Microsoft Stock’s Upside Now Runs Through Its Data Center Build

The bull case here turns less on whether AI demand arrives than on how fast Microsoft can deliver the orders it has already booked.

Microsoft (MSFT) has pulled ahead of the market over the past three months, returning 15.5% against 3.1% for the S&P 500, yet the stock is down 3.5% over the trailing twelve months and about 10% below its 52-week high. The upside from here does not need new demand. It needs delivery of what customers have already signed for.

A Commercial Order Book Worth Two Years Of Sales

The company’s commercial remaining performance obligation reached $678 billion in fiscal Q4 2026, against $331.84 billion of total company revenue over the trailing twelve months, a commercial order book worth roughly two years of everything the company sells. The book converts slowly, with roughly 30% expected within twelve months. The backlog is up 84% year over year but only 25% excluding OpenAI, and the gap is the significant OpenAI contracts signed during fiscal 2026. What changed is who added to it: management says the entire sequential increase came from customers outside frontier model companies. Microsoft 365 Copilot, now above 30 million paid seats, up from more than 20 million in fiscal Q3 2026, is the clearest read on those buyers.

Capacity, Not Customers, Sets The Pace

Microsoft’s constraint is not demand: by its own account customer demand continues to exceed available capacity, and has for a number of quarters. So the buildout throttles revenue. The company added 31 data centers across five continents in fiscal Q4 2026, 88 across the fiscal year, and over that year cut dock-to-live times for new GPUs in its largest regions by nearly half. Management says efficiency gains are monetized in the quarter they are won. Azure revenue grew 43% as reported in fiscal Q4 2026 against 40% as reported in fiscal Q3 2026, and the guide for fiscal Q1 2027 is roughly 45% in constant currency, which is not the same basis. That acceleration was bought with throughput, not with new orders. A bull case resting entirely on one buildout is a different proposition from the Trefis High Quality Portfolio, which does not depend on the handful of largest technology names to produce its returns.

The Margin Line That Decides Whether The Build Pays

The reward for being right is not small: MSFT has gained more than 30% inside two months four times since 2010, the earliest of them in 2015. The doubt is whether the buildout eats the return it creates. Capital expenditures were $41 billion in fiscal Q4 2026, a figure that already carries higher component pricing. Against that, management guides fiscal 2027 revenue and operating income to double-digit growth with full-year operating margins down less than 1 point, even as it expects capital expenditures to grow again.

Trailing-twelve-month operating margin is already 46.8%, level with its own three-year peak, so the guide asks the company to give up less than a point from a three-year high rather than climb back toward one. Windows OEM and Devices revenue is guided to decline in the high teens in fiscal 2027 and the low 20s in fiscal Q1 2027, and that drag is already inside the same guide. Stay inside that point and the upside case is real; lose more and the capacity is being bought at a worse price than advertised. A screen of the companies whose guidance keeps climbing is where the answer shows first.

Owning The Build Is Still Owning One Name

A company executing this well is still a single position, not a system for compounding. The Trefis High Quality Portfolio is built differently, spreading the work across a rules-based set of businesses. That portfolio has a track record of outpacing the three major indices – the S&P 500, S&P Mid-cap, and Russell 2000.