Hey there, bargain hunter.
Here is the question the market is refusing to ask out loud: how does a company post 18% revenue growth, 40% cloud growth, and $37 billion in annualized AI revenue — and still sit 25% below its 52-week high?
That is not a rhetorical question. That is the trade.
What the price action is actually saying
Microsoft closed around $384 on July 1. Its 52-week high is $555.45. The stock is well below its 52-week high, trading beneath its 200-day moving average. The broader market is at all-time highs. The S&P 500 is posting its longest weekly winning streak since late 2023. And MSFT is stuck, grinding sideways while its own fundamentals keep improving.
The disconnect is real. And it is the entire story.
What the stock is processing is a market that spent 2024 and 2025 pricing Microsoft as if the AI revenue would arrive immediately and cleanly, and is now repricing for the reality that this is a 3-to-5-year infrastructure cycle with messy free cash flow in the middle.
That repricing, though? It has gone too far.
The numbers that should be getting more attention
Microsoft reported revenue of $82.9 billion for the quarter ended March 31, up 18% year-over-year, with EPS of $4.27 on a GAAP basis, up 23%. EPS of $4.27 cleared the $4.06 adjusted consensus by 5.22%. Those are not the numbers of a company losing ground.
Azure and other cloud services revenue increased 40% year over year. That beat the roughly 39% analysts expected. Management then raised Q4 guidance to 39% to 40% constant currency — above the prior Street consensus of 37% — and said growth should modestly accelerate in the second half of calendar 2026.
Slight tangent, but it matters: The CEO explicitly stated that every per-user business at Microsoft — productivity, coding, security — would transition into a per-user and usage business. That is not a branding exercise. That is a structural rewrite of the revenue architecture. The traditional SaaS license model imposes a mathematical ceiling on revenue per seat, and once the seat base is saturated at the dominant player in a category, the only remaining lever is pricing. Consumption-based layering removes that ceiling entirely.
GitHub Copilot moved to consumption-based pricing effective June 1, 2026. That is the first visible proof of the model shift. More will follow.
The capex fear is real — but it is being priced wrong
Here is where the bears have a point. Microsoft reported $31.9 billion in fiscal Q3 capital expenditures and finance leases, up 49%. Gross margin, at 67.6%, was the narrowest since 2022, as depreciation costs mounted in connection with the company’s data center infrastructure build-out.
Free cash flow fell to $15.8 billion, down from $20.3 billion a year earlier, against reported net income of $31.8 billion. That gap between accounting profit and real cash generation is what the market keeps coming back to.
But here is the thing. Management raised the Q4 guide to 39% to 40% and explicitly indicated that growth should show modest acceleration in the second half of calendar 2026 compared with the first half. That language is incompatible with the framing of a business hitting a supply ceiling or running into demand fatigue. Companies that are genuinely supply-constrained manage expectations down. Microsoft just managed them up.
The backlog number is the one most people scroll past. Microsoft now has $627 billion in commercial remaining performance obligations, encompassing unearned revenue and amounts that will be recognized as revenue. That is more than two full years of current revenue sitting in contracted form. Whatever the near-term FCF noise looks like, the revenue pipeline is not the problem.
What the valuation actually looks like right now
Microsoft’s trailing P/E ratio is currently approximately 22x, against a five-year median of roughly 34x. Out of roughly 97 analysts covering the stock, most rate MSFT as a Strong Buy or Buy, with an average price target in the range of $589 to $592. Analysts highlight that the current valuation — trading at roughly 21x to 22x next-twelve-months earnings — is the stock’s cheapest forward multiple since 2023.
Wedbush’s Dan Ives reiterated outperform with a $625 target, stating the market is underestimating Azure’s growth trajectory. Bernstein raised its price target to $641, noting the engine of growth is strong and getting stronger, citing Azure growth exceeding expectations.
A Jefferies CIO survey found that Azure is now the primary cloud provider for 55% of U.S. CIOs, compared to 28% for AWS — a 27-point gap, massively wider than the 7-point spread in the December 2025 survey. That is enterprise market share shifting in real time.
The bull case, the bear case, and what invalidates both
The bull case is straightforward. Azure reaccelerates in H2 as new data center capacity comes online. Copilot scales past 4.4% commercial penetration. The company had over 20 million paid Copilot seats as of Q3, with more growth guided for Q4. Consumption-based revenue layers on top of the seat base, expanding revenue per customer without requiring new sales cycles. The stock re-rates from a 22x earnings multiple back toward its historical 30x to 34x range. That math produces a materially higher share price without requiring the business to do anything extraordinary.
The bear case is more uncomfortable. Azure capacity constraints are expected to persist through 2026, with only modest acceleration in H2 of the calendar year. The company is guiding $190 billion in capex for 2026, up 61% from 2025. If that spending does not produce the Azure growth acceleration management is guiding for, the FCF pressure continues and the multiple stays compressed. The litigation risk — class action lawsuits with a lead plaintiff deadline of August 11 — adds headline noise for at least another 12 months.
What invalidates the bull thesis entirely: if Copilot fails to break out of early-adopter territory and competing AI assistants offer comparable capabilities at lower cost, the per-seat expansion stalls and the consumption model never materializes at scale. That is the scenario the market is partially pricing. It is probably not the right scenario, but it is not zero either.
July 29 is the number that matters
The single metric to watch at Q4 fiscal 2026 earnings — expected July 29 — is Azure constant-currency revenue growth against management’s guided range of 39% to 40%. Any result above that range, paired with a sequential improvement in free cash flow margin, is the first concrete signal the infrastructure cycle has peaked.
That is three and a half weeks away.
The stock is at a three-year valuation low. The backlog is at an all-time high. The capex cycle is fully known and priced. The FCF trough is likely close. And Michael Burry disclosed a bullish long-dated position through December 2028 LEAP call options — not exactly a sign of someone expecting permanent impairment.
The market has been punishing MSFT for a spending cycle it can model while discounting a monetization ramp it cannot yet fully see. That gap between what is visible and what is incoming is where this trade lives.
July 29 is when the gap starts closing — or it does not. Everything until then is just waiting.