Today's AI Briefing

Anthropic just quietly became the highest-revenue AI lab on the planet, hitting $30B ARR while spending a fraction of what OpenAI burns on training. Google is internally testing a 24/7 personal AI agent called Remy that can make purchases on your behalf, Nvidia bet half a billion on fiber optics with Corning to wire up the next wave of AI data centers, and Arm's earnings reveal that agentic AI is about to quadruple the CPU demand nobody saw coming. Oh, and Congress unanimously agreed on something for once: banning AI companions for kids.

01IndustrySaaStr ↗

Anthropic Hits $30B ARR, Officially Surpasses OpenAI in Revenue While Spending 4x Less on Training

Anthropic's annualized recurring revenue reached $30 billion in April 2026, overtaking OpenAI's $24-25 billion and marking the first time the younger company has led on top-line run rate. The growth trajectory is staggering: from $1B in January 2025 to $30B just 15 months later, faster than Salesforce managed in 20 years. What makes this especially notable is that Anthropic achieved it while reportedly spending roughly a quarter of what OpenAI invests in model training, fueled instead by enterprise API contracts and cloud partnerships with AWS and Google Cloud. OpenAI's chief revenue officer disputes the $30B figure, claiming it's overstated by ~$8B due to different accounting treatments for cloud partner revenue.

02ProductDroid Life ↗

Google Internally Tests 'Remy,' a 24/7 Personal AI Agent That Can Make Purchases on Your Behalf

Google is dogfooding a new AI agent codenamed Remy inside Gemini that's designed to be a round-the-clock personal assistant capable of acting autonomously, including making purchases. Remy integrates deeply across Google services like Gmail, Calendar, and more, handling complex tasks and learning user preferences over time. It's Google's clearest answer yet to OpenAI's agent push and Anthropic's Claude Code/Cowork products. No public release date has been announced, but with Google I/O later this month, all eyes are on whether Remy gets an official unveiling.

03IndustryCNBC ↗

Nvidia Invests $500M in Corning, Plans 3 New US Factories and Up to $3.2B Total to Wire AI Data Centers with Fiber Optics

Nvidia bought $500 million in Corning stock warrants as the anchor of a multiyear partnership to 10x US-based optical connectivity manufacturing capacity and expand fiber production by over 50%. The deal will fund three new advanced manufacturing facilities in North Carolina and Texas, creating 3,000+ jobs. The total potential investment reaches $3.2 billion if Nvidia exercises all share options at $180. This is a telling infrastructure play: as AI clusters scale to hundreds of thousands of GPUs, the bottleneck is increasingly the fiber optics connecting them, not just the chips themselves.

04PolicyRoll Call ↗

Senate Unanimously Advances GUARD Act to Ban AI Companion Chatbots for Minors, House Files Companion Bill

The Senate Judiciary Committee voted 22-0 to advance the GUARD Act, which would ban AI companion chatbots for minors and require age verification for all chatbots. A bipartisan House companion bill was introduced the same day by Reps. Foushee (D-NC) and Moore (R-UT). The bill imposes $250,000 penalties for offering chatbots that encourage minors to engage in sexually explicit behavior or illegal violence, and requires all chatbots to disclose their non-human status at the start of each conversation. It's the strongest bipartisan signal yet that Congress is serious about AI child safety, and it now heads to the full Senate floor.

05IndustryArm Newsroom ↗

Arm Posts Record FY2026 Results, Launches AGI CPU as Agentic AI Set to Quadruple Data Center CPU Demand

Arm reported record full-year revenue of $4.92 billion (up 23%) with data center royalties more than doubling year-over-year for the second consecutive quarter. The company's new AGI CPU, its first production silicon product for data centers, saw customer demand double to over $2 billion. The key insight from the earnings call: as AI shifts from human queries to continuous agent-driven workloads, data centers will need more than 4x today's CPU capacity, creating a $100 billion+ market opportunity by 2030. Arm is betting that the agentic AI era makes CPUs, not just GPUs, the critical bottleneck.