Porritt Inc. Energy Desk | May 27, 2026 | Reading time ~7 minutes
Cold open
In the first five months of 2026, the U.S. advanced-nuclear thesis stopped being a thesis. Oklo got a Principal Design Criteria approval from the NRC, signed a 1.2 GW prepayment with Meta in Pike County, Ohio, and is moving on a 12 GW Master Power Agreement with Switch that was inked at the end of 2024. The Department of Energy then ratified the supply side: $94 million in Tier 2 cost-share awards across eight U.S. companies covering early site permits, reactor pressure vessel finishing, large-forging capacity, and uranium fuel fabrication. Demand and supply were each moving alone for two years. In May 2026 they finally rhyme.
This post calls the trade what it is: the Aurora Compact — a privately funded demand stack of roughly 13.2 GW underwriting one sodium-cooled fast reactor product line, plus a federally cost-shared supply lattice that solves the four things money cannot move on a calendar (siting, forging, RPV finishing, HALEU pellet capacity). The two halves were always the same trade. Until this quarter, only one half existed at a time.
The demand stack — 13.2 GW now under signature
- Switch, Ltd. — 12 GW Master Power Agreement. Signed December 2024. One of the largest corporate PPAs in the history of any power source. Tranched delivery across Aurora powerhouses sited near Switch’s hyperscale campuses.
- Meta Platforms — 1.2 GW Prepayment Agreement. Signed January 5, 2026. Meta prepays for power on a 206-acre Pike County, Ohio site that Oklo bought from DOE. Phase 1 targets 2030 first power, scaling to the full 1.2 GW by 2034. The prepayment funds nuclear fuel procurement and front-loads schedule risk.
- NRC Principal Design Criteria approval (April 2026). The first private advanced fission plant to clear that pre-application bar. It is not a Combined License Application, but it is the regulator validating that the safety case is the safety case Oklo says it is.
Add the two PPAs and you get 13.2 GW of named, tranched, contractually wired demand for a single reactor line that has yet to produce a watt. For context, the entire U.S. light-water SMR cohort — Holtec at Palisades, NuScale’s NRC-certified VOYGR, the Generation III+ designs about to receive DOE Tier 2 money — does not yet have a privately committed demand stack of that size. Oklo, at 15-to-75 MWe per unit, will need somewhere between 175 and 880 separate Aurora powerhouses to meet 13.2 GW. That math is the whole story: the bottleneck is not the order book. It is the fabrication line.
The supply lattice — DOE’s $94M is the un-glamorous half
On May 15, 2026, DOE announced eight Tier 2 awards under the Generation III+ SMR Program. None of them are reactors. All of them are the things that have to exist before reactors get poured.
| Company | Award | Function |
|---|---|---|
| Nebraska Public Power District | $27.86M | NRC-approved Early Site Permit, Nebraska |
| BWXT Nuclear Energy | $21.42M | RPV final assembly equipment, Mount Vernon, IN |
| Constellation SMR Development | $17.26M | NRC-approved Early Site Permit, New York |
| Scot Forge | $12.27M | Large vertical turning lathe + gantry mill, Spring Grove, IL |
| Framatome U.S. Government Solutions | $8.80M | +200 MTU/yr ceramic pellet capacity, Richland, WA |
| Global Nuclear Fuel Americas | $3.00M | Second BWR fuel rod fabrication line |
| American Forgemasters | $2.90M | New large-forging furnace, New Castle, PA |
| Container Technologies Industries | $0.55M | Nuclear quality assurance for SMR steel, Helenwood, TN |
Read that table the way an EPC reads it. The U.S. lost the heavy-forging supply chain to South Korea and Japan two decades ago. You cannot build an RPV without an open-die forge press, a vertical turning lathe of the right swing diameter, and a heat-treatment furnace big enough to normalize the part. Three of the eight Tier 2 awards — Scot Forge, American Forgemasters, BWXT — go directly to those three pieces of equipment. The Framatome award expands ceramic UO2 pellet capacity by 200 metric tonnes of uranium per year, which is exactly the chokepoint that has dragged AP1000 reload schedules into 2030. The two Early Site Permits cut roughly 24 months out of the front-end NEPA process for two Gen III+ host utilities. Container Technologies is the boring one — they get $548k to certify their welded steel containers for nuclear quality assurance — but it is also the bottleneck that delayed Vogtle 3 by months in 2022.
The DOE Tier 2 awards are not headline numbers. They are the foundry, the lathe, the furnace, the pellet press, and the early site letter. They are precisely the things 13.2 GW of demand cannot survive without.
