Oracle plunged over 7% during trading, with reports indicating that cloud profit margins are disappointing, and a loss of 100 million was incurred last quarter due to renting NVIDIA's Blackwell

Oracle's highly anticipated cloud business has been reported to have disappointing profit margins, facing profitability pressure due to the rental of advanced chips from Nvidia.

On Tuesday, October 7th, Eastern Time, media outlets cited internal documents stating that in the company's last fiscal quarter ending in August, Oracle's server rental generated revenue of $900 million and a gross profit of $125 million, equivalent to a profit of $0.14 for every $1 in sales, or a gross margin of 14%. This is lower than the gross margins of many non-tech retail companies and far below Oracle's traditional software business's overall gross margin of about 70%.

The documents indicated that in some cases, Oracle recorded "considerable" losses due to small-scale rentals of both new and old versions of Nvidia chips, with nearly $100 million in losses attributed to the rental of Nvidia's new Blackwell architecture chips in the last fiscal quarter.

Following the news, Oracle's stock price accelerated its decline towards the end of the U.S. stock market's morning session, hitting a daily low of $271, with an intraday drop of 7.1%, marking the largest intraday decline since September 11. Comments suggested that the stock price plunge was due to Oracle's cloud computing business profit margins being lower than many Wall Street analysts had expected.

The latest revelations have raised market doubts about Oracle's AI transformation profitability. As of the close on Monday this week, Oracle's stock price had risen nearly 75% this year, primarily benefiting from revenue growth expectations driven by surging AI computing demand.

When announcing its quarterly financial report a month ago, Oracle projected that its cloud computing business revenue would grow over 600% in the next three fiscal years. On the same day, it was reported that OpenAI signed a computing power contract with Oracle worth up to $300 billion over the next five years. Following these positive developments, on the first trading day after the news on September 10, Oracle's stock price surged by 36%, marking the largest single-day increase since 1992.

AI Cloud Business Faces Multiple Profitability Challenges

According to media reports on Tuesday, Oracle's AI cloud business has faced multiple financial pressures over the past year. In the past year, as sales in this business nearly doubled, the gross margin fluctuated between 10% and slightly above 20%, averaging around 16%.

The high costs of Nvidia chips are a major challenge. The cost of Nvidia server chips required for AI computing far exceeds that of traditional servers, and specialized networking equipment is also needed to connect these servers. To win large contracts from AI customers, Oracle and other cloud service providers have also offered significant discounts on graphics processing unit chip rental prices.

Reports indicate that financial analysts in the cloud computing industry generally believe that, given the high costs of Nvidia chips, AI computing is the lowest-margin business in the field. Moreover, Oracle's profit margins in this business have been lower than previous estimates from analysts at institutions like KeyBanc in recent quarters Oracle's disadvantages in data center operating models have also dragged down profit margins. Unlike Amazon AWS and Google Cloud, which own most of their data centers, Oracle primarily leases data centers from third parties. Oracle's Executive Vice Chairman Safra Catz stated in last month's earnings call that owning real estate "is really not our expertise."

Reports indicate that internal documents show Oracle's profit margins are under more pressure when deploying NVIDIA's latest chips. Due to the deployment of new NVIDIA chips for OpenAI in facilities in Abilene, Texas, in recent months, Oracle's AI cloud business gross margin has dropped from over 20% to below 15%.

In addition to whether it can build enough data centers to generate the promised revenue to investors, another challenge Oracle faces is that its cloud business heavily relies on a few customers. The top five AI cloud customers, including OpenAI and NVIDIA, contribute about 80% of Oracle's revenue from this business.

Oracle's profit margins are also affected by the actual usage and the number of paid servers by customers. According to internal documents, the utilization rate of Oracle's GPU cloud servers ranges between 60% and 90%, depending on the type of NVIDIA chips used.

Media reports have found that older NVIDIA chips seem to be boosting Oracle's profit margins, while newer versions of the chips are putting pressure on them. It was mentioned that in the last fiscal quarter, Oracle incurred nearly $100 million in losses due to leasing the newly launched NVIDIA Blackwell chips.

Documents show that part of the aforementioned losses stem from the time lag between when Oracle prepares the data centers for customers and when customers start using and paying for them. It is currently unclear what causes this gap and how Oracle plans to narrow it.

Strong Earnings Report but Market Skepticism

Oracle's earnings report released last month showed strong growth momentum. For the first fiscal quarter of 2026 ending in August this year, Oracle's cloud business total revenue reached $7.2 billion, a year-on-year increase of 28%. The revenue from cloud infrastructure (OCI) business was $3.3 billion, a year-on-year increase of 55%.

What caught the market's attention the most was that Oracle's remaining performance obligations (RPO), or backlog order value, surged 359% year-on-year to $455 billion in the first fiscal quarter. This was mainly benefited from large long-term contracts signed with AI companies like OpenAI.

When releasing the earnings report, Oracle projected that OCI business revenue for this fiscal year would reach $18 billion, a 77% increase from the previous fiscal year, and would grow to $32 billion, $73 billion, $114 billion, and $144 billion over the next four years. This means that by the fiscal year 2030, the annual revenue of this business will increase by 700% compared to this fiscal year.

However, Wall Street institutions have expressed concerns about the profitability and sustainability behind these impressive numbers JP Morgan analysts pointed out that new orders are mainly concentrated among a few AI giants such as OpenAI, leading to increased customer concentration risk. At the same time, most of the revenue comes from long-term contracts, with only about 10% belonging to short-term RPO, which will be recognized as revenue within the next 12 months. This means that the short-term performance contract amount has seen almost no growth quarter-on-quarter, and most of the revenue will only be recognized in the distant future.

Morgan Stanley also expects that out of the RPO of up to $455 billion in the first fiscal quarter, only about 10% will be recognized as revenue within the next 12 months. Additionally, current orders are mainly concentrated in large contracts from AI giants like OpenAI, and the rising customer concentration brings risks of single dependency. If major AI clients push for self-built solutions or switch cloud partners, the stability of Oracle's future contract revenue recognition will be in doubt.

Wall Street Questions Profit Outlook

Due to massive spending on chips and data center capacity, Oracle's gross margin fell to 67.3% in the most recent fiscal quarter, with media estimating this to be the lowest quarterly gross margin in over a year.

After Oracle released its earnings report, some Wall Street firms expressed concerns about its profitability while raising Oracle's target price. Goldman Sachs significantly raised Oracle's target price from $195 to $310 in last month's report but maintained a "neutral" rating, stating that the huge capital expenditures will delay cash flow breakeven until the fiscal year 2029.

Oracle's capital expenditure expectation for this fiscal year has been raised to $35 billion, a year-on-year increase of 65%. After deducting dividends, the free cash flow over the past 12 months was negative $10.9 billion. JP Morgan believes that the low profit margins of AI-related orders may drag down the growth rate of operating profit in the long term.

Morgan Stanley raised its revenue expectation for Oracle in fiscal year 2029 from $104 billion to $125 billion. However, due to the low profit margins of the newly added AI infrastructure business, it is expected that the operating profit margin will decline from 44% in fiscal year 2025 to 39% in fiscal year 2029.

Morgan Stanley analysts warned that Oracle's net debt exceeds $80 billion, with a leverage ratio maintained above 4 times, approaching the downgrade threshold set by rating agencies.

Analysts generally recognize Oracle's long-term potential in the AI field but emphasize that investors ultimately need to focus on fundamental issues such as profit margins and cash flow. Oracle's success depends on its ability to efficiently build data center capacity, maintain order growth momentum, and convert massive investments into sustainable profit growth