Microsoft Rewires Azure Around Artificial Intelligence

Scott Guthrie’s message to staff on Monday did not announce a product. It rearranged who reports to whom inside the part of Microsoft that rents out the machines artificial intelligence runs on.

Guthrie, who leads Microsoft’s cloud and AI group, told employees that the company is overhauling the leadership of Azure, consolidating internal units and reorienting the business around AI, according to The Information, which reported the changes the same day. The central move widens the brief of Arun Ulag, who previously ran Azure’s database products. He will also oversee the core Azure business and its main engineering teams, and supervise the executives who run Azure Core, the unit that supplies computing capacity to customers including OpenAI.

The executives under Ulag include Girish Bablani, Billy Anders, Douglas Phillips and Mark Russinovich, Guthrie told staff. Charlie Bell, who stepped down as head of Microsoft’s security business earlier this year and took an advisory role, now reports to Ulag as well, according to people familiar with the reorganization.

Microsoft has not confirmed the changes. How many teams or employees they cover, when they take effect and which units are being merged have not been made public, and the company has not explained why it is making them now. Reorganizations of this kind usually matter to customers over years rather than weeks, and nothing disclosed ties the reshuffle to service, price or support changes.

The appointment fits a pattern set earlier in the year. Microsoft cut its reporting segments from three to two in September, creating Agents and Infra and Devices and Consumer, and began disclosing Azure as a separate figure for the first time. In April, Eric Boyd, who led the Azure AI platform, left to become head of infrastructure at Anthropic, a departure that opened the role Ulag is now filling. Chief Executive Satya Nadella has been handing a newer group of managers broader authority; Ulag has attended board meetings this year and is regarded inside the company as one of Nadella’s closest lieutenants, according to people familiar with the matter.

The organizational change arrived on a day Microsoft also staked out its silicon strategy. In a blog post published the same Monday, Guthrie said the company will scale its Azure Maia family to multi-gigawatt deployments beginning in 2027 and make Maia the primary system for running its own products and services, with the third generation of the chip scheduled to reach production that year.

Putting databases, core infrastructure and engineering under one executive concentrates the platform and the capacity it sells in a single chain of command. Azure Core holds the scarce resources: the accelerators, the power contracts and the buildings. One manager now answers for what customers build with and for whether the computing exists to run it. That arrangement suits a company selling AI services whose margin depends on filling data centers and moving more of the work onto its own silicon, which would reduce the share of each dollar of revenue that reaches Nvidia.

The reshuffle also reflects a change in what cloud providers compete on. The first phase of the cloud business was about regions, uptime and developer tools. The current phase is about capacity and the price of it. Microsoft, Amazon and Google are each reorganizing around the assumption that demand for AI computing keeps rising, and each is funding that bet with capital spending that will take years to earn back. In that setting the internal question is who controls the scarce capacity, and at Microsoft the answer has been narrowed to a smaller group in one reporting line.

For customers the practical list is short. Watch whether the new structure changes how capacity is allocated to the largest AI workloads, and whether Microsoft reports the AI portion of Azure more clearly once its segment structure settles. Watch, too, whether the Maia timeline holds; a chip program that slips turns a cost advantage back into a purchasing problem at exactly the moment when accelerators remain the tightest part of the supply chain.

The promotion also has an internal constituency to satisfy. Ulag comes from the data and analytics side of the business, and he is now over the teams that design and operate the physical plant, a pairing that matters when the scarce commodity is not software but electricity and silicon. Microsoft is spending more on capital projects than at any point in its history, and the executive who can tie demand forecasts to construction schedules is the one the business needs.

Two personnel details say something about priorities. Security experience is now inside the infrastructure organization rather than beside it, and the executive who ran the AI platform for Azure is doing the same work for a competitor. Neither tells a customer what to do this quarter. Together they describe a company arranging itself around the assumption that the computing behind its models is now a product in its own right.

Guthrie’s note told employees how the work will be organized. It did not tell customers what changes for them. That answer will come from the products.

Related Posts

  • October 13, 2026
  • 9 views
IMF Sees a Bond Market With Little Room for Error

The warning came from Bangkok, where more than 18,000 delegates had gathered for the annual meetings of the International Monetary Fund and the World Bank, and it was aimed at…

  • October 13, 2026
  • 10 views
Microsoft Puts Its Own Silicon at the Center of Azure

For four years Microsoft has run the largest rental business in computing on accelerators designed by Nvidia. On Monday it told customers and rivals where it intends to stand in…