Microsoft is tightening its approach to data center deals with so-called “neocloud” providers, a move that sent shares of several AI infrastructure companies tumbling this week. The software giant is weighing fewer third-party projects and pulling back from partnerships it views as higher risk, according to a Bloomberg News report.
Neocloud providers—companies that build and operate data centers specifically for AI workloads—include names like Nebius, IREN, and Nscale. These firms often rely on a handful of large customers, such as Microsoft, to anchor their projects. When a hyperscaler like Microsoft signals it may sign fewer such deals, the market takes notice.
After the report, Nebius fell about 14%, IREN dropped about 17%, and CoreWeave declined roughly 10%. The sharp moves reflect how sensitive these stocks are to any hint that the biggest buyers of AI computing capacity are becoming more cautious.
Why Microsoft is pulling back
Microsoft is not abandoning data centers—it continues to build its own. But it is reconsidering how many third-party projects it wants to rely on, and it is demanding more flexibility in how it secures computing capacity. The company wants “flexibility and optionality,” rather than being locked into projects that can slip on permitting, power, or design changes.
Execution has been a concern. IREN said some deliveries for Microsoft AI infrastructure are moving back after Microsoft requested network-design changes. IREN said those change-driven delays won’t trigger financial penalties, but the shift still signals that Microsoft is willing to alter plans midstream.
Microsoft has also backed away from specific commitments. It had planned to anchor Nscale’s Loughton, Essex site, but power-connection delays pushed usable supply into the 2030s, so Microsoft stepped back. It also chose not to expand an existing agreement tied to a Texas site.
These moves are part of a broader pattern. As we noted in our coverage of Microsoft's tightening terms on neocloud deals, the company is increasingly focused on controlling costs and reducing risk in its massive AI infrastructure buildout.
What it means for neocloud providers
For companies like Nebius, IREN, and CoreWeave, Microsoft’s pullback is more than just a lost contract. When a hyperscaler agrees to be an “anchor tenant,” it doesn’t just promise future demand—it often provides the predictable revenue that lenders need before they’ll fund a build. If that anchor steps back, third-party developers can look less “bankable,” meaning financing may get pricier or harder to secure, and timelines can stretch.
That’s why the news matters beyond one UK site or one Texas option. It raises the funding and delivery risk across neocloud pipelines, which can feed into lower valuations and more volatile share prices for operators.
The broader AI data-center buildout has already been under scrutiny. Earlier this year, doubts about the pace of construction erased $2 trillion from the Nasdaq-100, as we reported. And permitting issues can stall projects—a Texas permit freeze could affect a quarter of US data-center plans, as we explained.
What investors should watch
For everyday investors, the key takeaway is that the AI infrastructure boom is not a one-way bet. While demand for computing power is real, the companies building that capacity face execution risks—power delays, design changes, and the whims of their largest customers.
Microsoft’s move is a reminder that even the biggest tech players are willing to walk away from deals if terms don’t suit them. That can be good for Microsoft’s bottom line, but it adds uncertainty for the neocloud providers that depend on those contracts.
Investors in these stocks should watch for signs of further pullbacks or renegotiations. The market will also be listening for any commentary from Microsoft or the neocloud companies about the pace of new deals. As we saw with IREN's revenue beat despite Microsoft delays, even good news can be overshadowed by concerns about the relationship.
Ultimately, Microsoft’s tighter approach is a signal that the era of open-ended AI spending may be maturing. That’s not necessarily bad for the sector long-term, but it means more volatility and more scrutiny of every deal announcement.
