OpenAI is accelerating a fundraising round of about 30 billion dollars that could value the company at more than 1.4 trillion dollars, Reuters reported on Monday, citing people familiar with the matter. The push is deliberately timed ahead of Anthropic's expected stock market debut in mid-November. Existing backers SoftBank Group and Abu Dhabi-based MGX are in talks to join, and OpenAI met with long-only funds and sovereign wealth funds last week, in the same week Anthropic was holding its own investor day. OpenAI and SoftBank declined to comment to Reuters. Nothing is finalised, but the compressed timetable, described as unusually fast for a deal this size, is the story in itself.
The valuation jump is steep even by this market's standards. OpenAI was last valued at 852 billion dollars in March, in a 122 billion dollar round that included an Amazon commitment of up to 50 billion dollars. The company is now pitching annualized revenue of 50 billion dollars, expected to reach 70 billion dollars by year-end, helped by its latest models and price cuts on cheaper models that increased usage. A public Anthropic listing would give investors a market-tested comparison OpenAI would rather avoid for now. As one capital markets lawyer put it to Reuters, OpenAI may want to set its own valuation and not have to worry about a comp.
There is a practical reason for the hurry beyond bragging rights. OpenAI needs the capital for its massive infrastructure commitments, and a stronger balance sheet matters if weaker public markets delay its own planned listing in 2027. The backdrop is not friendly: investors have grown more selective amid rising interest rates and tech stock volatility, shares and debt of cloud provider CoreWeave came under pressure last week over nearly 50 billion dollars in borrowings, several planned listings have been pulled or paused, and a widely watched semiconductor index has shed 1.6 trillion dollars in market value in less than two weeks, according to the Reuters reporting. Raising at this scale, in this tape, is a test of whether the frontier labs still set their own terms.
While the money race ran, Microsoft was selling the other half of the enterprise story: control. On October 19 the company expanded Project Perception, its agentic security system, to discover, risk-score and supervise the AI agents companies run internally, treating them, in its words, as agentic insiders. New Guardian Agents monitor other agents continuously and can throttle permissions, demand human approval for high-risk steps, or shut an agent down, while new agents start with no authority and receive only task-scoped, time-limited permissions. Microsoft said its detection now covers 60 agent types, up from 25, including Claude Code, OpenAI Codex and GitHub Copilot, with the controls in preview through Microsoft 365 E7.
A smaller deal the same day pointed the same direction. Reuters reported on October 20 that Boston-based ASI raised 325 million dollars in a Series C led by Spark Capital, valuing the company at 3 billion dollars. ASI builds simulation and monitoring software to forecast disruptions across air traffic, power grids and military operations, and it already holds a five-year, 99 million dollar U.S. Army contract and a 12-year, 875 million dollar Federal Aviation Administration contract. For builders, the pattern across all three stories is consistent: capital is flowing to whoever can prove control, whether that means securing compute before a rival prices the market, supervising agents before they act, or monitoring physical systems before they fail. The model is no longer the whole product. The controls and the balance sheet around it are part of the product too.
