Microsoft latest earnings report has caused a few tremors in the market over fears that the company has over-committed its resources when investing in AI, and that same issue could also have repercussions for enterprises.
In addition, some financial analysts have concerns about the future of Open AI, in which Microsoft holds a 27% stake. Writing in the New York Times earlier this month, Sebastian Mallaby, senior fellow at the Council on Foreign Relations, said “My bet is that over the next 18 months, OpenAI runs out of money.”
In fact, AI research organization Epoch AI has suggested that the very nature of its research expenditure means that OpenAI will always be running out of money, as there’s a need to spend on new development as soon as one iteration starts earning.
Customer impact
What does this mean for Microsoft customers. There will be only the smallest minority of businesses that won’t be investing in AI in the future, so the question is, should they be wary of Microsoft’s products?
Not at all, said Scott Bickley, advisory fellow at Info-Tech Research. “The whole ‘AI trade’ is a classic ‘heads I win, tails you lose’ goldilocks situation for Microsoft. The AI business is already bleeding billions of dollars in losses for Microsoft and all the major infrastructure players, driving the current AI bubble hysteria in the markets. The company realizes that the AI arms race needs to be fought and won today, as ROI on these multi-billion dollar investments are likely to be realized over the course of many years.“
However Sanchit Vir Gogia, chief analyst at Greyhound Research, thinks that Microsoft’s customers will need to make some firm decisions after the latest earnings report.
“Microsoft’s own infrastructure costs are ballooning. Azure gross margins have compressed in recent quarters due to AI investment, and OpenAI has reportedly incurred over $12 billion in Azure usage alone. That’s not sustainable unless customer monetization ramps aggressively,” he said.
This will have implications for the way that customers use products. “What happens when revenue lags and margins squeeze? Microsoft shifts posture,” he noted. “Already, we’ve seen the removal of long-standing volume discounts in Enterprise Agreements. There’s also a clear reduction in pricing flexibility during renewals. Bundling is getting tighter, and standalone products are increasingly being framed as incomplete unless paired with AI.”
Customers who don’t opt into AI risk seeing stagnation in the innovation roadmap for the tools they use daily, he pointed out. “CIOs must read these signs for what they are: Microsoft is not threatening to cancel Word or Teams, but it is subtly recasting these as vehicles for AI consumption. Non-AI product investments will slow. Commercial terms will harden.”
However, customers should not worry about the long-term future of the company, said Bickley. “Microsoft customers can be content that the company is sitting comfortably. Should the AI hype not deliver in the near to mid term, Microsoft will be uniquely situated to curtail ancillary spending such as that in the neo-cloud [data center] ecosystem, with their core model providers, or to scoop up core infrastructure assets and IP rights should smaller players in the AI ecosystem become distressed and need to sell assets in a fire sale.”
Think about options
So, what should Microsoft customers consider when it comes to future purchasing? Gogia said they need to think carefully about their options. “Buying Microsoft today is about making an active choice to align with Microsoft’s AI-first worldview. And that means CIOs need to interrogate every part of that buying decision with new eyes,” he explained. “The first dimension to scrutinize is cost, not just upfront pricing, but total lifecycle cost. Microsoft’s AI offerings come with layered fees.”
They must consider the base Copilot license, required upgrades to higher M365 tiers, and the Azure compute needed to run inference, he noted. Then there’s the integration and rollout effort. “And,” he said, “all of that assumes steady adoption and usage, which in reality is uneven across functions and regions. Some users love Copilot. Others barely touch it. But pricing rarely reflects that nuance.”
He warned that companies would also need to be more flexible. “CIOs must push for terms that allow model-switching, service deprecation remedies, and exit clauses,” he said. “What happens if Microsoft changes its partnership with OpenAI again? What if they start prioritizing their own models over GPT? What if access terms shift mid-contract? These are live issues already impacting customers”.
Finally, CIOs should consider their own company’s technology. “How easy is it to integrate third-party AI models? Can you containerize workflows to reduce switching costs? Are you tracking where Copilot features are becoming the default interface across apps? Buying Microsoft today means committing to an architecture where Microsoft is the brain, not just the backbone,” Gogia said. “The bottom line is this: Microsoft is no longer just a vendor. It is becoming a copilot in every sense of the word.”