rate limits will be higher for paid subscribers. Developers will also have access to Gemini 3 Pro via Google's Vertex AI and AI Studio.
• Google also introduced a new agentic tool called Antigravity that lets developers describe apps more conceptually, instead of coding it line by line.
• A more advanced Gemini 3 Deep Think will remain in testing before launching first for AI Ultra subscribers.

The big picture: Gemini 3 arrives amid fierce competition among the leading AI players. OpenAI released GPT-5 in August and updated it last week to 5.1, with additional personality options.

• Elon Musk's xAI released Grok 4.1 on Monday, saying the update is far less likely to hallucinate than previous versions.

Between the lines: The Gemini 3 rollout came later than some expected. Google said it wanted to allow for more testing and integration time.

• "One of the things we really wanted to optimize for with this model is putting it out broadly," Google senior director Tulsee Doshi told Axios.

Google DeepMind CEO Demis Hassabis was more direct.

• "We are the engine room of Google, and we're plugging in AI everywhere now," Hassabis told Wired.

What we're watching: Doshi said to expect future versions of Gemini 3 optimized for running locally or delivering results that are faster and more cost-efficient.

• "We'll build the family out for Gemini 3."

Growing concern about an AI bubble — fueled by companies investing in and buying from each other — serves as the pressing backdrop to Nvidia's critical earnings report later this afternoon.

Why it matters: Nvidia shares — and the broader stock market — have slipped in recent days amid debate over whether investors have gotten too bullish on the AI trade.

Between the lines: Nvidia CEO Jensen Huang offered something of an earnings preview at the Nvidia GTC event in Washington, D.C., in October, saying the company had already booked $500 billion in revenue for its Blackwell chips through the end of 2026.

• Then yesterday Nvidia announced a deal with Anthropic. The chipmaker offered at least 1 gigawatt in compute capacity, while also agreeing to invest $10 billion in the Claude maker. (Microsoft is investing $5 billion in Anthropic and bringing Claude to its Azure cloud.)
• The spotlight will also be trained on China. Huang has argued that the U.S. will lose the AI race if American companies can't sell chips to China, but defense hawks say such chips could jeopardize national security.

The big picture: What investors want to know is whether the whopping demand for Nvidia chips from the likes of Meta, Oracle, OpenAI, Tesla and other tech giants can continue unabated.

• Some observers are concerned about a circular loop in which AI giants are investing in each other and buying each other's products, potentially setting up the ecosystem for a dramatic collapse if something goes wrong.
• "Concerns regarding vendor financing loops, financial sustainability of several key AI players, and recent supply side issues have increased," Stifel analyst Ruben Roy wrote in a research note.

Fortune 500 leaders are panicked after Anthropic said it found evidence of Chinese state-sponsored hackers using Claude Code to automate parts of their espionage campaigns.

Why it matters: Major cyberattacks move security budgets.

Driving the news: Anthropic released a report last week detailing what it called the first known instance of a nation-state using AI agents to automate an espionage campaign.

• Anthropic said roughly 30 organizations were targeted and Claude Code automated up to 90% of the workload.

Zoom in: Since the report's release, executives have been flooding SecurityPal, a company that uses AI agents to vet the security of third-party vendors, with questions about the safety of their own tools and whether they rely on similar coding agents, SecurityPal CEO Pukar Hamal told Axios.

• "They were already asking a lot of questions about AI, but that's only gone up now since the news," Hamal said.
• SecurityPal's customers include major companies in the aviation, health care and financial services sectors, among others.

Reality check: Security researchers are questioning whether Anthropic's findings are truly the watershed moment the company suggests.

Keep reading.

• President Trump called for the federal government to have exclusive purview to regulate AI, saying state regulation is threatening to undermine AI-driven economic growth. (Axios)
• Microsoft debuted Agent 365, a new system that gives security leaders a way to track all the autonomous AI agents roaming on their systems and the data they're tapping. (Axios)
• Clara Shih, the former Salesforce AI chief who joined Meta a year ago to lead its business AI push, is leaving Facebook's parent company. (Bloomberg)
• An AI-infused teddy bear has been pulled off the market after reports of troubling interactions with kids. (Futurism)

I love the ingenuity of the person responsible for the TV graphics during a recent NBA game hit with some technical challenges. "Stats are down but he played great," read one graphic highlighting the play of Nuggets forward Tim Hardaway Jr.