Oracle's AI cloud business faces margin compression due to Nvidia's pricing power for AI chips.

Oracle's (NYSE: ORCL) ambitious foray into the artificial intelligence cloud market is facing significant headwinds, with the company's AI cloud business reporting substantially lower profit margins due to the high cost of Nvidia (NASDAQ: NVDA) chips. This revelation, surfacing on October 7, 2025, has sent Oracle's stock tumbling and ignited widespread investor concern regarding the overall profitability of AI infrastructure, prompting a re-evaluation of the sector's financial viability. The immediate market reaction saw Oracle's shares decline by as much as 7.1%, prompting analyst downgrades and considerable investor apprehension, though Nvidia's shares saw only a slight dip as the chip giant continues to benefit from overwhelming demand for its chips across the AI industry.

Internal company documents reveal a stark contrast between Oracle's traditional software business, which boasts gross margins around 70%, and its burgeoning AI cloud segment. For the three months ending in August 2025, Oracle generated approximately $900 million in revenue from renting out servers powered by Nvidia chips, yet its gross profit stood at a mere $125 million, translating to a 14% gross margin. Oracle reportedly incurred an operating loss of nearly $100 million in its latest quarter specifically from the rental of Nvidia Blackwell chips for AI cloud services. The primary culprits behind these compressed margins are the substantial cost of acquiring and operating high-performance Nvidia GPUs, such as Blackwell, H100, and H200, coupled with Oracle's aggressive pricing strategies designed to capture market share in the fiercely competitive AI cloud infrastructure space. Operational expenses, including energy consumption, labor, and data center depreciation, further exacerbate the profitability challenges.

Oracle's journey into the high-stakes AI cloud arena has been marked by rapid expansion under CEO Safra Catz and Chairman and CTO Larry Ellison. In March 2024, Oracle first highlighted burgeoning demand for its Gen2 AI infrastructure and expanded its collaboration with Nvidia to deliver sovereign AI solutions. Throughout late 2024 and early 2025, Oracle continued to report "record level AI demand," with key milestones including the deployment of one of the world's largest AI SuperComputers, scaling up to 65,000 Nvidia H200 GPUs, and a deepened alliance with Nvidia to make its AI Enterprise software and NIM microservices natively accessible through the OCI Console. In September 2025, Oracle announced a dramatic surge in its remaining performance obligations to $455 billion, largely driven by securing multi-billion dollar contracts with leading AI entities, including a widely reported five-year, $300 billion agreement with OpenAI for data center capacity.

Oracle's struggle with thin AI cloud margins is sending ripples across the technology sector. Other major cloud providers face similar challenges, with companies like Amazon Web Services (NASDAQ: AMZN), Microsoft (NASDAQ: MSFT) Azure, and Google Cloud (NASDAQ: GOOGL) also making immense investments in AI infrastructure, which is driving up operating expenses and compressing profit margins. AWS saw its operating margin drop in Q2 2025, largely due to significant spending on AI infrastructure. The "Big Three" hyperscalers are collectively projected to invest around $240 billion annually in data centers, power, and servers to meet AI demand, underscoring a broader industry trend where immense investment in AI infrastructure is colliding with the realities of profitability.