OpenAI Files for IPO, Microsoft Launches MAI Models, Google-SpaceX $30B Deal — AI News Briefing

Top 7 Stories

1. OpenAI Files Confidential IPO Paperwork at Potential $1 Trillion Valuation

OpenAI has submitted a confidential S-1 filing to the SEC, according to multiple reports from Reuters and Fortune. The company is reportedly targeting a valuation in the $850 billion to $1 trillion range, with annualized revenue around $25 billion. Fortune reports that OpenAI is “frontrunning the narrative” around its IPO, anticipating that details would inevitably leak.

The filing comes just days after Anthropic also filed for its own IPO, marking a historic moment where the two leading frontier AI labs head to public markets in the same week. CNN reports that “the hard part is about to begin” for these companies as they face new scrutiny from public market investors over profitability, growth sustainability, and the enormous capital costs of training frontier models.

2. Microsoft Launches MAI Model Family to Reduce OpenAI Dependence

Microsoft has unveiled its own family of AI models called MAI, including MAI-Thinking-1 — its first in-house reasoning model trained without OpenAI data. The move, covered by CNBC and Mashable, signals a strategic pivot to reduce Microsoft’s reliance on OpenAI and Anthropic for the AI capabilities powering its products.

Microsoft’s AI chief stated the company was “set free” from OpenAI to pursue its own superintelligence research. The MAI models span reasoning, voice, coding, and image generation — essentially covering the same capability space as the models Microsoft has been licensing. This development raises questions about the future of the OpenAI-Microsoft partnership as both companies prepare for public markets.

3. Google Signs $30 Billion AI Compute Deal with SpaceX

Google has entered a massive $30 billion deal with SpaceX for AI computing power, according to The New York Times and TechCrunch. The arrangement will see Google paying SpaceX approximately $920 million per month for compute capacity at xAI data centers. The deal comes shortly after SpaceX signed a similar compute agreement with Anthropic.

The arrangement is raising eyebrows ahead of SpaceX’s own anticipated IPO, with analysts questioning the pricing — one report notes Google is paying roughly 2.2x per GPU more than what Anthropic pays SpaceX. The deals underscore the intense competition for AI compute capacity and SpaceX’s emerging role as a critical infrastructure provider for the AI industry.

4. Trump Administration Moves to Pre-Check AI Models Before Release

The Trump administration is developing a framework to pre-check frontier AI models from OpenAI, Google, and Anthropic before public release, according to Cybernews. The initiative represents a shift toward proactive government oversight of AI capabilities rather than the post-deployment enforcement approach that has characterized most tech regulation.

All three AI giants are expected to attend a G7 meeting next week to discuss the framework. The pre-check system would likely focus on national security risks, cybersecurity vulnerabilities, and potential misuse scenarios. Industry response has been mixed, with some companies seeing it as an opportunity for regulatory clarity while others worry about delays and government overreach.

5. Anthropic Calls for AI Pause Framework as Capabilities Accelerate

Anthropic has publicly called for the establishment of a formal pause framework as AI capabilities approach what the company describes as a potential “point of no return.” The proposal, covered by Yellow.com, would establish clear criteria for when AI development should be temporarily halted to allow safety research and policy to catch up.

The call comes amid a week of intense activity for Anthropic, which filed for its own IPO while simultaneously navigating the ongoing U.S. export control order that forced it to disable foreign access to its most powerful models. The pause framework proposal positions Anthropic as the industry’s leading voice on responsible development, even as the company races to scale commercially.

6. Florida Sues OpenAI Over Chatbot Safety Concerns

Florida has filed a lawsuit against OpenAI over chatbot safety, according to The New York Times. The suit alleges that OpenAI failed to implement adequate safeguards to prevent its chatbots from producing harmful content, particularly for minors. The legal action adds to a growing patchwork of state-level AI regulation efforts.

The lawsuit arrives as OpenAI faces scrutiny from multiple state attorneys general and prepares for its IPO. Legal analysts say the case could establish important precedents around AI company liability for user harms, particularly as chatbots become increasingly integrated into consumer products and services.

7. Bipartisan Opposition Blocks $130 Billion in Data Center Projects

More than 75 data center build-outs worth a combined $130 billion have been successfully blocked in the first three months of 2026, according to Tom’s Hardware. The opposition is notably bipartisan, driven by fears of soaring power and water costs in host communities.

The wave of blocked projects highlights a growing tension between the AI industry’s insatiable demand for compute infrastructure and the physical limits of energy and water resources. Several major tech companies have been forced to reconsider their data center expansion plans, potentially creating bottlenecks that could slow AI model training and deployment timelines.

Trend Watch

What to Watch

• IPO window dynamics: With both OpenAI and Anthropic filing simultaneously, the market’s reception of these listings will determine AI valuations for years to come. Watch for S-1 details on burn rates and revenue growth.
• Microsoft-OpenAI relationship: The MAI model launch could signal a fundamental restructuring of the partnership that defined the early AI boom. Any public response from OpenAI will be closely watched.
• SpaceX as AI landlord: If SpaceX becomes a primary compute provider for multiple AI labs, it gains unusual leverage over the industry. Regulatory scrutiny of this concentration is likely.
• Pre-check framework details: The G7 meeting next week could produce the first concrete proposal for government pre-approval of AI models — a concept that would fundamentally change how frontier labs operate.