Why the two halves matter together — and why prior cycles failed
The last three U.S. nuclear demand cycles failed for the same structural reason: the demand side and the supply side never overlapped in time. South Texas 3 & 4 had utility demand but no forging supply chain by 2011 and was canceled by 2018. The Carbon Free Power Project at INL had federal grant supply but the utility off-take collapsed in 2023 and the project shut down. Vogtle 3 & 4 finished, but the cost overrun was almost entirely supply-chain attrition that started in 2011 and never recovered until the units were nearly mechanically complete in 2023.
The Aurora Compact is the first cycle in which a privately financed demand stack and a federally cost-shared supply lattice are landing inside the same fiscal year. The signal we are watching is not whether Oklo’s first powerhouse at Idaho National Laboratory goes critical by the DOE Reactor Pilot Program’s 4 July 2026 target. It is whether the BWXT Mount Vernon line is finishing RPV shells by Q4 2027, the Framatome Richland pellets are flowing by Q2 2028, and Scot Forge has the second large gantry mill cutting chips by the end of 2027.
If those four data points fire on time, the next ten years of U.S. fast-reactor deployment do not look like Vogtle. They look like Boeing 737 line manufacturing — repeatable, paced, and bound by the slowest fabrication station rather than by NRC docket length.
The Aurora architecture in one paragraph
Aurora is a sodium-cooled fast neutron reactor in the 15-75 MWe range. Metallic fuel, HALEU enrichment (less than 20% U-235), heat pipes carrying core heat to a supercritical CO2 Brayton cycle for power conversion. The fuel is the elegant part: DOE granted Oklo 5 tonnes of HALEU for the first INL core, downblended at INL via electrorefining from used fuel material. The reactor is not just consuming HALEU — it is, in principle, the front end of a fuel-recycle loop that no other U.S. advanced design has yet operationalized at commercial scale. The supercritical CO2 cycle improves thermal-to-electric efficiency by 3-5 percentage points over a steam Rankine cycle in the relevant temperature range. The module size targets the 50-75 MWe envelope that fits a hyperscaler campus block, a remote industrial site, or — and this is where it matters for industrial buyers — a modular refining or chemical processing complex looking for stable process heat below the cost of natural gas.
What this means for distributed industrial users
Refiners, chemical operators, and remote process-heat consumers have spent two decades hedging on natural gas as the only honest base-load heat source. Aurora-class microreactors at $80-110/MWh delivered power (Oklo’s modeled range) and an outlet temperature usable for low-grade process heat change the unit economics of any industrial facility under about 250 MWe of total thermal demand. That is the entire modular-refinery footprint, the entire mid-Atlantic green-hydrogen electrolyzer footprint, and the front-end heat duty for most distillation columns operating below 500°F. The Aurora Compact is not just hyperscaler infrastructure. It is the first cost-credible alternative to a natural gas combined-cycle utility connection for distributed industrial customers.
Risk register — what would break the trade
Three things will determine whether the Aurora Compact survives the next 36 months. First, NEPA — accelerated permitting paths cut paper, but environmental challenges to a sodium-cooled fast reactor will be the first major lawful test. Second, HALEU rate — DOE’s downblending capacity at INL is bounded; commercial HALEU production via Centrus and proposed enrichment expansions has to ramp before the demand stack pulls more than the DOE pilot supply can deliver. Third, the fabrication ramp at BWXT Mount Vernon, Scot Forge, and Framatome Richland. If any one of those three stalls by more than 18 months, the 2030 Pike County Phase 1 date slips and the Switch tranche timing follows it.
None of those three risks is fatal. All three are observable in public data. We will keep a quarterly tracker.
Dual-use angle
The same modular reactor that powers a Meta inference campus heats a 10,000-bpd modular distillate refinery, a remote ammonia plant, or a Tier-1 DoD forward operating base. Porritt Inc. is the industrial-heat tenant the Aurora Compact has not yet named. We build the AI-native refining, compliance, and process design stack that turns a 75 MWe microreactor outlet into a federally bankable industrial facility.
Talk to us. If you are a hyperscaler nuclear team, a Tier 2 awardee, a DOE Loan Programs Office partner, or a federal program officer working the Reactor Pilot Program, Porritt Inc. is taking introductions for industrial co-tenant teaming. Contact: porrittinc.com/contact.
About Porritt Inc. Porritt Inc. is a Salt Lake City–based industrial AI and modular refining company. UEI WQWHVJ1J98S9 | CAGE 9Z7B9. Active in DOE, USDA, NSF, and ARPA-E funding cycles.