AI News Roundup: Anthropic Files for IPO, NVIDIA Unveils RTX Spark, Trump Signs AI Cybersecurity Order

Anthropic Files for IPO, Setting Up a Race to a Trillion-Dollar Listing

Anthropic confidentially filed its S-1 prospectus with the SEC on June 1, making the Claude AI maker the first major frontier lab to officially begin the process of going public. The filing comes on the heels of a $65 billion funding round that valued Anthropic at $965 billion — eclipsing OpenAI for the first time and crowning it the world’s most valuable startup.

The numbers behind the filing are staggering. Anthropic’s revenue run-rate hit roughly $47 billion in May, up from about $10 billion a year earlier, and the company has told investors it expects to report its first profitable quarter this month. Investment bankers widely expect the listing to debut above the $1 trillion mark, likely targeting an October 2026 window — well ahead of OpenAI’s September IPO target. Fortune called it the potential “opening of the floodgates for the IPO market.”

NVIDIA Unveils RTX Spark Superchip at Computex, Reinventing the Windows PC

NVIDIA CEO Jensen Huang took the stage at Computex Taipei to announce RTX Spark, a new superchip designed to turn Windows laptops and desktops into personal AI agent machines. The chip pairs 20 Arm-based Grace CPU cores with a Blackwell GPU packing 6,144 CUDA cores, fifth-generation Tensor Cores, and up to 128GB of unified LPDDR5X memory — all connected via NVIDIA’s NVLink-C2C interconnect.

The result: 1 petaflop of AI performance in laptops as slim as 14mm and as light as 3 pounds. Over 30 laptops and 10 desktops from ASUS, Dell, HP, Lenovo, Microsoft Surface, and MSI will ship this fall. The move marks NVIDIA’s aggressive push into the $200 billion CPU market, aiming to own every layer of the AI stack from data center to desktop.

Trump Signs Executive Order on AI Cybersecurity

President Trump signed “Promoting Advanced Artificial Intelligence Innovation and Security” on June 2, directing federal agencies to establish a framework for the secure deployment of frontier AI models. The centerpiece: a voluntary early-access program where AI developers would give the government access to new models for up to 30 days before broader release.

The order also creates an AI cybersecurity clearinghouse to identify and share information on software vulnerabilities at scale, and directs agencies to develop benchmarks assessing AI models’ cyber capabilities. The order arrived two weeks after Trump scrapped an earlier version following industry pushback, reflecting a careful balance between national security concerns and the administration’s desire to avoid stifling innovation.

Microsoft Launches MAI-Thinking-1 and Copilot Workspace GA at Build 2026

Microsoft Build 2026 brought a flurry of AI announcements, headlined by MAI-Thinking-1 — the company’s first in-house reasoning model, built from scratch on commercially licensed data with zero OpenAI involvement. The model runs 35 billion active parameters in a sparse Mixture of Experts architecture with roughly one trillion total parameters, and a 256,000-token context window. It matches Claude Opus 4.6 on coding benchmarks and was preferred over Claude Sonnet 4.6 in blind human evaluations.

GitHub Copilot Workspace also graduated to general availability, adding Fleet mode for autonomous codebase tasks and multi-agent mode in VS Code that spawns parallel subagents for linting, testing, docs, and security review simultaneously. Starting August 2026, Project Polaris — Microsoft’s homegrown model — will replace GPT-4 Turbo as Copilot’s default engine.

Microsoft and Mayo Clinic Partner on Frontier Healthcare AI

Also announced at Build: Microsoft and the Mayo Clinic are collaborating to develop a frontier AI model purpose-built for healthcare. The model will combine Mayo’s de-identified clinical data and physician expertise with Microsoft’s AI infrastructure, targeting earlier diagnoses and more personalized treatment decisions.

Critically, Mayo Clinic will own the model, reinforcing patient trust and clinical rigor. Microsoft plans to distribute it through Azure Foundry APIs, making it accessible to healthcare organizations worldwide. It’s a significant bet that domain-specific frontier models — not general-purpose chatbots — are the future of AI in regulated industries.

‘Disrupted or Dead’: AI Crushes Pre-ChatGPT Startup Valuations

A stark CNBC analysis reveals the collateral damage of the AI boom: startups that last raised capital before ChatGPT’s November 2022 launch are watching their valuations collapse. Companies that last raised in 2021 are worth 68% less on average; those from 2022 have seen 52% declines. More than 220 former unicorns have fallen below the billion-dollar threshold.

AI companies captured 80% of Q1 2026 venture funding at roughly $242 billion, squeezing everyone else out of the capital market. Enterprise SaaS is the hardest hit: 75 SaaS companies appear on PitchBook’s fallen unicorn list, double the next-largest category. As one VC put it, “Now you’re seeing 50 engineers do what it would’ve taken 500 engineers to do five years ago.”