Microsoft's AI Spending Spree Meets a Pivotal Monetization Test

Microsoft's Q3 earnings face scrutiny over AI monetization and $625B backlog. Key questions: Copilot adoption, OpenAI dependency, and massive infrastructure spend.

As Microsoft prepares to report earnings on April 29, the central question for investors is no longer about the scale of its artificial intelligence investment, but the pace of its return. The stock, trading near €356, has shed roughly 12% year-to-date and sits nearly 24% below its 52-week high, reflecting market unease over the colossal sums being spent.

The company’s cloud division, Azure, continues to be a powerhouse, posting growth of at least 39% in each of the first two quarters of the current fiscal year. Underpinning this performance is an astonishing order backlog of $625 billion, a figure that has surged 110% year-over-year. However, a significant portion of this future revenue—$281 billion—is tied to a single partner: OpenAI. Recent reports that OpenAI has more than halved its planned computing capacity investment through 2030, from $1.4 trillion to $600 billion, suggest this backlog may be overstated, a point analysts will scrutinize during the earnings call.

Monetizing AI for the broader customer base remains the critical hurdle. While Microsoft has 450 million paying Microsoft 365 subscribers, only 15 million were using the Copilot AI assistant by the end of 2025, representing an adoption rate of just over 3%. GitHub Copilot shows stronger momentum with 4.7 million Pro-Plus subscribers, a 75% annual increase, but cross-selling potential is largely untapped. In response, CEO Satya Nadella has initiated an internal overhaul of Copilot, referred to as “Code Red,” centered on the upcoming Microsoft 365 E7 suite set to launch in May 2026.

The financial burden of this AI race is immense. Last quarter alone, Microsoft funneled $37.5 billion into its AI infrastructure, primarily for new graphics processing units. The company is accelerating its build-out, having brought the $3.3 billion Fairwater data center in Wisconsin online in April, ahead of schedule. Internally, about 30% of Microsoft’s capacity is currently dedicated to Copilot and language model development, with analysts noting this could rise to 50% while still leaving room for Azure upside.

This spending has reshaped Microsoft’s valuation. The stock now trades at a price-to-earnings multiple of around 26, down from nearly 40 a year ago and below its historical five-year average. For the first time in years, it commands a slight discount to the S&P 500. Wall Street consensus anticipates 16% revenue growth for the March quarter, with analysts like BNP Paribas’s Stefan Slowinski pointing to potential free cash flow margins around 20% as a foundation for re-rating.

With 31 analysts largely maintaining buy ratings, albeit with trimmed price targets, the upcoming report is a pivotal moment. Investors will judge whether the billions spent on data centers can swiftly translate into higher software sales and whether the E7 suite can finally unlock Copilot’s vast, dormant potential.