Microsoft’s AI business has accelerated significantly, with the company’s stock rallying nearly 30% following quarterly earnings that beat Wall Street estimates. However, according to Bloomberg analysis of Microsoft’s regulatory disclosures, the company’s AI growth carries substantial concentration risk.
Analysis of Microsoft’s filings suggests that OpenAI accounts for approximately 70% of Microsoft’s total AI-related revenue. According to the disclosures, GPU and infrastructure consumption specifically from OpenAI drives between 65 to 70% of Microsoft’s total AI business, based on projected growth and previous revenue-related disclosures.
Microsoft has reported that Copilot has millions of paid seats in Microsoft 365 and is being gradually adopted across governmental, public, and private sector businesses. However, when compared against previous disclosures, GPU and infrastructure consumption from OpenAI remains the cornerstone of Microsoft’s AI business, with training platforms like ChatGPT and serving it to customers driving the majority of revenue.
The concentration presents a significant risk profile. OpenAI is estimated to be losing between 10 and 20 billion dollars per year while generating $2-5 billion in revenue, according to the source material. Microsoft itself is among OpenAI’s largest backers, having invested billions in cash, infrastructure, compute credits, and other instruments.
As one analyst noted, “Microsoft’s AI bookings simply represent OpenAI’s compute bill packaged as revenue.” The arrangement mirrors earlier business models—Spotify, for example, ran losses for years before eventually achieving profitability. The strategy involves scaling rapidly, creating new markets, and locking in users before monetizing.
Microsoft is attempting to diversify its AI operations by training its own “more efficient” models for consumer applications and boosting external enterprise adoption. The company has also begun devoting more attention to legacy consumer businesses like Xbox and Windows.
However, the analysis raises questions about the sustainability of the current model. If OpenAI experienced significant deceleration or failure, or if competitors like Claude or Gemini gained user preference, Microsoft would face billions in sunk costs and excess data center capacity. Regulatory scrutiny, shifts in public opinion, or political opposition to new data centers could also impact the trajectory.
Key facts
- Bloomberg analysis of Microsoft’s filings indicates OpenAI accounts for around 70% of Microsoft’s total AI revenue
- GPU and infrastructure consumption from OpenAI drives 65-70% of Microsoft’s total AI business
- OpenAI is estimated to be losing $10-20 billion annually while generating $2-5 billion in revenue
- Microsoft has invested billions in cash, infrastructure, and compute credits to support OpenAI
- Microsoft is attempting to diversify by developing its own AI models and boosting enterprise adoption