Oracle invokes force majeure on its $165B data center — the power bottleneck just became contractual; Nebius locks 17-21% hikes for Oct 1 and TSMC locks 3-6% for 2027, all against a 5.16% 10Y
Today in one sentence: The power bottleneck just became contractual — Oracle invoked force majeure on its $165B Project Jupiter, handing the trillion-dollar AI-infrastructure financing market its first real stress test; the same week, Nebius locked 17-21% GPU price hikes and TSMC locked 3-6% for 2027 — a three-stage supply repricing against a 5.16% 10Y.
01 — TOP SIGNALS
1. Oracle invokes force majeure on its $165B Project Jupiter: the power bottleneck moves from narrative to contract clause — the first real stress test for AI infrastructure financing
What happened. Bloomberg first, Reuters confirmed: Oracle sent a force majeure notice to Blue Owl's STACK Infrastructure over Project Jupiter, the New Mexico campus being built to support OpenAI (project size ~$165B per 24/7 Wall St). The proximate cause: an Energy Transfer natural gas pipeline feeding the site slipped from August 2026 to February 2027 on New Mexico regulatory issues. Key contract details: securing power is Oracle's responsibility under the deal; the lease cannot be terminated; Blue Owl has ~$3B of equity in the project while Oracle pays the debt costs — and the project debt already trades below 90 cents on the dollar. Oracle's official line: "Project Jupiter remains on our planned schedule" (Yahoo Finance). The first casualty on the chain was power equipment: Bloomberg — Bloom Energy, contracted for 2.45GW of fuel cells at the site, fell as much as 8% intraday (closed -3.1% at $266.65). ORCL finished -3.5% at $139.54, down 6.5% over two days (intraday low $133.48).
Why it matters. This is the first time "power = bottleneck" has entered the legal-execution layer: not a cancellation, but a redistribution of delay risk. Oracle spent $28.5B of capex last quarter (more than 3x YoY), ran -$5B of FCF, and sits on $664B of RPO — all of which assumes the buildings actually get electrons on schedule. The Reuters analysis piece: the ripples are spreading through the trillion-dollar AI-infrastructure financing market — SB Energy, developing an Ohio campus for OpenAI, delayed its IPO this week; lawyers say force majeure clauses are "becoming more common" in data center development agreements; "the financing side of the AI buildout is starting to ask much harder questions than the demand side." Note the causal direction: demand is not the problem. At a 5.16% 10Y, "who bears delay risk" is the problem.
Investment implication. Transmission: FM clauses spread → lenders reprice AI DC project debt (sub-90¢ is the first reading) → marginal funding costs rise for anyone without an investment-grade or convert channel → supply growth gets constrained while demand (Nebius hikes, MU guidance) stays strong. A "strong demand, expensive capital" regime favors players with locked power and cheap funding (NBIS-class) and punishes leveraged builders (IREN -1.9% the same day as the counter-case). Timing risk for the memory chain: if 2027-28 grid-connection delays cluster, the "when" of 2027 memory demand gets re-sequenced (SNDK -3.5% broke below its 9/18 close; FXLeaders attributes part of the memory-chain pressure to Oracle DC concerns — single source, unverified). Watchpoints: a second FM notice, the Blue Owl project debt price, whether SB Energy's IPO restarts. Falsification: project debt back near par with no new FM notices = "one-off, contained."
The full signal is members-only
Members get the complete daily edition — all Top Signals, attribution, what the market may be missing, and the catalyst calendar — plus the full searchable archive.
Get the free daily previewReady for everything? Become a member on Patreon — from $15/mo · Already a member? Sign in