When an enterprise CX or IT leader signs a Microsoft 365 Copilot agreement today, they are not simply licensing an AI productivity tool. They are, whether they know it or not, placing a bet on a financial structure that ties Microsoft's future revenue projections to the continued spending habits of a single external company: OpenAI. Understanding that structure and the risks embedded within it is quickly becoming part of responsible enterprise procurement.
The numbers are striking. According to recent financial analysis, OpenAI's activity on Azure accounts for approximately 49% of Microsoft's $627 billion in Remaining Performance Obligations (RPO) - the metric that captures future contracted revenue.
Put simply, nearly half of what Microsoft has told investors it expects to earn is contingent on OpenAI continuing to route its workloads through Azure.
What Is Circular AI Investment - and Why Does It Matter to CX Buyers?
To understand the risk, it helps to understand the mechanism. Bloomberg's detailed mapping of AI circular deals lays out the structure: tech behemoths invest in AI labs; those labs are contractually required to spend the capital back on the investors' cloud infrastructure; that compute spend is booked as revenue; and the cycle repeats.
Microsoft invested over $13 billion in OpenAI. OpenAI runs its models on Azure. Microsoft books that as cloud revenue.
Approximately 96% of Microsoft's OpenAI investment has already returned as Azure revenue - a figure that underscores just how closely the two companies' fortunes are intertwined.
Yet, with OpenAI’s recent legal issues, their wave of ambitious spending commitments totalling $600 billion, and Anthropic’s latest $1 trillion valuation, their long-term stability seems more tenuous than it did a year ago.
For tech mainstay Microsoft, such close ties may prove troublesome in the long term and have downstream effects on consumers. Microsoft has positioned Dynamics 365 Contact Center - its Copilot-first CCaaS platform - as a cornerstone of its enterprise customer experience strategy.
The system’s entire AI layer sits on Azure, meaning the product roadmap, pricing trajectory, and feature velocity for these tools are all downstream of a financial structure that most procurement teams have never stress-tested.
Ranjan Roy, Technology Journalist:
"The risk isn't that Microsoft collapses. It's that if OpenAI's trajectory changes, a substantial portion of Microsoft's forward revenue projections gets repriced, and Copilot's roadmap sits inside that uncertainty."
Has the April 2026 Restructure Change the Risk Picture?
On April 27, 2026, Microsoft and OpenAI announced an amended partnership that ended their exclusivity arrangement - a deal that had bound OpenAI exclusively to Azure since 2019. The headlines suggested a loosening of ties, but the reality is more nuanced.
Microsoft retains its license over OpenAI's models through 2032. OpenAI's payments to Microsoft continue through 2030, now subject to a cap. The catalyst for the renegotiation is OpenAI's $50 billion deal with Amazon, making AWS the exclusive distribution channel for its enterprise "Frontier" platform, signalling that OpenAI is actively diversifying its cloud dependencies.
Analysts suggest that lock-in has not disappeared; it has moved. The risk is now less about cloud infrastructure dependency, and more about what industry observers describe as "AI ecosystem alignment, agent orchestration, workflow control, and data governance."
For contact center and CX deployments - where agentic AI is deeply woven into workflow automation, CRM integration, and customer journey orchestration - that is a meaningful distinction.
Is Copilot Delivering Enough Value to Justify the Risk?
The adoption data compounds the concern. Despite Microsoft committing to a $150 billion annual capex run rate, only 15 million users held paid Copilot seats as of January 2026, representing just 3.3% of the 450 million-strong Microsoft 365 commercial installed base. An independent Recon Analytics survey of over 150,000 enterprise users found Copilot's preferred tool share at only 8% when users had simultaneous access to Copilot, ChatGPT, and Gemini.
For CX teams that have invested in Dynamics 365 Contact Center or Copilot Studio integrations, this gap between spend and adoption is not just a Microsoft problem - it is their problem. Satya Nadella has pledged to "save hundreds of millions with contact center GenAI.” But at $30 per user per month for Microsoft 365 Copilot, a 10,000-seat contact center deployment represents $3.6 million in annual licensing before a single agent workflow is automated.