AI’s commercial momentum met practical limits this week. OpenAI put business revenue and a possible public listing in focus; Microsoft reportedly declined to expand a major compute commitment; and Meta bought an AI coding company. Meanwhile, Google’s school AI policy drew criticism, and US voters challenged the pace of data-center construction. Here are five developments as of the morning of November 5.
1. OpenAI says enterprise customers now generate most of its revenue
OpenAI president Greg Brockman said at a Wall Street Journal technology conference in Napa, California, on November 4 that business customers now account for a majority of OpenAI’s revenue. The company has been expanding enterprise sales, including promotion of its Codex coding agent, as it competes with Anthropic. Asked whether OpenAI would be ready for an initial public offering in 2027, Brockman replied, “I think so,” according to Reuters.
That is an assessment of readiness, not an IPO filing or a confirmed listing date. The revenue mix matters because the company’s position is increasingly measured by recurring corporate demand, not simply the popularity of ChatGPT with consumers. Reuters reported that OpenAI has also sought to accelerate fundraising following a delay to its listing plans.
2. Microsoft reportedly passes on a larger Nscale compute deal
Microsoft declined an option to more than double its existing agreement with Nvidia-backed Nscale for capacity at a Texas data center, according to a Financial Times report relayed by Reuters on November 5. The potential expansion was worth tens of billions of dollars, the report said. Reuters explicitly said it could not independently verify the account; neither company immediately responded to its requests for comment.
The reported decision follows Microsoft’s withdrawal from a separate Nscale project in Essex, England, after delays to its power connection pushed the expected supply date into the 2030s. Nscale filed for a US IPO in September and has also discussed a potential compute agreement with Anthropic. Taken together, these developments underscore how power availability, delivery schedules and financing can determine whether even substantial AI infrastructure commitments proceed.
3. Meta buys Conductor to bolster its AI coding push
Meta has acquired Conductor, a startup whose software lets engineers coordinate multiple coding agents in parallel workspaces. Meta confirmed the deal to TechCrunch, but did not disclose financial terms. Conductor founder and CEO Charlie Holtz said the team is joining Meta and expects to offer something new to users in early 2027. The existing product uses models from suppliers including Anthropic and OpenAI rather than models developed by Conductor itself.
The transition has immediate consequences for customers: Conductor says new sign-ups stopped November 4, its cloud workspaces will become read-only November 13, and the cloud service will shut down November 20, when deletion of cloud data begins. Users are advised to export or move their workspaces before then. Local software is expected to keep receiving new models and critical fixes until early 2027. For Meta, the acquisition adds a workflow interface and team to its existing coding-agent efforts; it does not establish what its next product will be.
4. Google’s school AI controls spark a Search access dispute
Google plans to switch off certain AI tools by default for K–12 student accounts next year unless administrators opt in. But schools that choose to restrict AI altogether would also lose student access to Google Search through those accounts, because Search and AI features are integrated, Google education vice president Chris Phillips told The New York Times, as reported by TechCrunch. Google says it will provide schools with more details before the changes take effect.
The policy drew criticism on November 4 from Senator Bernie Sanders, Senator Richard Blumenthal, former Federal Trade Commission chair Lina Khan and New York City mayor Zohran Mamdani. Their objection is that schools should be able to retain conventional search while limiting AI-generated answers. Google’s position is that the two are now deeply integrated; the resulting choice puts classroom autonomy and platform design in direct tension.
5. Midterm ballot measures put AI data-center expansion on notice
Opposition to data centers made a strong showing in the US midterms. Of 41 data-center-related ballot measures, all but one went in an anti-data-center direction, according to Ballotpedia figures cited in a Reuters analysis. Reuters also reported gubernatorial victories in Texas, Georgia and Ohio for candidates promising tougher rules or pauses on new projects. Investment bank Jefferies estimates Texas represents about 26% of planned US data-center capacity, making its permitting decisions particularly consequential.
Texas governor-elect Gina Hinojosa wants a pause to remain in place until lawmakers establish protections for communities and utility customers, Reuters reported. This is not a nationwide ban, and demand for computing capacity has not disappeared. But project developers may face more uncertain approval timelines and higher financing costs, especially if they have already ordered equipment or committed to customer delivery dates. The election results turn electricity, water and local consent into central AI business questions.
The common thread is execution: enterprise revenue, compute contracts, product transitions and public permission will shape what AI companies can actually deliver next.
