AI market correction: CIOs step away from the hype and vendors adjust

It may be due to an overheated AI hype cycle or a decision by CIOs to scale back on their purchasing plans, but sales projections are being lowered at both Microsoft and OpenAI, as well as potentially at a host of other AI providers.

On Wednesday, The Information reported that Microsoft has reduced AI quotas for certain products after multiple sales teams failed to hit their goals, and stated that it is not the only firm that is “adjusting expectations for revenue for AI agents that automate complex tasks.” The article noted that OpenAI, for instance, recently lowered its projections for AI agent revenue by $26 billion over the next five years.

According to Sanchit Vir Gogia, chief analyst at Greyhound Research, “the pullback in AI sales quotas is not a warning sign for the market. It is a signal that the enterprise technology world is finally returning to reality after a year that felt more like a gold rush than a structured industry shift.”

Over the past 18 months, he said, “many vendors pushed targets that were far ahead of what customers could reasonably absorb. Enterprise buyers were asked to make multi-year AI commitments before they had a fair chance to test the tools, examine their integration complexity, or evaluate whether the promised gains would hold up inside their own messy, interconnected systems.”

Buyers are ‘stepping away from hype’

The slowdown in sales pressure, Gogia said, is therefore healthy because it restores balance in conversations that had tipped too far towards urgency.

He pointed out, “the gap between vendor promises and enterprise experience is the heart of this correction. Buyers are not turning away from AI. They are stepping away from hype. They are choosing to invest only where they have already seen evidence of value.”

Greyhound’s research between 2023 and 2025, he said, “shows that most organizations reached the same point at roughly the same time. They discovered that building sustainable AI outcomes requires far more groundwork than early marketing suggested. Data preparation takes time. Model behaviour needs tuning.”

Governance frameworks, said Gogia, “cannot be improvised. In many cases, the imagined benefits were simply too quick and too broad compared to what the technology could deliver once it met real production systems.”

Scott Bickley, advisory fellow at Info-Tech Research Group, suggested that the real cause for Microsoft’s reduction in quotas could be self-inflicted: “My take on Microsoft’s approach to the market with AI is one of arrogance, and one of leveraging their market position.”

From the get-go, he said, the company has offered extremely high list pricing and minimal discounting, even when a customer purchases at scale. And, he noted, “they present these products, whether you’re talking about Copilot or Azure Foundry, as if they’re fully baked-in solutions, turnkey, ready to roll, and that they drive tons of return on investment.”

Although Microsoft charges a premium price for these products, said Bickley, “the reality is they’re half baked, they’re not ready for prime time, and they’re vastly overpriced. And that doesn’t even take into consideration the talent required at the end user organization to use these tools, and re-engineer their business processes.”

He added that if he were a CIO, “I would want to take this bread crumb, this clue, and zoom out a little bit and [determine], am I really building out a proper AI strategy that encompasses all of the different components outside of the technology itself? What am I trying to accomplish with the technology?”

He added, “productivity is one piece of the equation, but to really drive value, you need to have personalization, predictive value that you don’t have today, performance, revenue-driving performance that you don’t have today.”

The AI landscape has changed in other ways as well, noted Keith Fitzpatrick, research director, enterprise software and digital workflows at The Futurum Group. He wrote in an analysis released Wednesday, “the enterprise software market shifted decisively from AI hype to embedded, operational agentic AI, with major vendors integrating intelligence directly into workflows, data layers, and multi-agent orchestration frameworks.”

AI progress comes from discipline, not drama

As deployments have scaled, he said, “the conversation moved beyond capabilities to focus on value realization, governance, interoperability, and evolving AI pricing models. Looking ahead to 2026, buyers will prioritize measurable business outcomes, rewarding vendors that can demonstrate AI-driven revenue gains, cost reductions, and operational scale enabled by unified data foundations and well-governed multi-agent architectures.”

Fitzpatrick observed that buyers are increasingly fatigued by what he referred to as “claims wars” and superlatives. In 2026, he said, procurement teams will reward vendors that can demonstrate more than just task-level efficiency, so vendors should monitor their competitors’ emerging customer case studies tied to business KPIs.

Meanwhile, Bickley’s advice to CIOs about anything involving AI is this: “Try and come to the realization that you don’t have to rush into this vortex of AI hype. You can take your time and methodically plan out what makes sense for your business, and you’re not really losing ground. The hype cycle has been so loud and so ubiquitous that it has drowned out rational logic and sound reasoning.”

Gogia echoed those sentiments. “The fever of the initial hype cycle has passed,” he said. “The technology is still powerful, but it is being evaluated with a clearer eye and steadier hand. Vendors are adjusting to this new rhythm because they recognize that trust built slowly is more valuable than revenue booked quickly.”

“[Organizations that] embrace this maturity, on both sides of the table, will shape the next decade of enterprise AI in a way that is sustainable, credible, and rooted in operational reality rather than marketing velocity,” he said. What is being witnessed currently by Microsoft and others, he noted, “is not a loss of momentum. It is a shift from performance to substance. It is a market discovering that progress in AI comes from discipline, not drama. And for the first time in this cycle, that discipline is starting to show.”