Goldman Sees Microsoft's AI Payoff Zone: $640 Target as Copilot Starts to Scale

Goldman's call reflects a broader shift from AI infrastructure to enterprise workflows

This is where the AI trade gets more selective. After two years of rewarding picks-and-shovels names, the market is starting to look harder for AI that lands in enterprise workflows and converts into recurring revenue the focus is beginning to shift. Goldman Sachs signaled that shift by adding Microsoft to its U.S. Conviction List, keeping a Buy rating, and raising the price target to $640 from $610.

The core debate has also changed. Bears argued that Copilot is not as good a product as it could be. Goldman now says product quality is improving, while Microsoft says Microsoft 365 Copilot has passed 30 million paid seats, up from roughly 20 million three months earlier. Goldman is also pointing to Azure acceleration, improving AI unit economics and Copilot monetization. That combination makes the bull case more concrete: Microsoft may be moving from AI investment narrative toward measurable monetization.

The monetization case is becoming easier to trace

Goldman's thesis is no longer just about building AI capacity. It is about showing how that capacity turns into revenue, margin support, and potentially stronger earnings growth.

Scale is showing up across cloud and productivity

Microsoft says its AI business surpassed an annual revenue run rate of $37 billion. Azure has also crossed a meaningful threshold, now above $100 billion in annual revenue. Taken together, those numbers suggest AI demand is feeding back into Microsoft's core cloud and software platforms rather than staying trapped in a capital-intensive buildout phase.

Revenue per user matters as much as seat growth

The stronger signal is not just more Copilot seats. It is higher revenue per user. Microsoft 365 Commercial cloud grew 19% while seats grew 6%, and the company pointed to revenue per user driven by Microsoft 365 E5 and Microsoft 365 Copilot. That is the classic software-leverage pattern: sell a higher-value bundle inside an existing distribution network instead of starting from scratch.

Goldman is connecting that operating trend to future earnings power, projecting EPS growth to accelerate from 12% in fiscal 2027 to more than 20% by 2029. Whether that target proves accurate, the direction of the thesis is clear: investors are being asked to look at recurring AI-powered revenue, not just spending.

The key test is whether Copilot becomes a workflow staple

Goldman is putting Microsoft on the U.S. Conviction List, the bank's highest-conviction ideas. In doing so, it is backing the view that Microsoft is shifting from infrastructure spend toward recurring AI-powered subscription revenue.

The bullish case rests on distribution as much as product quality. Microsoft already has AI embedded in tools millions of workers use daily, which gives it a commercial advantage over a standalone AI product trying to win adoption from scratch. Goldman is also leaning on operating proof, including product quality is improving and revenue per user driven by Microsoft 365 E5 and Microsoft 365 Copilot.

The bear case is simpler: if Copilot remains merely better rather than essential, enterprises may treat it as optional rather than embedded. In that scenario, AI investment could stay expensive without delivering the subscription scale investors are hoping for.

Signals to watch

Validation - Copilot adoption broadens beyond early adopters. - Revenue per user continues to outpace seat growth. - Operating trends begin to support the earnings-acceleration thesis.

Invalidation - Quality improves, but adoption still looks optional. - AI infrastructure investment keeps rising faster than monetization. - The market stops rewarding Microsoft for recurring AI-powered subscription revenue and returns to punishing AI spend on principle.