All the Ways Oracle's Bet on AI Data Centers Could Go Wrong

Oracle is burning through tens of billions of dollars to build AI data centers, intended for clients like OpenAI. The company surprised everyone last September when it announced orders worth $455 billion, causing its stock price to skyrocket. However, the company has now communicated that there is a risk these orders might not translate into actual earnings, meaning it may not recoup its colossal investments.

There is a risk that Oracle is betting on a losing horse: AI data centers.

"To grow our OCI business, which requires additional computing capacity, we must incur significant capital and operating expenses to increase our existing capacity in data centers and to establish data centers in new geographic locations," Oracle writes in a filing to the SEC, the regulatory agency for financial markets in the USA. The company warns that "if we are unable to secure new capacity in data centers at affordable prices or if we do not carefully plan and manage our infrastructure capacity requirements, our profitability could decline."

The reason is straightforward. Oracle is betting everything on its future as a cloud operator powering AI, based on theoretical orders from OpenAI that amount to over $300 billion over the next five years and an additional $155 billion from other clients. In theory, Oracle has a bright future ahead.

However, the problem is twofold: on one hand, there is the risk of underestimating demand, which would leave Oracle lagging behind competitors; on the other hand, there is the risk of overestimating demand or that clients become insolvent, which would prevent the company from recovering expenditures made for building new data centers and equipping them with everything necessary. OpenAI, for example, is still extraordinarily in the red, making its failure a possibility that cannot be overlooked. There is always the risk that "if customers do not renew contracts, we may not be able to re-lease, reuse, or allocate such capacity on acceptable terms, or at all."

But the risks associated with clients are not the only ones. There are also risks related to powering the data centers.

"We have faced, and may continue to face, challenges in securing reliable and cost-effective energy sources for our energy needs related to data centers, which are globally constrained by significant increases in demand and limited energy availability to power AI computations. In addition, energy prices can be unstable, partly due to extreme weather events and market structure in certain regions, and increases in energy prices can negatively impact our margins, especially where the price offered to customers is fixed or locked in."

And if energy is not sufficient, there’s also the problem of physical space on which to build new computing centers and local laws.

"Our expansion into data centers depends on access to suitable construction sites with the right permits; reliable and predictable energy sources; networking hardware; and availability of servers, including GPUs, memories, and other critical components. Data centers in the areas we rely upon may not be available on commercially reasonable terms or at all. Government-imposed limitations or moratoriums on data center construction in certain markets could reduce our ability to execute our expansion plans or prevent us from completing scheduled data center plans. Even where sites and capacity are available, our data center expansion plans are complex and subject to execution risks, including, among other things, delays or increases in costs related to design, engineering, permitting, construction, utility connections, device delivery, and contractor effectiveness. Our ability to build and operate data centers may also be affected by existing and evolving laws, regulations, and policies related to land use and zoning, environmental permitting, energy use, grid reliability, greenhouse gas emissions, water usage, building codes, health and safety compliance, tax incentives, and data localization."

In summary, the number of variables that can go wrong is enormous. However, Oracle states that "we have made significant investments in AI-related initiatives, including investments in infrastructure and personnel, and we expect to continue investing significant resources to build and support our AI products in support of our growth strategy. If we do not continue to invest significant resources to develop and support our AI products, we could fall behind in technological advancements and evolving industry standards, which would similarly harm our ability to compete."

In doubt, Oracle has stated it intends to spend $70 billion in capital during fiscal year 2027, up from $55 billion in fiscal year 2026. To do this, the company plans to take on new debt of $40 billion. Investors do not seem pleased, as the stock has lost 40% of its value over the last month.