Anthropic Files for IPO at Near-Trillion Valuation
In what could become the largest AI IPO in history, Anthropic confidentially submitted a draft S-1 registration statement to the U.S. Securities and Exchange Commission on June 1. The filing comes just days after the Claude maker closed a massive $65 billion Series H round that pushed its valuation to $965 billion — surpassing OpenAI's $852 billion valuation from March.
The company's growth trajectory has been staggering: Anthropic's annualized revenue run rate hit $47 billion in May, up from $10 billion a year ago. While the number of shares and pricing remain undisclosed, analysts project the company could go public as early as mid-August. The confidential filing process allows Anthropic to prepare without immediately disclosing detailed financials, giving the company time to finalize its offering strategy.
The filing underscores a remarkable shift in the AI landscape — Anthropic, founded just five years ago in 2021, is now poised to become one of the most valuable public companies in the world, a testament to the breakneck pace at which the AI industry has matured.
White House Issues Executive Order on AI Innovation and Security
On June 2, the White House issued a sweeping executive order titled “Promoting Advanced Artificial Intelligence Innovation and Security” that introduces a voluntary review framework for frontier AI models. Under the new policy, AI developers are asked to submit cutting-edge models to a group of federal agencies for review 30 days before public release, with the government using that window to identify potential cybersecurity and national security vulnerabilities.
The order takes a two-pronged approach: strengthening government and private-sector cyber defenses against advanced AI threats, and developing voluntary benchmarking frameworks for the secure development of frontier models. Notably, the administration opted for a collaborative rather than prescriptive approach, emphasizing that the U.S. leads in AI precisely because it avoids overly burdensome regulation.
The order also directs federal agencies to identify and challenge state AI laws deemed inconsistent with national policy, setting up a potential clash with states like Colorado, whose comprehensive AI legislation takes effect June 30, and Vermont, which just banned AI therapy bots from making mental health decisions.
Canada Launches “AI for All” National Strategy with $2B+ Investment
Prime Minister Mark Carney unveiled “AI for All,” Canada's new national AI strategy, on June 4 in Toronto. The five-year plan pledges over $2 billion in funding and aims to create up to 250,000 AI-related jobs by 2031 through a combination of legislation, infrastructure investment, and workforce training programs.
Key elements include building a world-leading public supercomputer for researchers and businesses, partnering with private capital to construct data centers that can scale to at least 100 megawatts, and launching a National AI Literacy Initiative to provide free entry-level AI training. Canada also plans to update its privacy laws to prevent surveillance pricing and is investing $50 million to expand the Canadian AI Safety Institute, plus $100 million for a Health Sector Data Space.
Carney struck a cautious tone about foreign AI platforms, warning they could be used against Canadians. Critics noted the strategy's emphasis on job creation and economic competitiveness came at the expense of detailed safety provisions and environmental protections.
MiniMax M3: Open-Weight Model Challenges Frontier Leaders
Chinese AI company MiniMax released MiniMax M3 on June 1, the first open-weight model to combine frontier-level coding, a 1-million-token context window, and native multimodal capabilities in a single package. Built on the novel MiniMax Sparse Attention (MSA) architecture, M3 delivers over 9x faster prefill and 15x faster decoding at 1M-token context compared to its predecessor — at just 1/20th the per-token compute cost.
The numbers are turning heads: M3 scores 59.0% on SWE-Bench Pro, surpassing both GPT-5.5 and Gemini 3.1 Pro on coding benchmarks. The model also supports image and video input natively, as well as desktop computer operation. Model weights and a technical report are expected within 10 days of launch.
M3's release continues the trend of Chinese AI labs releasing increasingly competitive open-weight models, putting pressure on both closed-source providers and the broader open-source ecosystem to keep pace.
GitHub Copilot's Token-Based Pricing Overhaul Sparks Developer Revolt
Since June 1, all GitHub Copilot plans have moved to usage-based token billing, replacing the previous flat-rate model. While base subscription prices remain unchanged ($10/mo for Pro, $39/mo for Pro+), developers now burn through AI credits based on actual token consumption — and the backlash has been immediate and fierce.
Power users report projected cost increases of 10x to 50x, with one Reddit user claiming their monthly bill would jump from $29 to $750, and another projecting an increase from $50 to roughly $3,000. The pain is especially acute for developers running agentic coding sessions, which consume tokens at a significantly higher rate than simple code completions.
Adding fuel to the fire, GitHub removed the fallback experience that previously let users who exhausted their allocation drop to a lower-cost model. Now, once credits run out, they simply stop. The timing is notable: Microsoft simultaneously launched MAI-Code-1-Flash, its own 137B-parameter coding model for Copilot, suggesting the token billing shift may be partly designed to steer usage toward Microsoft's more cost-efficient in-house model.