It’s an AI boom, not a bubble…, but is that true at Microsoft?

It’s become conventional wisdom that we’re in the midst of an AI bubble. As evidence, AI naysayers point to a McKinsey report that found “nearly eight in 10 companies report using gen AI — yet just as many report no significant bottom-line impact.” They also cite an MIT report, The GenAI Divide: State of AI in Business 2025, that claimed 95% of generative AI (genAI) pilots in businesses are failing.

There’s growing evidence, though, that enterprises are now getting big payoffs from AI and we’re not in a bubble at all. 

A study in December from the VC firm Menlo Ventures, 2025: The State of Generative AI in the Enterprise, found that “for all the fears of over-investment, AI is spreading across enterprises at a pace with no precedent in modern software history…. Our latest market data shows broad adoption, real revenue, and productivity gains at scale, signaling a boom versus a bubble.”

Where is AI booming?

That deep-dive report offers plenty of good news for AI boosters. It found that AI spending in 2025 was $13.8 billion — more than six times the $2.3 billion spent on it in 2023. The report calls that “a clear signal that enterprises are shifting from experimentation to execution, embedding AI at the core of their business strategies…. [Seventy-two percent] of decision-makers anticipate broader adoption of generative AI tools in the near future.”

It’s not just the amount of money, but how it was spent that shows genAI is starting to boom, the report claims. The boom was led by chatbots like Microsoft’s Copilot and OpenAI’s ChatGPT, the brains behind Copilot. Now, though, the report says \ AI in enterprises is “increasingly distributed across departmental solutions in coding, sales, customer support, HR, and verticals from healthcare and legal to the creator economy.”

Equally important: enterprises have dramatically shifted the way they develop and deploy AI. Rather than build their own tools internally, they’re buying them from AI companies. In 2024, the report says 53%of AI solutions were bought rather than built internally, a number that jumped to 76% in 2025. The report notes “ready-made AI solutions are reaching production more quickly and demonstrating immediate value,” unlike internal solutions that take more time and don’t necessarily pay off.

How does Microsoft benefit?

What does all this mean for Microsoft? There’s evidence the company might not be helped by the boom as much as other AI companies. The financial site The Motley Fool warns, “The math behind Microsoft’s AI boom doesn’t add up.” It notes that on paper, Microsoft’s AI business is barreling along, with $625 billion in orders for services from its Azure cloud-based platform. But, it cautions, $281 billion of that is from OpenAI, which has been losing ground to competitors and might face financial trouble.

Microsoft frequently touts the success of Microsoft 365 Copilot, its AI add-in for the Microsoft 365 enterprise office suite. At the end of the last fiscal quarter, the company claimed Microsoft 365 Copilot had 15 million paid seats. That sounds substantial. But, as The Motley Fool points out, there are 450 million paid seats for Microsoft 365. SO, “despite Microsoft’s massive investments in AI, Copilot’s penetration rate is barely over 3%. Commercial customers appear to be extremely reluctant to pay for Copilot.”

Similarly, it warns that the AI coding tool GitHub Copilot has just 4.7 million paid subscribers even though GitHub itself has approximately 150 million users – a penetration rate also barely over 3%.

Microsoft fights back

Microsoft knows those numbers need to rise. So it’s introduced several new Copilot products aimed at enterprises. Copilot Cowork for Microsoft 365 does more than answer questions or work with Microsoft 365 applications; it takes actions based on what you need done. Here’s an example of how it can purportedly help people prepare for a client meeting: “Cowork pulls relevant inputs from email, meetings, and files, schedules prep time on the calendar, then produces a connected set of deliverables: a briefing document, supporting analysis, and a client-ready deck.”

The company on Monday also introduced Microsoft Agent 365, which manages enterprise AI agents, automatically determining every agent’s identity, permissions, activity, and risk level. Microsoft is also rolling out a new enterprise suite, 365 E7: The Frontier Suite, which combines Microsoft 365 with Microsoft 365 Copilot and Agent 365.

That’s all fine. But is it enough to boost the low uptake of Copilot? Unlikely. Eventually, expect Copilot to become a free or almost-free part of enterprise subscriptions for Microsoft 365, because Google does that with Gemini and Google Workspace. That will increase the use of Copilot, but it won’t bring in additional revenue.

So, are there boom times instead of a bubble bust ahead for Microsoft? A lot will depend on how well Azure does, and whether OpenAI continues to need the Azure services it claims it needs. 

Perhaps the biggest wildcard is whether Microsoft’s bet on what it calls “Humanist Superintelligence” pays off. Human Superintelligence is a series of AI-based technologies, each pointed at solving an important problem and aimed at bettering people’s lives. If it succeeds, real boom times are ahead for the company. If not, Microsoft might eventually be displaced as the world’s leader in AI.