Microsoft’s supply chain is being forced to keep pace with an artificial intelligence boom that has rewritten the company’s old assumptions when it comes to demand.
The recent surge in AI has created what Microsoft Cloud corporate vice president Joanna Kostecka calls an “AI arms race,” where demand for cloud capacity can jump by hundreds of percent, instead of following the steady patterns that planners once relied on.
That has made forecasting far more challenging. At the same time, Microsoft is dealing with longer lead times and tight supplies of components, along with power constraints brought on by data centers. Factor in the constant waves of disruptions to supply chains across the globe, and it becomes increasingly difficult to plan around anything resembling a stable supply-and-demand picture.
“The normal world of predictability of accurate forecasts does not exist,” Kostecka says.
For Microsoft, responding to that environment came down to how quickly it could add the data center capacity needed to meet demand. Dhaval Desai, partner engineering manager for Microsoft Cloud Supply Chain, says companies that have that capacity are able to run AI models, experiment with new applications and begin delivering value faster.
“In this AI arms race, the biggest thing is how fast you can deliver the demand or the capacity that your data center needs worldwide,” Desai says.
Cleaning Up the Process First
Microsoft’s AI journey started with a basic question: Which processes were actually worth automating?
Desai says the company first looked at its existing workflows to identify any waste, as well as unnecessary steps and areas where employees were spending too much time on manual work. The idea was to improve those processes before layering AI on top of them.
“We don't want to automate a bad process using AI as a technology,” he says.
Once those processes were simplified, Microsoft was able to use AI to automate more of the work, shorten cycle times and add intelligence to decisions that previously required more manual effort.
The company looked across nearly every part of its cloud supply chain. That included product design and planning, sourcing for components, assembling equipment in factories, delivering server racks to data centers and managing replacement parts for equipment failures.
The scope mattered because Microsoft was trying to solve for speed across an entire system. Improving one isolated step offered limited value if the process slowed down somewhere else.
Planning has gone through a similar change as well. Kostecka says the traditional planning model could take a month from the initial demand forecast through supply, inventory and executive decision-making. That pace may still work for some companies, but Microsoft found that it could no longer keep up with the swings its cloud business was experiencing.
Instead, the company began looking at the decisions that needed to happen weekly, the ones that needed to happen daily, and those that needed to reach executives immediately. It also started looking several years ahead in strategic discussions with suppliers and other partners, extending beyond the 24- to 36-month window that was common in traditional planning.
“We had to speed up, and what that meant was that this monthly cycle was no longer valid,” Kostecka says.
Why Microsoft Turned to SAP
Microsoft did not have to look far for a technology partner. Microsoft and SAP have worked together for more than three decades, with a long history of connecting their technology and business systems. But Desai says the decision went beyond familiarity.
SAP’s strength, he says, is the depth of the business processes built into its applications. Microsoft saw those established processes as a way to create more consistency across its own operation as the cloud business expanded.
The concern was that teams could otherwise build one-off processes that worked for a specific problem or group but became difficult to maintain as the company grew.
“We definitely wanted to use the SAP solutions to standardize our processes as we scale our business,” Desai says.
There was also a practical technology fit. Desai says Microsoft’s technology stack and SAP’s applications already worked closely together, which allowed the company to draw from both as it built out its AI capabilities.
Faster Decisions, With People Still in the Loop
While there’s always a financial consideration in a process like this, the benefits described by Kostecka and Desai are centered more on speed, scale and the ability to make better use of planners’ time.
Desai explains how AI has helped reduce process cycle times, all while taking on more of the tedious tasks that planners have traditionally done manually. That includes much of the spreadsheet work that often occupies far too much of their time, giving planners more room to focus on decisions that require business judgment.
The human role also remains central. Desai expects AI agents to continue to work alongside planners, performing analysis and other routine tasks while people continue to review critical business decisions.
Watch: Microsoft's AI Journey
Kostecka sees that changing what companies need from planners, too. Instead of specializing narrowly in one portion of the supply chain, planners will have to understand how decisions ripple across the business, and weigh trade-offs between power, hardware, components, customer demand and financial targets.
That’s becoming especially important as Microsoft plans around constrained supplies, creating a situation where the company needs to run multiple scenarios quickly, determine where available capacity should go, then revisit those choices as conditions change.
For Microsoft, that may be the biggest lesson from its AI journey so far. The technology can help the company move faster, but only after the underlying work is simplified enough to keep pace.
With demand changing by the week or even the day, Microsoft no longer has the luxury of spending a month building a plan and then sticking with it until the next cycle.
“You have to adjust the rhythm of the business to actually fit what your company is experiencing,” Kostecka says.
Watch the related video here.
