Microsoft stock surged 5.45% on Friday, May 29, closing at $450.24 as investors rewarded a week that brought both a landmark Wall Street research note and the opening of Microsoft Build 2026 — the annual developer conference that opened Tuesday at Fort Mason Center in San Francisco and serves as the year's highest-stakes test of whether the company's AI agenda can convert ambition into revenue. For investors trying to decide whether the rally has legs, the answer Morgan Stanley published on May 27 is unusually specific.
Morgan Stanley analysts Keith Weiss and Josh Baer framed Microsoft's AI infrastructure spending through a new lens: revenue per megawatt of installed datacenter capacity. Their conclusion was that Wall Street has been tracking Microsoft's capital expenditure correctly but drawing the wrong conclusion from it. The market, Weiss argued, is treating a build-out that precedes demand as evidence that monetization is failing — when it is, in fact, evidence that monetization has not yet begun.
The gap between those two readings is where Weiss sees the investment opportunity. At 40% gross margins on Azure AI, Morgan Stanley's capex-implied revenue forecast for fiscal year 2027 exceeds the firm's own bottom-up estimate by 59%. At 50% margins, that gap reaches 91%. Morgan Stanley maintained its Overweight rating and a $650 price target on MSFT.
Morgan Stanley's $650 Price Target: Revenue Per Megawatt Explained
The revenue-per-megawatt framework works like this: Weiss divided Microsoft's cloud revenue across its full ecosystem — Azure, Microsoft 365 Commercial Cloud, Dynamics 365, LinkedIn commercial operations — by its installed datacenter capacity. The result is a single productivity figure that can be tracked as the company expands.
Today, Microsoft's cloud ecosystem generates roughly $20 million to $30 million in annualized revenue per megawatt of installed capacity. That figure is expected to decline toward the high teens by fiscal year 2028. Weiss explicitly rejects the interpretation that this decline signals trouble. The company is deploying AI-specific capacity ahead of demand, he wrote in the note — meaning the megawatts go online first, and the revenue follows as customers activate workloads. The declining ratio simply reflects the timing gap.
The scale of what is being built makes the timing gap consequential. Microsoft's installed datacenter footprint is projected to grow from roughly 5 gigawatts in fiscal year 2024 to approximately 20 gigawatts by fiscal year 2028 — a fourfold increase. That commitment, confirmed in Microsoft's own earnings disclosures, includes a 2026 capital expenditure budget of $190 billion — up 61% from 2025 — partly driven by higher memory component prices flagged by CFO Amy Hood on the April 29 earnings call.
Build 2026 Puts the Thesis on Stage
Build 2026 arrives at a precise moment for that thesis. Microsoft CEO Satya Nadella is scheduled to take the stage at 12:30 p.m. ET, and the announcements already confirmed by Microsoft ahead of the keynote align closely with the monetization engine Weiss is modeling.
The centerpiece is a fundamental restructuring of Copilot from a sidebar assistant into an autonomous-agent platform. At Build, Microsoft has confirmed AI agents as the new core of productivity: Office 365 Copilot is gaining persistent multi-agent capabilities in a feature called Agent Mode, scheduled to roll out to Microsoft 365 subscribers in late June 2026. GitHub Copilot is evolving from code autocomplete into an autonomous developer agent. Azure AI Foundry is being repositioned as an enterprise control tower for managing agent orchestration at scale. Windows Local AI is bringing on-device agents to qualifying PCs running Windows 11.
Among the developer infrastructure announced at Build, Copilot Runtime for Windows enables any Windows application — including legacy Win32 software — to call the same on-device AI models that power the operating system's native Copilot, eliminating the need for a cloud round-trip on every inference call. For enterprise customers in regulated industries, the new AgentGuard governance layer will enforce role-based permissions, data loss prevention, and audit logging across all agent interactions, with full integration into Microsoft Purview expected by end of 2026.
How Does Microsoft Make Money From AI Agents?
For Nadella's investor narrative, these announcements answer the question Morgan Stanley's model raises: where, specifically, does the revenue come from? Microsoft's Q3 fiscal year 2026 earnings disclosed that nearly 90% of the Fortune 500 now have active agents built with Microsoft's low-code tools, and that enterprise demand for autonomous agent capabilities has led product request lists for 18 consecutive months. That kind of installed base is precisely what Weiss is betting on. An enterprise customer already paying for Microsoft 365 does not need to evaluate a new vendor to adopt Agent Mode; the capability arrives in the product they already use, and Microsoft bills it as a feature of the tier they already pay for.
