AI News: Physical Intelligence Eyes $11B Valuation, Claude Popularity Strains Infrastructure

Daily roundup for March 30, 2026 covering robotics funding surge, Anthropic infrastructure challenges, Google Gemini chat import tools, and OpenAI shutdowns

Top Stories

Physical Intelligence Nears $11 Billion Valuation in New Funding Round

Physical Intelligence, the two-year-old robotics startup founded by former Google DeepMind researchers, is negotiating a new $1 billion funding round that would more than double its valuation to over $11 billion. The deal would mark a stunning rise—just four months ago, the company raised $600 million at a $5.6 billion valuation.

Founders Fund is set to participate, with Lightspeed Venture Partners also in talks alongside returning backers Thrive Capital and Lux Capital. The company’s pitch: “Think of it like ChatGPT, but for robots.” Physical Intelligence is building general-purpose AI models that can power robots to perform diverse tasks, part of a broader industry shift toward “physical AI.”

The timing aligns with growing investor appetite for robotics. At NVIDIA’s GTC 2026 conference, physical AI dominated the agenda, and ABB announced that its integration with NVIDIA’s Omniverse simulation achieves 99% correlation between simulated and real-world robot behavior—a breakthrough for industrial automation.

Source: TechCrunch

Claude’s Popularity Surge Strains Anthropic’s Infrastructure

Anthropic is facing a good problem to have: Claude is too popular. Paid subscriptions have more than doubled this year, according to an analysis of credit card transactions from 28 million U.S. consumers. But the demand is outpacing GPU capacity.

Starting in late March, Anthropic tightened usage limits during weekday peak hours (8 a.m. to 2 p.m. ET). The company acknowledged the constraints stem from high demand exceeding available compute resources.

This comes after a turbulent March for Anthropic—14 major product launches, 5 service outages, and an accidental data leak that revealed the existence of “Claude Mythos,” a new model the company describes as “a step change” in AI performance. An Anthropic spokesperson confirmed Mythos is the most capable model they’ve built to date, currently being trialed with early access customers.

Source: TechCrunch

Google Gemini Adds Chat History Import to Poach Competitors’ Users

Google released new tools that let Gemini users import chat history and context from rival AI assistants like ChatGPT and Claude. The feature aims to eliminate the friction of starting fresh when switching services.

The move is strategic: rather than compete solely on model capability, Google is targeting the switching costs that keep users locked into competitors. Combined with the recent rollout of “Personal Intelligence” to free U.S. users—which connects Gemini to Gmail, Photos, and YouTube for more context-aware responses—Google is positioning Gemini as the hub for users’ AI interactions.

Source: Bloomberg

OpenAI Shelves Sora and Adult Mode, Pivots to Robotics

OpenAI shut down Sora, its AI video generation app, and indefinitely paused its “adult mode” ChatGPT project this week. The Financial Times reported that challenges including sexual datasets and content filtering issues led to the indefinite adult mode pause.

Meanwhile, OpenAI launched a Safety Bug Bounty program focused on AI misuse and agentic risks, offering up to $100,000 for identifying vulnerabilities. The company is also expanding shopping features in ChatGPT with richer product discovery and a new Walmart in-app experience, available to free and paid users.

Source: 9to5Mac

Quick Hits

  • AI layoffs accelerating: Tech layoffs hit 59,000 in 2026 so far. Amazon leads with 16,000 cuts, Block slashed 40% of its workforce, and Meta is reportedly planning reductions affecting up to 15,000 workers to fund $135B in AI spending. IBTimes

  • White House AI framework: The administration released a National Policy Framework for AI on March 20, recommending Congress avoid creating a new federal AI regulator. The framework prioritizes child safety, free speech protections, and preempting state-level AI laws. Nextgov

  • OpenAI’s “Spud” finishes training: OpenAI’s next model, codenamed “Spud,” reportedly completed pretraining on March 25. No release date announced. The Neuron

  • Google-Agile Robots partnership: Google DeepMind partnered with Agile Robots to integrate Gemini Robotics foundation models with Agile’s hardware, expanding Google’s physical AI footprint. CNBC

  • Humanoid robot costs dropping: Material costs for humanoid robots could fall from $35,000 in 2025 to $13,000-$17,000 by 2035, according to a new analysis. Fortune

Worth Watching

The robotics funding frenzy shows no signs of slowing. Physical Intelligence’s potential $11B valuation comes just weeks after RoboForce raised $52M and Agile Robots locked in Google DeepMind integration. Combined with NVIDIA’s simulation-to-reality breakthroughs, 2026 is shaping up as the year physical AI moves from labs to factories.

Meanwhile, Anthropic’s Claude Mythos remains under wraps but appears imminent. If it delivers on the “step change” claims, expect the AI assistant wars to intensify. Google’s chat import feature suggests they’re already preparing for competitive churn.