Microsoft's Record Single-Day Market Cap Gain: What the $678 Billion Backlog Really Says

TL;DR $90 billion in revenue (+18%), net income +31%, EPS of $4.81, Azure +43%, and capital expenditures of $41 billion that came in below expectations. Commercial remaining performance obligations hit $678 billion, up 84%. My ten-year model gives a $363 to $883 range with a $571 midpoint.

Same AI Spending, Opposite Reaction

If Meta's earnings spooked the market, Microsoft's did the exact opposite. The stock jumped nearly 10% after the print, and as far as I can tell it was the largest single-day market cap gain for an individual company in history.

Start with the numbers. Revenue of $90 billion, up 18%. Net income up 31%. Earnings per share of $4.81, comfortably ahead of expectations. Total cloud revenue reached $59.3 billion in a single quarter, up 27%.

But what actually calmed Wall Street down wasn't the revenue, it was the spending. Capital expenditures came in at $41 billion, below what analysts expected. Microsoft is spending heavily and showing the returns at the same time. In the same quarter, Meta raised its spending plan and watched cash flow collapse. I'd argue the opposite stock reactions weren't about business quality at all. They were about how far the gap between spending and payback had narrowed.

There were soft spots. Personal computing declined, with both Windows and Xbox down. Nobody is buying Microsoft for Xbox, though. They're buying it for cloud and AI.

The Real Story This Quarter: $678 Billion

Commercial remaining performance obligations, meaning contracted work that hasn't been billed yet, hit $678 billion, up 84%. I think this is the single most important line in the release.

The reason is simple. Revenue is history. Backlog is the future. That $678 billion isn't projected revenue, it's signed revenue. It means Fortune 500 companies are locked into multi-year commitments for Azure and Microsoft software, and almost nobody in tech has visibility like that right now.

Then there's Azure itself. Cloud revenue grew 43% year over year against the 40% Wall Street expected, and Azure crossed $100 billion in annual revenue for the first time. The part that matters is direction: the biggest growth engine is speeding up, not slowing down. At this scale, that usually goes the other way.

The third leg is Copilot, which has scaled to 30 million paid enterprise seats. This is the line that separates Microsoft from most companies burning money on AI. They aren't acquiring new users, they're cross-selling premium upgrades to users already sitting inside Office 365. The distribution was already built, so the AI investment pays off now rather than someday.

What the Numbers Look Like Underneath

Behind the $450 share price sits a $3.4 trillion market cap and a $3.6 trillion enterprise value. The roughly $200 billion difference is net debt. Against $67 billion of free cash flow last year and $133 billion of net income over the trailing twelve months, that debt load isn't something I'd call a problem.

Return on invested capital has averaged 25% over five years and 17.4% last year. Margins are pointed firmly upward: 34% over ten years, 37% over five, and 40% over the last one. I like that shape. Profitability rising as scale rises means pricing power is intact.

Revenue growth tells a similar story: 13.8% over ten years, 14.5% over five, 16% over three. Not as flashy as Meta, but accelerating at this size is arguably more impressive.

The stock trades around 25 times earnings. Compared with other names in this market, that isn't expensive, provided net income keeps climbing.

The dividend is the item that nags at me. The yield is only 0.77% because the market cap is so large, but in dollars it consumes $26 billion of cash flow. If the plan is to spend far more on capex from here, the question of where that dividend gets funded from doesn't go away.

Running the eight checks, Microsoft looks a lot like Meta: two fail. They also haven't repurchased many shares. Which is a little funny, because this stock traded as low as $350 not long ago. More on that in a moment.

The Ten-Year Model and the Price I'd Pay

Analyst estimates first. Earnings per share go from $17 to $40 over the next seven years, better than 10% annually, with one down year in the middle. Revenue more than doubles from $335 billion to $760 billion, roughly 11 to 12% a year.

My assumptions: revenue growth of 7%, 10%, and 13%, reflecting that Microsoft's historical growth is lower than Meta's but has been improving. Net margin and free cash flow margin at 34%, 37%, and 40%. Exit multiples ten years out of 19, 23, and 27 times earnings. Required return of 9%, same as I used for Meta.

The output: a low price of $363, a midpoint of $571, and a high of $883.

Here's the part I actually care about. This stock traded at $350. That's below even the low case I just calculated. Same company, same business, same Azure, same Copilot, and yet buying at $450 and buying at $350 produce completely different expected returns.

The story matters in investing. But if you overpay for the story, you can be right about the story, even more right than you expected, and still end up with a bad investment. Price determines return. The more you pay, the less you get. The less you pay, the more you get. If that sounds like a cliché, go run the $350 versus $450 math and it stops sounding like one.

The Bear Case: Three Risks

First, the spending is still enormous. This quarter looked great, but infrastructure commitments are tracking somewhere between $175 and $190 billion. That's a serious bet, and the concern is straightforward: what if returns on invested capital can't keep pace with how fast the asset base is growing? Working now doesn't guarantee the math works forever at this scale. Think back to the fiber build-out 25 years ago, where demand eventually proved real and investors still suffered for years.

Second, what they're buying doesn't last. A large share of the spend goes into GPUs and hardware that depreciate quickly. These are not twenty-year assets. Technology moves fast, so they lose value fast and need replacing. That cycle creates constant margin pressure and is already weighing on free cash flow relative to operating income. Cash isn't flowing as freely as the headline numbers suggest.

Third, and this one is more philosophical but still real: as AI agents get smarter, people may stop using software directly. If users interact with bots instead of apps, the traditional software model gets threatened, and Microsoft ends up funding the thing that disrupts its own business.

Of the three, I think the third is the most underpriced. The first two are already discussed everywhere and can be tracked in the financials. Nobody has put a number on the third one yet.

I've looked at Microsoft from other angles in Is Microsoft a Dying Business? The Real Reason It's Down 20% This Year and Microsoft: A 52-Week Low and the OpenAI Stake the Market Isn't Pricing In.