Everyone has been watching the wrong race. While the AI world obsesses over which lab builds the best model, a fresh CIO survey from Morgan Stanley points to a quieter fight playing out beneath the surface, one Microsoft may already be winning.
In a note shared with TheStreet on Oct. 12, Morgan Stanley published findings from its third-quarter 2026 AlphaWise CIO survey. The bank maintained its Overweight rating and $600 price target on Microsoft, against a closing price of $541.15 that day. But the more telling data was not about the stock. It was about where enterprise IT chiefs think AI money is actually heading.
Also read: Wall Street doubles down on Microsoft stock for investors
Microsoft leads where it matters most for AI
One of the more interesting questions Morgan Stanley threw into this survey was about AI orchestration. Not which model do you use, but who sits above the models, directing traffic, managing data permissions, keeping an eye on costs. Microsoft pulled 20% of responses on that one. LLM providers collectively were right behind at 17%. Google got 5%, Amazon 4%.
That finding runs counter to the popular narrative. The assumption has been that whoever wins the model race wins the enterprise AI race. But a growing number of companies are spreading workloads across multiple models rather than committing to one.
The moment a business is running different models for different tasks, it needs something above the models to keep things organized. That is the orchestration layer, and right now Microsoft is where most CIOs say they’d go for it.
Microsoft’s setup for this is Azure at the infrastructure level, Fabric for enterprise data, Foundry to select and route AI tasks, and Copilot to surface the results inside everyday applications. A company could run a competitor’s model through that stack and Microsoft still collects on the infrastructure, the data layer, and the software. Three revenue opportunities from one AI workflow.
Bloomberg / Getty Images
Spending intentions still favor Microsoft, but the gap is closing
Thirty percent of CIOs named Microsoft as the vendor most likely to capture the biggest share of generative AI spending in 2026, up from 27% the prior quarter, according to the survey. LLM providers came in at 16%, Google at 8%, Amazon at 7%.
Pull back to a three-year horizon and things get messier. Microsoft is at 28%, which was 30% just last quarter. LLM providers climbed from 17% to 21% in that same window. The vendor consolidation story that once looked inevitable hasn’t materialized.
Companies are running multiple models, and the labs behind those models are now showing up in procurement conversations they weren’t part of a year ago.
You could read that as bad news for Microsoft. Morgan Stanley doesn’t see it that way. Juggling several models creates its own headaches, and someone has to build the tooling that keeps everything organized. That is the business Foundry is going after.
The ROI problem is becoming a Microsoft sales pitch
Here is something that should give AI bulls pause. Forty-two percent of CIOs told Morgan Stanley they cannot yet measure the returns on their AI token spending. Another 28% say returns are running below what they spend. Only 13% are ahead.
Businesses are not waiting around on this. Forty-three percent already have token spend controls in place. Another 31% said they’re putting them in before the end of 2027. Add those up and you’re looking at most of the enterprise market actively trying to get a grip on what AI is actually costing them.
The implication for Microsoft is straightforward. A tool that helps a company route cheaper models for simpler tasks and save the expensive ones for work that actually needs them has a pretty clear value proposition when IT chiefs are being asked to justify every dollar.
That is Foundry’s pitch, and the survey data suggests the appetite for it is growing.
Azure is still the dominant cloud, though growth is leveling off
On the cloud side, 79% of surveyed CIOs said their organizations use Azure, ahead of AWS at 66% and Google Cloud at 43%. That maps closely to broader market data showing Azure holding the second largest share of global cloud infrastructure spend behind AWS.
Microsoft’s net score for expected IT budget share gains in 2026 came in at plus 43%, compared with plus 26% for Amazon and plus 4% for Google. Over the next three years, that gap stays wide: Microsoft at plus 35%, Amazon at plus 11%.
More Microsoft:
- Microsoft’s $665 target hinges on a new AI advantage
- Meta and Microsoft just sent employees a memo about Anthropic
- Microsoft just sent a strong message to Anthropic
The caveat is the growth rate. CIOs who use Azure expect spending to grow 15% in 2026, down from 21% in last year’s survey for 2025. AWS is right alongside at 15%, with Google Cloud at 12%. Microsoft is still the biggest cloud by usage, but the growth advantage it once had over Amazon is largely gone.
The longer runway is in AI workloads themselves. CIOs in the survey put AI and machine learning at 3.8% of their cloud spending today. In three years, they expect that to hit 9.9%. If even half of that growth flows through Azure, the demand picture looks pretty different from what current growth rates suggest. Getting there is another question.
What the data cannot tell you
CIO surveys are a snapshot of intentions. They are not contracts. A respondent who says Microsoft will get the biggest share of their AI budget in 2026 could face a CFO in January with a very different view. The orchestration preference data is especially early, given that Foundry is still a relatively new product.
Morgan Stanley has a $600 price target on Microsoft, but the range around it tells the real story. Their bull case goes to $795. Their bear case drops to $250. A $545 gap between best and worst is a pretty honest admission of how much nobody actually knows. Survey data shows where CIOs say they want to spend. Whether that translates into the numbers Microsoft needs to justify its valuation is a different conversation entirely.
What this survey really captures is a change in the question enterprises are asking. It used to be: Which AI model should we use? Now it’s: Who keeps all our models working together? That fight doesn’t get nearly the attention the model race does, but it might end up mattering more. And right now, when CIOs try to answer it, Microsoft is the name that comes up first.
Related: Meta and Microsoft just sent employees a memo about Anthropic
