Microsoft’s AI Revenue Hits $37 Billion Run Rate, but $190 Billion Capex Plan Clouds the Picture

Microsoft faces critical AI monetization test at Build 2026 amid $190B capex and $37B AI revenue run rate, yet stock is down 24% from its high. Analysts see a valuation disconnect.

Microsoft will face its next critical test of artificial intelligence monetization when the Build 2026 developer conference kicks off on June 2 in San Francisco with an online component. The two-day event will pivot away from glossy presentations toward developer tooling, Azure workloads, and enterprise AI agents. For investors, the key question is whether developers can translate Copilot, Azure, Windows, and Foundry into repeatable, commercial workflows.

The infrastructure powering these ambitions is expanding at a breathtaking pace. Morgan Stanley analyst Keith Weiss projects Microsoft’s data-center capacity will balloon from roughly 5 gigawatts in fiscal 2024 to 20 gigawatts by 2028. The capital required is staggering: capital expenditures are on track to hit around $190 billion in calendar 2026, with approximately $118 billion falling in the second half alone. By comparison, full-year capex for fiscal 2025 stood at $64.5 billion. The investment is already yielding tangible products — the company recently unveiled MAI-Image-2.5, a text-to-image model that grabbed third place on the global Arena Leaderboard. Mustafa Suleyman, head of Microsoft AI, called it a “quality leap” in text rendering, layout coherence, and stylised illustrations. The model will roll out on MAI Playground and Microsoft Foundry within two weeks, targeting enterprise customers needing high-resolution commercial visuals.

Financial results underscore the operational strength behind the buildout. Microsoft reported revenue of $82.9 billion, up 18 percent, while operating income reached $38.4 billion. Net income hit $31.8 billion, with diluted earnings per share of $4.27 — both figures climbing 23 percent year over year. The cloud segment generated $54.5 billion in revenue, a 29 percent gain, and the Intelligent Cloud division brought in $34.7 billion. Azure and other cloud services advanced 40 percent, and commercial remaining performance obligations — contracted but not yet recognised revenue — surged 99 percent to $627 billion. Most notably, the AI business surpassed an annualised revenue run rate of $37 billion, marking a 123 percent jump from the prior year.

Yet the stock has failed to reflect that momentum. On Thursday, shares traded at €363.60, up 2.45 percent on the day, but that modest bounce does little to repair the damage of recent months. The stock has shed roughly 12 percent since the start of the year according to one account, or 9.91 percent by another measure, and sits more than 24 percent below its 52-week high of $467.45. The 200-day moving average remains 7.47 percent above the current price, underscoring the distance to a meaningful trend reversal.

Analysts see an unusual disconnect between Microsoft’s fundamentals and its valuation. The forward price-to-earnings multiple now hovers between 21 and 25 times — far below the five-year median of 34 times. Morgan Stanley maintains an overweight rating with a $650 price target, while HSBC analyst Stephen Bersey sets a target of $571, pointing to an annual revenue opportunity of $43 billion from the Anthropic partnership by 2030. Zacks expects earnings per share of $17.33 for the current fiscal year, representing 27.1 percent growth.

Skeptics, however, warn that the capital spending splurge is squeezing near-term margins. The restructuring of Microsoft’s OpenAI agreement — eliminating revenue-sharing payments in exchange for securing IP rights through 2032 — is viewed by Wedbush as a positive catalyst for future cash flow. But whether the AI infrastructure can convert its massive scale into profits quickly enough will be decided by upcoming quarterly results, and the market will use those numbers to judge the valuation gap.

Build 2026 therefore serves as more than a developer gathering. Microsoft’s program promises “real code, real systems, real workflows,” and sessions will cover everything from building and operating AI agents with Windows 365 — including cloud PCs, governance, identity, and enterprise controls — to GitHub Copilot CLI and coding agents that take a project from an empty command line to a deployed Azure application. The company is betting that production-grade agents, Copilot automation, and Azure consumption will reinforce each other, creating a narrative that matches the operational strength of the numbers.

For now, the technical picture remains mixed. One supportive trading session does not erase the weaker medium-term chart context, and the stock needs sustained buying to turn the overarching trend. The real test will come on June 2 and 3, when Microsoft must prove that its sprawling AI investment is finally translating into durable, commercial usage — and convince the market that the valuation discount is an anomaly rather than a warning.