Summary

  • Oracle's Q1 FY2027 (quarter ended August 31, 2026) reported revenue of $19.3 billion, up 30%, and remaining performance obligations of $664 billion, up $209 billion year over year, with more than $30 billion of additional AI cloud contracts booked in the quarter (PR Newswire).
  • The strongest diversification evidence is verbal: co-CEO Clay Magouyrk said at Oracle AI World in October 2025 that Oracle contracted $65 billion of new commitments in 30 days across seven contracts from four customers, and that "none of those customers are OpenAI" — with Meta named as one (CNBC).
  • Oracle's 10-Q schedules only about 13% of RPO for recognition within 12 months and roughly half beyond three years, and discloses no customer concentration figure (Regardsofwallstreet).
  • No gigawatt figure, contract term, or dollar size for the Meta agreement has ever been published; Reuters reported roughly $20 billion in talks in September 2025, citing a person familiar with the matter (Reuters).
  • The verifiable ledger therefore stops at aggregate megawatts and GPUs: 850 MW of additional datacenter capacity delivered in the quarter and more than 300,000 GPUs shipped since the end of Q4 FY26 (Converge Digest).

The diversification story of Oracle Cloud rests on a sentence spoken at a conference. On October 16, 2025, co-CEO Clay Magouyrk told Oracle AI World attendees that the company had contracted $65 billion of new cloud infrastructure commitments in 30 days, spread across seven contracts from four different customers, and stated that "none of those customers are OpenAI." Meta, he confirmed, was one of the four (CNBC). Nine days earlier, Reuters had reported — citing a person familiar with the matter — that Oracle and Meta were discussing a multi-year cloud computing deal worth about $20 billion, under which Oracle would provide capacity for training and deploying AI models alongside Meta's existing providers (Reuters).

Neither statement has a paper trail. Oracle has never issued a press release sizing the Meta contract, naming its duration, or converting it into megawatts. When Oracle reported Q1 FY2027 results on September 10, 2026, the release disclosed $19.3 billion of revenue (up 30%), OCI IaaS revenue of $7.4 billion (up 121%), RPO of $664 billion (up $209 billion), and more than $30 billion of additional AI cloud contracts booked — and named no counterparty for any of it (PR Newswire). The same release confirmed operational delivery: 850 MW of additional datacenter capacity and more than 300,000 GPUs delivered to AI Cloud customers since the end of Q4 FY26, nearly triple the prior quarter's volume. But delivery is an aggregate; customers are unnamed.

The 10-Q for the quarter ended August 31, 2026 completes the opacity. Its RPO recognition schedule allocates about 13% of the $664 billion backlog to the next twelve months, 37% to months 13–36, 34% to months 37–60, and the remainder thereafter — roughly half of the backlog is scheduled for recognition beyond three years. The filing attributes the year-over-year increase to "certain significant cloud contracts" without naming them, and discloses no customer concentration figure (Regardsofwallstreet). Prior BTW analysis established that the widely reported ~$300 billion OpenAI contract has never been confirmed or sized by Oracle itself (BTW).

The result is a specific structural condition: a company whose largest revenue commitment is publicly believed to be concentrated in one customer, while its management verbally asserts four-customer diversification, and where no instrument in the public record can verify either claim at the level of a named contract. The mechanism that converts announcement into capacity — Zettascale10 multi-gigawatt clusters, offered initially targeting up to 800,000 NVIDIA GPUs — attaches its gigawatt language to Stargate and OpenAI, not to Meta (Oracle).

What can be verified is one-sided. Meta is simultaneously self-building: it committed $1.5 billion to an El Paso, Texas data center, its 29th globally, expected online in 2028 and scalable to a 1-gigawatt site (Reuters). That hedge matters for reading Oracle's backlog: a customer that can substitute its own capacity has negotiating leverage that a pure captive cannot exert, which bears on how much of the $664 billion RPO converts at contract prices versus renegotiated ones.