This website uses cookies

Read our Privacy policy and Terms of use for more information.

KEY POINTS

- Oracle filed its fiscal year 2026 10-K on June 22, capping a year in which AI-driven cloud infrastructure demand reshaped its revenue mix and competitive positioning.

- The filing lands as Big Tech has collectively run up $750 billion in AI-related debt, and Oracle's aggressive data center expansion puts it directly inside that spending cycle as both a buyer and a beneficiary.

- Watch the $16-per-share earnings power thesis against a 10-year Treasury yield at 4.46% — if rates stay elevated, Oracle's capital-intensive infrastructure buildout faces a higher hurdle rate than the bull case assumes.

Oracle filed its fiscal year 2026 10-K with the SEC on June 22, and the document lands at a moment when the company's decade-long infrastructure transformation is finally colliding with real AI demand. The filing covers the fiscal year ended May 31, 2026 — and the numbers inside will either validate Larry Ellison's all-in data center bet or expose a capital allocation story that the market has been too generous in pricing.

The Infrastructure Pivot That Defines the Bull Case

Oracle's cloud infrastructure unit — OCI — has been the central argument for ORCL bulls for the past 18 months, and the fiscal 2026 10-K is the definitive scorecard on whether that argument held. The setup entering this filing was unusually strong: hyperscale AI demand was accelerating globally through the fiscal year, and Oracle had positioned OCI as the enterprise-grade alternative to AWS and Azure at a moment when capacity constraints were pushing customers to secondary providers. Gartner projects 40% of domestic AI data centers will face severe power constraints by 2027, and over $1.5 trillion in proposed physical infrastructure is currently trapped in permitting bottlenecks — both dynamics benefit an incumbent with operational facilities and existing grid connections.

Ellison has publicly committed Oracle to a capital expenditure trajectory that mirrors the ambitions of companies three times its size. That commitment shows up in the balance sheet mechanics of the 10-K: the debt load, the depreciation schedule, and the remaining performance obligations — the contracted future revenue that is the real-time demand signal for cloud infrastructure. Remaining performance obligations are the number traders should pull directly from the filing. If that figure accelerated meaningfully in the May quarter, it means enterprise customers are locking in multi-year OCI commitments at a pace that justifies the capex. If it decelerated, Ellison's bet is running ahead of actual booking velocity.

The competitive dynamic inside this filing is also worth isolating. Oracle is not just competing with AWS and Azure for workloads — it is competing for the same Nvidia GPU allocation that every hyperscaler needs to attract AI tenants. NVIDIA's fiscal year 2026 revenue hit $215.94 billion, up 65.47% year over year, and the company carries $145 billion in supply commitments this quarter alone. Where OCI sits in that GPU queue — and what the 10-K discloses about compute capacity expansion timelines — is a material variable for the AI revenue growth rate in fiscal 2027.

The Rate Environment That Complicates the Math

Oracle's infrastructure buildout is capital-intensive by design, and the financing backdrop for that buildout has become meaningfully more expensive than the original business case assumed. The 10-year Treasury yield is sitting at 4.46% as of June 18, and the Fed funds effective rate is 3.63% — a rate environment that directly inflates the cost of the long-duration debt Oracle issues to fund data center construction. Big Tech has collectively run up $750 billion in AI-related debt, and Oracle is a meaningful participant in that figure. Every 50 basis points of additional yield on long-term debt is a real drag on the net present value of infrastructure assets with 15-to-20-year useful lives.

The broader sector context makes this rate sensitivity more acute. Nvidia itself borrowed $25 billion on June 15 — against what it characterized as decades of AI demand — even though the company generates cash at a rate that would make the borrowing unnecessary under normal capital allocation logic. When the most profitable chip company in history is accessing the debt markets at 4.46% 10-year yields, it signals that the capex commitments across the AI supply chain are outrunning near-term cash generation for everyone in the ecosystem, including Oracle. The 10-K's interest expense line and the company's weighted average cost of debt will tell traders exactly how much of OCI's incremental revenue is being consumed by financing costs before a dollar reaches operating income.

CPI inflation running at 4.2% year over year as of May compounds the problem on the construction side. Data center builds are materials- and labor-intensive, and a 4.2% inflation rate means the cost of physical infrastructure that was underwritten at 2% inflation assumptions is running 200 basis points hot. If Oracle's fiscal 2026 capex came in above the original guidance range — which the 10-K will disclose — that overage is the first indicator that cost inflation is eroding the return on invested capital that justified the buildout in the first place.

What Traders Watch Next

The immediate trading setup around the ORCL 10-K is a read-the-filing event, not a reaction-to-guidance event — Oracle does not hold an earnings call concurrent with the annual filing. The actionable data points are sequenced: remaining performance obligations first, then cloud revenue growth rate, then capex versus prior guidance, then interest expense as a percentage of EBITDA. Those four numbers, in that order, tell the complete story of whether Oracle's fiscal 2026 was a validation year or a stretch year.

The SpaceX-Google GPU rental deal — disclosed at $920 million per month for 110,000 Nvidia chips beginning October 2026 — is directly relevant context for reading Oracle's filing. It establishes a market price for premium AI compute capacity and sets a benchmark for what OCI can charge enterprise tenants who need GPU access at scale. If Oracle's cloud revenue per unit of compute capacity is expanding toward that market rate, the margin story improves materially. If it is not, Oracle is subsidizing AI tenant acquisition to hold occupancy, which is a very different business than the one the market is pricing.

The 40-day countdown to EU AI Act enforcement on August 2 adds a forward regulatory variable. Oracle's cloud business has significant European revenue exposure through OCI Frankfurt and OCI Amsterdam regions, and enterprise AI workloads running on OCI infrastructure will be subject to the same transparency and data governance obligations as the model providers themselves. The penalty structure — up to 7% of global annual revenue — creates a compliance cost that the fiscal 2027 guidance will need to absorb. Traders should mark August 2 on the calendar as the next hard deadline for Oracle's European cloud business. The stock's next scheduled catalyst after the 10-K is the fiscal Q1 2027 earnings call, expected in mid-September — and by that point, the EU compliance cost will be a live operating expense, not a future risk.

Keep Reading