Meta reserves 1 GW of Noon Energy's 100-hour storage as hyperscalers become the marginal buyer for unproven chemistries

The deal is structurally unusual before it is technically interesting. Meta has reserved up to 1 GW / 100 GWh of long-duration storageEnergy storage systems designed to discharge power continuously for 10 or more hours, filling gaps in renewable generation that span multiple days rather than just evening demand peaks. from Noon Energy, a Palo Alto company that has not shipped a commercial system, against a first-of-kind delivery of 25 MW / 2.5 GWh due in 2028. A hyperscalerA massive cloud service provider, typically Amazon Web Services, Microsoft Azure, or Google Cloud, capable of provisioning computing infrastructure at a global scale. has pre-paid for capacity from a chemistry that does not yet exist at grid scale, at a deal size that dwarfs anything most utilities have signed for storage in the same week.

The technology is a reversible solid-oxide fuel cellAn electrochemical conversion device that produces electricity directly from oxidizing a fuel. Reversible versions can run backwards, using electricity to produce stored chemical fuel. rated for 100+ hour discharge — a regime where lithium-ion is structurally unsuitable, because Li-ion degrades when held at 0% or 100% state-of-charge for the stretches multi-day storage requires. Noon's cell stores energy in carbon and oxygen rather than lithium, nickel, or cobalt, which is the part of the press release that matters: a 100 GWh deployment of this chemistry does not have to compete with the supply chain for electric vehicles. The bottleneck moves from mined metals to manufacturing throughput, which is a problem Meta can underwrite directly.

AI data centers have become the marginal buyer for first-of-kind storage chemistries. Form Energy's iron-air system has a 30 GWh Google deal in Minnesota and a 12 GWh contract with Crusoe; Energy Dome's CO2 battery has Google equity and a Texas data-center MOU; Noon now has Meta. Each contract is larger than the entire installed long-duration-storage base that existed two years ago. The hyperscalers are doing what regulated utilities cannot: writing capacity-reservation cheques against unproven technology and treating first-of-kind risk as the cost of decoupling compute siting from grid-firm-power constraints.

Reversible solid-oxide cells stack as carbon-oxygen reservoirs — lithium nowhere in the bill of materials.

×

The winners are the half-dozen LDES vendors whose chemistries were unfundable at utility timelines and are suddenly fundable at hyperscalerA massive cloud service provider, typically Amazon Web Services, Microsoft Azure, or Google Cloud, capable of provisioning computing infrastructure at a global scale. timelines — Noon, Form, Energy Dome, and a long tail of carbonate and metal-air candidates that follow them through the door. The losers are the gas-peaker developers whose business model assumes data centers keep paying premium $/MWh for firmed capacity through the late 2020s, and the turbine OEMs whose order books reflect that assumption. The integrated utilities sit in an awkward middle: hyperscaler-owned storage behind the meter is, technically, demand-side load shaping, which is not the category most rate casesFormal proceedings where regulated utilities justify their investments and operating costs to public utility commissions in order to set the prices they can charge customers. are written against.

What the Meta–Noon contract opens is the question of who actually owns the 2030s grid: the regulated utility selling firmed kilowatt-hours, or the hyperscalerA massive cloud service provider, typically Amazon Web Services, Microsoft Azure, or Google Cloud, capable of provisioning computing infrastructure at a global scale. that has built its own multi-day battery and treats the utility as a transmission backstop. What it forecloses is the assumption — still embedded in most integrated-resource plans — that long-duration storageEnergy storage systems designed to discharge power continuously for 10 or more hours, filling gaps in renewable generation that span multiple days rather than just evening demand peaks. at relevant scale is a 2032-or-later problem. It is being procured now, in gigawatt-hour blocks, by buyers who have already decided the answer.