Oracle expects to increase OCI margins by 30-40%

Oracle has posted revenue for Q2 FY26 of $16.1bn, a 13% increase year-over-year, which it said marks three consecutive quarters of double-digit growth.

The enterprise software provider said it experienced significant growth in both cloud infrastructure and applications. Its second quarter cloud revenue, which includes infrastructure as a service and software as a service, was $8bn, up 34%. Overall, cloud revenue now accounts for 50% of the company’s overall revenue.

Nevertheless, during the earnings call, some of the questions directed at Oracle executives suggested that the financial markets are concerned about the costs involved in supporting this growth. “We incur no cash expenses until that’s fully delivered and provisioned and fit for purpose,” said Oracle CEO Clay McGuirk in response to a question about how the company’s cash flow relates to datacentre builds.

He discussed three ways Oracle finances its datacentre growth. The first is a traditional hosting model, where Oracle customers bring their own hardware into Oracle datacentres. In other cases, McGuirk said customers may rent capacity once it is provisioned. Then there is the model where Oracle pays upfront for the hardware.

“That’s obviously the most cash intensive upfront,” he said. “Then there’s a depreciation schedule over the next several years.”

When asked about margins for artificial intelligence (AI) workloads for Oracle Cloud Infrastructure (OCI), which are expected to increase by between 30-40% over the life of a customer contract, McGuirk said Oracle’s focus has been to accelerate capacity delivery to improve margins and achieve its target profitability for AI workloads.

“As we go through this build-out phase, right now we’re in a phase of very rapid build-out without the majority of the capacity online,” he said. “As we actually get the majority of this capacity online – and that’s really our focus – the best way to improve margins quickly is to actually go out and deliver capacity faster. That ends up very rapidly ensuring that we get to that 30-40% gross margin profile for all of the AI datacentres.”

Other questions focused on Oracle’s ability to upsell to customers building AI on OCI. Responding to this line of questioning, Oracle executive chairman and chief technology officer Larry Ellison spoke about Oracle’s AI Data Platform.

“We built an AI lake house we call the AI Data Platform that actually points to and vectorises all of your data, whether it’s in an object store in different clouds, a bespoke application or another database,” he said. “It really will take the universe of your data, catalogue that data, vectorise it, and allow an AI or LLM [large language model] to do multi-step reasoning on all of that data.”

According to Ellison, this means Oracle is able to take all of a business’s enterprise data and unify it in a way that a user is able to ask a single question, and the AI models can find the answer to that question regardless of how the data has been stored in. “That’s really a unique proposition, and we think that thing is going to turbocharge the use of our database and the use of our cloud dramatically,” he added.

Among the interesting questions posed by financial analysts during the earnings call was one asking about Oracle’s ability to repurpose datacentre infrastructure if a customer is unable to remain in contract. McGuirk responded, discussing the benefits of bare metal virtualisation.

“What we deliver for our AI infrastructure is exactly the same cloud that we deliver for all of our customers,” he said. “We made specific choices at the beginning of OCI around bare metal virtualisation and the way in which we do things like secure wipe of hardware.”

As an example, McGuirk said that anyone with a credit card can sign up for any one of the OCI regions. “You can spin up a bare metal computer as quickly as a few minutes,” he said. “At the end of that, you can turn it off, and I will recycle that and hand it to another customer in less than one hour.”