AI isn’t being rejected – it’s being reassessed. As Microsoft acknowledges the gap between promise and proof, enterprise buyers are pushing for fewer, better-defined AI use cases tied to real productivity and workflow outcomes.
Omdia view
Summary
Microsoft CEO Satya Nadella recently highlighted the importance of drawing a clear line between AI promises and proof. Whilst not declaring AI a bubble, Nadella was clear that if AI economics and benefits remain concentrated amongst technology companies and if productivity gains fail to materialize across the broader economy, then slow adoption and scrutiny will only intensify. This aligns well with the sentiment of enterprise buyers, as this piece will show.
This isn’t an AI rejection – it’s an AI reality check
What’s emerging isn’t a bubble bursting or a wave of never-ending anti-AI sentiment and hysteria; it’s a recalibration. It’s a classic pattern in enterprise technology adoption. We saw similar things in the late 1990s and early 2000s with the dot-com recalibration; in the 2010s with cloud computing; and, more recently, with big data being sold as transformational. Each pattern looks different and has its own nuances and characteristics. Still, each has shown hype outrunning execution, and business buyers have paused to reassess how value can become more specific, not abstract.
Recent Omdia workplace transformation research, conducted with more than 500 senior IT and business leaders, highlights that AI adoption isn’t stalling; instead, buyers are becoming more selective. They get the grandiose AI vision and benefit, but they are increasingly demanding solid proof points and propositions over where it demonstrably removes friction, cost, or delay. The resistance here isn’t to the technology, but to AI being positioned as transformational before it proves useful, a critical distinction. The business AI market is early: things are uneven, and messaging is crowded, all of which are contributing to the weak execution we are experiencing.
This mirrors something else the survey data clearly showed: organizations have an appetite for platforms, not point solutions. Enterprises, especially, are far more interested in coordination and orchestration than in owning yet another AI feature or solution.
Microsoft is financially protected, but credibility is on the line
Microsoft is one of the least financially exposed players in the current AI ecosystem, as its vast infrastructure spend here continues to reinforce its core business and operations (data centers, cloud consumption, productivity ecosystem, etc.). This essentially means that the vendor can capture value from the money it has invested in AI, even if AI apps struggle commercially. Our research consistently shows that Microsoft remains the dominant platform inside enterprise IT. This isn’t because enterprise buyers love every feature offered, but because they recognize the challenges and risks of fragmentation as a bigger risk than imperfect execution. This means that ensuring AI is being optimally introduced, integrated, and experienced inside existing products and workflows is vital.
From a workplace productivity perspective, Copilot is where Microsoft’s strategy and approach are, and will continue to be, judged. For enterprises, AI is most valuable when it is deeply embedded into existing workflows, grounded in enterprise context, and aligned to specific tasks. Results from our recent workplace transformation survey emphasize this, with just under half of respondents stating that integrating with existing platforms and workflows is their biggest AI adoption headache. What is vital for Microsoft and other vendors delivering AI capabilities to improve employee productivity is a shift away from feature-led rollouts and hysteria toward outcome-led execution. Enterprise buyers are looking for AI designed to remove friction from specific tasks, not to demonstrate technical capability simply. Vendors must ensure the value delivered is immediate and measurable. If AI solutions feel generic, interruptive, or insufficiently grounded in the platforms and workflows they are meant to enhance, business traction will suffer.
Our research shows that enterprise buyers are increasingly intolerant of technologies that demand investment without clearly displacing cost, effort, or risk elsewhere. That applies to AI as much as anything else.
So, what happens now?
The current media talk of bubbles and AI backlash doesn’t signal the decline of the technology, but more a grounding of where businesses are in reality. We are already seeing a shift in sentiment and adoption towards fewer, better-defined AI use cases and solutions that are less about novelty and more about improving accessibility, practical productivity gains, measurable workflow improvements, and removing friction from everyday work. For vendors, this means delivering AI solutions that are richly contextualized within existing platforms, embedded directly into workflows, tied clearly to defined outcomes, and supported by evidence that value is being created, not merely promised. AI confidence and adoption will improve among enterprises when vendors deliver these capabilities credibly, repeatedly, and in the language of business outcomes.
Appendix
Further reading
Survey data taken from Omdia’s Workplace Transformation & Enterprise Platforms Study (email below for information on access)
Frank Landymore, “The CEO of Microsoft Suddenly Sounds Extremely Nervous About AI,” Futurism (retrieved January 22, 2026)
Kevin Okemwa “OpenAI might torch $14 billion in 2026, hitting bankruptcy by next year — burning through wads of cash, can it keep its operations afloat?,” Windows Central (retrieved January 22, 2026)
Author
Adam Holtby, Principal Analyst, Workplace Transformation
