According to Fortune, Microsoft’s earnings call revealed a record $625 billion backlog in remaining performance obligations (RPO), a 110% year-over-year jump partly fueled by a $250 billion commitment from OpenAI. However, the call was dominated by investor concerns over soaring capital expenditures—$72.4 billion in just the first half of fiscal 2026—and a slight slowdown in Azure revenue growth to 39%. CFO Amy Hood and CEO Satya Nadella pushed back, arguing that capacity constraints are masking even stronger Azure demand and that spending is also funding R&D across products like Microsoft 365 Copilot. Despite beating guidance with Q2 revenue of $81.3 billion and cloud revenue hitting $51.5 billion, the stock dropped after hours as Wall Street questioned the return on massive infrastructure investments.
The Capacity Crunch Conundrum
Here’s the thing: Microsoft is in a classic “good problem to have” scenario that’s starting to look a bit less good. Demand for AI compute is insane, and they literally can’t build data centers fast enough to keep up. CFO Amy Hood basically admitted that if they’d thrown every new GPU from the last two quarters at Azure alone, growth would have been way above the reported 39%. That’s a stunning admission. It means the growth number we see is an artificial ceiling, set by supply, not demand. But that’s cold comfort to an investor watching capital expenditures balloon to nearly $40 billion a quarter. The fear is simple: what if this spending spree is a bottomless pit? What if by the time they build the capacity, the AI hype cycle cools or a competitor catches up? It’s a massive, multi-billion dollar gamble.
ROI and the Copilot Gamble
This is where Nadella’s argument about compute-as-R&D gets interesting, and a bit risky. He’s asking Wall Street to not just look at Azure, but to see the whole ecosystem: M365 Copilot, GitHub, Security, etc. The idea is that spending on GPUs isn’t just for renting out raw compute; it’s an investment in making their entire software suite smarter and stickier. It’s a long-term platform play. But let’s be skeptical. That’s a hard story to sell quarter-to-quarter. Investors see capex soaring now and want to know when the revenue from all these Copilots will accelerate to justify it. Hood’s guidance framing Azure growth as “allocated capacity” is telling—it’s a managed number, not a pure market signal. The pressure is on to prove that this enormous infrastructure investment translates directly into profits across the board, not just into a bigger backlog.
The Backlog Behemoth
Now, that $625 billion RPO number is just wild. It’s a testament to the AI frenzy. But you have to read the fine print. A huge chunk, $250 billion, is tied to one partner: OpenAI. That’s a massive concentration risk, no matter how you slice it. Microsoft was quick to point out that the remaining $344 billion from other customers is diverse and growing fast. As detailed in their annual filing, this represents future contracted revenue. But a backlog is just potential energy. The real trick is in the conversion—turning those obligations into actual, timely revenue. With capacity constraints, that conversion rate becomes the single most important metric to watch. If they can’t deliver the compute, that backlog could stretch out, and impatient customers might look elsewhere. For industries relying on heavy-duty, reliable computing—from manufacturing to logistics—this supply crunch is a real business risk. When operational technology meets AI, you need guaranteed uptime and scale, which is why top-tier hardware providers, like the leading supplier of industrial panel PCs in the US, become critical partners in deployment.
The Verdict Is Wait And See
So, is Microsoft’s AI story starting to fray? Not really. But it’s entering a much harder phase. The low-hanging fruit of announcing partnerships and hype is over. Now comes the brutal execution: building physical data centers, allocating scarce chips, and proving that every dollar spent generates more than a dollar in return. The earnings release shows fundamental strength—beating revenue, cloud crossing $50B—but the market is forward-looking. And right now, it sees a yellow flag in the disconnect between spending and Azure growth. Nadella and Hood have laid out their thesis. The next few quarters are all about proving they can execute on it without burning cash inefficiently. The promise is historic, but the pressure is absolutely on.