That monetization architecture is one of the structural reasons Morgan Stanley's framework implies so much potential revenue from existing capacity. The incremental cost of activating Agent Mode for an existing enterprise is low. The marginal revenue, multiplied across thousands of Fortune 500 accounts, is not.
Bear Case: $190 Billion Is a Large Check to Write
The counterargument, and it is a serious one, is that $190 billion in capital expenditure is a commitment that leaves little margin for monetization to arrive slowly. Stifel analyst Brad Reback downgraded MSFT to Hold in February 2026, writing that consensus expectations for fiscal year 2027 appear too optimistic, and flagging Azure gross margin compression as AI spending outpaces near-term revenue. Melius Research moved to Hold separately, arguing that Microsoft's flagship Microsoft 365 segment could face structural displacement from AI-native alternatives.
The Register noted in April that over the prior four quarters, Microsoft had spent roughly $97 billion on infrastructure to generate $37 billion in annualized AI recurring revenue. At that pace of capital intensity, the payback period depends heavily on whether Azure AI gross margins expand toward the 40–50% range that Weiss's upside cases assume — margins that Microsoft has not yet demonstrated at scale. CFO Amy Hood, on the April 29 earnings call, acknowledged that demand continues to exceed supply and said the company expects to remain supply-constrained at least through 2026.
Even the timing of Microsoft Build 2026 carries a data point about reliability risk: Copilot experienced a significant outage on Monday, June 1, with more than 2,600 users reporting problems on Downdetector before noon, leaving enterprise users locked out of workflows that organizations have restructured around the AI assistant.
Read more: OpenAI, Microsoft End Exclusive Partnership After 7 Years—Azure Remains As Primary Cloud Partner
The broader Wall Street consensus is, for now, firmly behind the bull camp. Based on 34 Buys and 2 Holds among analysts polled by TipRanks, MSFT carries a Strong Buy consensus. The average price target across that group sits near $557, implying roughly 24% one-year upside from current levels — a figure that itself lags Morgan Stanley's $650 by a considerable margin. Only two analysts in the full coverage universe carry ratings below Buy.
For investors sitting on the sidelines, Weiss's message heading into Build week is structural: the infrastructure is already in the ground, the enterprise customer base is already installed, and the monetization is a function of activation pace, not a question of whether demand exists.
Frequently Asked Questions
Why did Microsoft stock MSFT surge on May 29, 2026?
MSFT climbed 5.45% on May 29 to close at $450.24, on above-average volume of 77.2 million shares — roughly 124% higher than its three-month average. Investor sentiment was boosted by reports confirming $37 billion in annualized AI run-rate revenue, in-house AI model development aimed at reducing costs, and strength in Xbox and software. The move also followed Morgan Stanley's May 27 research note arguing Wall Street was underestimating the revenue potential embedded in Microsoft's AI datacenter buildout.
What is the Morgan Stanley price target for MSFT stock?
Morgan Stanley analyst Keith Weiss maintains an Overweight rating and a $650 price target on MSFT, which implies approximately 44% upside from the stock's current price of around $452. Weiss co-authored a May 27, 2026 note with analyst Josh Baer introducing a revenue-per-megawatt framework for evaluating Microsoft's AI datacenter productivity, concluding that Wall Street revenue estimates may lag capex-implied potential by as much as 91%.
What was announced at Microsoft Build 2026?
Microsoft confirmed at Build 2026 a broad expansion of AI agent capabilities across its product stack, including Office 365 Copilot Agent Mode for persistent multi-agent tasks, GitHub Copilot's evolution into an autonomous coding agent, the Azure AI Foundry enterprise agent orchestration dashboard, and Copilot Runtime for Windows enabling on-device AI inference for any Windows application. The AgentGuard governance layer for regulated enterprise use is also expected to integrate into Microsoft Purview by end of 2026. The keynote by CEO Satya Nadella is scheduled for 12:30 p.m. ET on June 2.
Is MSFT a good investment now?
Wall Street's consensus is strongly bullish, with 34 Buys and 2 Holds on MSFT as of early June 2026. The bull case — led by Morgan Stanley's $650 target — rests on the argument that Microsoft's $190 billion AI infrastructure buildout will generate far more revenue than current analyst models project. The bear case, held by analysts including Stifel's Brad Reback, argues that near-term Azure growth expectations are too optimistic and that margin compression from capex could disappoint. Investors should conduct their own analysis before making any investment decision.
ⓒ 2026 TECHTIMES.com All rights reserved. Do not reproduce without permission.
