Summary

  • Oracle’s proposed El Paso Electric service is for late construction, initial commissioning, emergency backup and life safety. Oracle says Project Jupiter’s planned primary supply remains a Bloom fuel-cell microgrid behind the meter.
  • The announcement combines a proposed $100 million minimum power spend over ten years with a separate proposal for up to $55 million a year in customer credits for up to three years. Those are different payment channels; neither is yet evidence of an executed agreement or a lower bill.

The revealing part of Oracle’s New Mexico announcement is the scale of the connection it describes. A planned data-centre campus with as much as 2.45 gigawatts of installed Bloom fuel-cell capacity still needs a utility agreement, but the proposed El Paso Electric service is not presented as the campus’s main source of electricity. It sits at the edge of a power design Oracle says will run primarily on a behind-the-meter microgrid.

That distinction changes how to read the headline. Oracle’s 13 October proposal says the EPE agreement would cover late-stage construction, initial commissioning, emergency backup and life-safety needs. The company says it expects to pay $10 million a year for power over ten years even if use falls below contracted levels. It says that minimum would help offset shared electric-system costs otherwise borne by EPE’s New Mexico customers. The announcement does not disclose the detailed contract or show how a particular customer’s bill would change.

Oracle’s April power announcement described up to 2.45 GW of installed Bloom capacity as the plan to power the campus. In a July explanation, Oracle said the fuel-cell system would operate behind the meter and the site’s grid connection would run through a “small, residential-scale” substation, used mainly for the office building, initial ramp and equipment commissioning, with limited backup in an emergency. These are descriptions of a planned system, not proof that the capacity is installed, operating or supplying measured output. They also make it inaccurate to describe the proposed EPE service as a 2.45 GW grid supply.

The second amount in the announcement is separate. Oracle proposes up to $55 million annually in direct credits for EPE residential and small-business customers for as long as three years; the company says credits could begin in 2027. The release’s “at least $155 million” framing joins the ten-year $100 million power-payment floor with one year at the proposed $55 million credit ceiling. If the credits reached the full ceiling for all three years, the arithmetic across both channels would be $265 million. That is a conditional, nominal sum of unlike proposed payments, not a signed contract value or a forecast of household savings.

The power payment goes to the utility under a proposed service arrangement; the credit proposal is directed to customer accounts.

There is a live rate case nearby in the calendar, but it is not the same instrument. In case 25-00082-UT, EPE seeks approval for a net $70.5 million increase in New Mexico base rates, phased in two steps. One year of Oracle’s proposed $55 million credit ceiling is about 78% of that request as a simple ratio. The comparison says nothing about an offset: the rate case covers a utility filing across affected rate classes, while the credit proposal has its own recipients, timing and agreement. The New Mexico Public Regulation Commission’s case page lists the request and a December evidentiary hearing; it does not say Oracle’s proposed credits have been filed as a rate-case adjustment or approved.

For the market, the useful question is what the contract will make auditable. Oracle says more detail will follow filing. Until then, the public record does not show the service’s contracted demand, cost allocation, exit rights, collateral, customer-credit eligibility or how EPE would report credits applied. A large-load policy guide prepared for the commission discusses contract duration, ramp periods, exit fees, collateral and cost responsibility as general design questions. It is consultant guidance, not a binding decision and not evidence that any particular term appears in Oracle’s agreement.

The announcement is therefore a proposal about the economics of a boundary: an internally powered campus still seeks a small utility interface, promises a minimum payment around that interface and separately offers customer credits. Its commercial significance will depend on the filed terms and their treatment by EPE and regulators, not on adding the maximum headline figures. The strongest later evidence would show the signed service terms, the actual power and minimum payments, the credit rules and disbursements, and whether the credit programme has any formal relationship to the pending rate case.

Until those records exist, the plan may be material, but its claimed customer benefit remains prospective.