Power approval set to delay Oracle's Wisconsin AI datacenter

Oracle's 902MW Wisconsin AI datacenter, requiring 1.3GW of electricity, could miss its planned 2027 capacity start because its grid connection awaits Wisconsin regulatory approval. Aterio's base case forecasts partial power only in December 2027 and full supply in October 2028, with commissioning likely pushing meaningful customer load ramp-up into Q1 2028; an extended review could delay full supply to April 2029. The risk follows Oracle's force majeure notice on its New Mexico Project Jupiter and comes as fiscal-2027 capex guidance rises to as much as $95B, though Oracle says the New Mexico project remains on schedule and Tencent reportedly signed a $7B processor-leasing deal.

Analysis

The key equity issue is not AI demand but conversion timing: delayed energization pushes OCI revenue recognition and GPU utilization to the right while construction, equipment deposits, and power-capacity commitments continue. That combination can widen Oracle’s free-cash-flow deficit and raise incremental financing costs precisely when investors are underwriting a steep AI-cloud revenue ramp. A several-quarter delay would also increase the risk that customer contracts require alternative capacity, credits, or reprioritization across Oracle’s existing fleet.

Oracle’s competitive disadvantage versus MSFT, AMZN, and GOOGL is increasingly grid optionality rather than accelerator access. Larger hyperscalers can redirect workloads across established regions; Oracle’s concentrated, very large campuses create greater exposure to a single transmission approval, commissioning sequence, and utility interconnection timetable. The second-order beneficiary is not necessarily a Wisconsin utility equity, but competing cloud platforms with immediately deployable capacity that can absorb overflow demand from customers unwilling to wait for dedicated Oracle capacity.

The force-majeure posture matters because it suggests Oracle is preserving contractual flexibility while its capex program remains exceptionally large relative to internally generated cash flow. That is more material for ORCL’s valuation multiple than for OWL unless the project subsidiary has meaningful recourse, impairment, or fee-related exposure at the parent. Near-term sentiment can remain supported by large AI contract announcements; the 1-3 month catalyst is evidence of revised commissioning dates, financing terms, or cloud-capacity guidance, while the 6-18 month risk is lower-than-modeled return on invested capital from underutilized delivered hardware.

Contrarian view: the market may overreact if Oracle can redeploy GPUs to other live campuses or if counterparties bear a meaningful portion of delay costs. The bearish thesis is falsified by unchanged OCI growth guidance, disclosure that committed capacity is being delivered through alternate regions, and financing that does not materially worsen net-debt/EBITDA or interest expense.

Key Decisions for Investors

• Buy ORCL 6-9 month put spreads rather than establish an outright short ahead of the next earnings/capex update; target a structure beginning 5-10% below spot and funded with a 20-25% downside strike. The thesis is a guidance or cash-flow reset, while defined risk protects against AI-contract headlines and successful capacity substitution.
• Run a 3-6 month relative-value hedge: short ORCL versus long MSFT or AMZN, sized beta-neutral. The trade isolates Oracle’s execution and funding sensitivity against hyperscalers with greater regional capacity flexibility; exit if Oracle reiterates OCI growth and demonstrates alternative live capacity for delayed workloads.
• Do not short OWL on this information alone. Monitor for disclosures of project-level guarantees, balance-sheet marks, deferred management fees, or recourse to the parent; absent those, the likely economic exposure is too indirect for a high-conviction equity position.
• Set a catalyst watch for regulatory docket milestones, revised power-delivery dates, and Oracle’s next capex/financing disclosure. Escalate the ORCL downside position only if management maintains revenue timing while capex or interest expense rises, or if customer commitments are explicitly deferred rather than reallocated.