Just one customer will double Oracle’s cloud sales in 2028 • The Register
Oracle has landed a mystery customer that will add more than $30 billion to the database giant’s annual revenues, more than doubling the size of its current cloud business.
“Oracle is off to a strong start in fiscal year 2026. Our multi-cloud database revenue continues to grow at over 100 percent, and we signed multiple large cloud services agreements including one that is expected to contribute more than $30 billion in annual revenue starting in fiscal year 2028,” CEO Safra Catz said in a filing lodged with the US Securities and Exchange Commission on Monday.
The database giant posted $57.4 billion revenue for FY 2025, and reported $24.4 billion of cloudy revenue for the year, $10.3 billion of which came from sales of infrastructure-as-a-service.
The filing does not identify the mystery customer. However, on Oracle’s Q4 2025 earnings call founder Larry Ellison mentioned a new customer which “said we’ll take all the capacity you have wherever it is. It could be in Europe, could be in Asia, we’ll just take everything.”
Or perhaps the giant new customer is Chinese e-tailer TEMU, which Oracle has announced is a customer. The database giant also remains a candidate to become TikTok’s future American home.
Oracle’s approach to cloud is unique, as it happily places its own infrastructure in rival hyperscalers’ datacenters. Those rigs mostly run Oracle databases, so customers of rival clouds can enjoy low-latency access to their preferred DB running in the environment Big Red has tuned for performance.
Revenue from those deals is surging – in Q4 2025 it grew by 115 percent. “We currently have 23 multi-cloud datacenters live with 47 more being built over the next 12 months,” Oracle Chair and CTO Larry Ellison said earlier this month, adding that he expected to continue growing the segment by triple digit percent headed into the 2026 fiscal year.
Since ChatGPT’s debut in 2022, Oracle has also become a go-to destination for GPUs for hire. As you may recall, in January, OpenAI announced its Stargate initiative to blow as much as $500 billion of mostly other people’s money on AI infrastructure over the next four years with SoftBank, Oracle, and investment firm MGX playing key roles in the spending spree.
To support Stargate, Oracle will reportedly shell out $40 billion to pack the first of OpenAI’s Stargate datacenters with around 400,000 of Nvidia’s GB200 superchips. Though, as we reported at the time, packing that many GPUs into a 1.2 gigawatt datacenter campus is easier said than done.
Oracle had previously disclosed plans to deploy several hundred thousand GPUs from Nvidia and AMD. This includes a “zettascale” cluster of 131,072 Nvidia Blackwell GPUs along with an equal number of AMD Instinct MI350 accelerators.
Not all of that, we’ll note, is for OpenAI. Oracle has also signed cloud agreements with several model builders of note, including xAI and Meta. Even long-time archrival Microsoft has previously used Oracle for burst capacity for its AI-enhanced Bing search. ®
