AI & Tech News Roundup — May 22, 2026
OpenAI is going public at an $852 billion valuation. SpaceX filed its S-1 targeting a $1.75-2T valuation. Anthropic eyes a late 2026 IPO. Google debuts Gemini Spark as a persistent background agent, and Trump signs an AI cybersecurity directive.
AI & Tech News Roundup -- May 22, 2026
OpenAI Is Going Public. SpaceX Filed Its S-1 and It Is Full of AI Bets. Google's New Agent Never Clocks Out. The AI IPO Era Has Officially Begun.
OpenAI Targets September IPO -- Prediction Markets Already Pricing It at $1.4 Trillion
OpenAI is filing its IPO prospectus confidentially this week, with Goldman Sachs and Morgan Stanley advising the process and a September 2026 public listing as the target. The company's last private raise established a valuation north of $852 billion -- but prediction market traders on Polymarket are already pricing the IPO-day valuation at $1.4 trillion or higher, reflecting expectations of significant institutional demand when the shares actually price.
The listing would be the largest tech IPO in history by a significant margin and the clearest validation yet that frontier AI has crossed from research project to enterprise-scale business. OpenAI is generating $2 billion in revenue per month and has a stated 2030 revenue target of $280 billion.
SpaceX Files Its S-1 -- And the Filing Is Really an AI Story
SpaceX unveiled its public S-1 on May 21 for a Nasdaq listing, targeting a valuation of approximately $2 trillion. The company reported $18.7 billion in 2025 revenue but posted a $4.9 billion loss -- a number that reflects how heavily SpaceX is investing in AI infrastructure. The S-1 says SpaceX believes its total addressable market could reach $28.5 trillion, with the majority of that tied to AI.
The filing's most striking disclosures are about compute. Anthropic has committed to paying SpaceX's xAI division $1.25 billion per month through May 2029 for data center capacity -- a deal Bloomberg pegs at nearly $45 billion total. AI coding assistant Cursor is also renting xAI data center space. Through these arrangements, SpaceX is quietly becoming one of the largest AI cloud providers in the world, with aerospace as the headline and AI infrastructure as the actual business model.
Anthropic Eyes a Public Listing as Its Compute Business Takes Shape
With Anthropic committing nearly $45 billion to SpaceX for compute over three years, the company's capital requirements are becoming more visible. Anthropic is reported to be targeting late 2026 for its own IPO, with enterprise demand for its Claude model family accelerating -- particularly in regulated industries that prioritize safety documentation and interpretability over raw benchmark performance.
Three frontier AI companies are now all targeting public markets within roughly the same window. The combined capital raised from OpenAI, SpaceX, and Anthropic listings would reshape where AI infrastructure investment flows for the next decade -- and signal that the era of purely private frontier AI is ending.
Google I/O Debuts Gemini Spark: An AI That Never Clocks Out
Google used its I/O 2026 developer conference to introduce Gemini Spark, a persistent AI agent designed to run continuously in the background across Gmail, Docs, Calendar, Drive, and connected third-party apps. Unlike previous Gemini implementations that required a specific prompt to activate, Spark operates proactively -- summarizing meeting notes without being asked, flagging deadline conflicts across calendar and email, and connecting to external tools via API.
The product is a direct response to Microsoft's Copilot 365 integration and a signal that Google is done letting AI be a feature add-on. Gemini Spark is being positioned as the central operating layer across Google Workspace, with local file access rolling out later this year. For enterprise users already in the Google ecosystem, the question becomes less about which AI to use and more about whether to use anything else at all.
Trump Expected to Sign AI Cybersecurity Directive
President Trump is expected to sign an executive directive this week expanding existing cybersecurity information-sharing programs to formally include AI companies. The move would bring AI developers under the same data-sharing frameworks that currently cover critical infrastructure operators -- meaning AI companies would be expected to share threat intelligence about adversarial attacks on their models with federal agencies, and receive threat data in return.
The directive represents the clearest signal yet that the administration views AI systems as national security infrastructure rather than simply commercial software. How this intersects with the CLARITY Act -- currently heading for a Senate floor vote -- remains an open question as both bills move through Congress simultaneously.