Diagram of the HALEU fuel cycle showing five production steps with conversion and deconversion marked as the real bottlenecks, not enrichment

How HALEU Is Made and Why Building More Centrifuges Won’t Fix the Shortage

ByElliot Marsh

Every conversation about the high assay low enriched uranium (HALEU) shortage lands in the same place, we need more centrifuges. It’s not wrong. It’s just the middle of the story.

Spend any time doing nuclear fuel cycle planning and you learn that a reactor fleet doesn’t stall on the step everyone photographs, the spinning machines, it stalls on the two quiet steps on either side of them. The centrifuge takes a gas in and puts a gas out. Something has to make that gas and something has to turn the enriched gas back into a solid a reactor can actually burn. Those two steps “conversion and deconversion” are where the U.S. chain is thinnest and neither is fixed by ordering more centrifuges.

This piece walks the whole thing as physical steps, from ore to fuel pellet and shows where it breaks. For what HALEU actually is and why advanced reactors need it, see our explainer on what HALEU is , how it differs from the LEU in today’s reactors and for the enrichment economics , why HALEU costs almost six times more per kilogram see the cost breakdown. Here we’re after a different question, how is HALEU made and which link gives out first.

How HALEU Is Made, Step by Step (the 60-second version)

Short answer: HALEU is made in five physical steps. Mine and mill the ore into yellowcake, convert that into uranium hexafluoride (UF₆) gas, enrich the gas to just under 20% U-235, deconvert it back into a solid oxide or metal, then fabricate that solid into fuel. Enrichment is only step three of five.

Uranium leaves the ground as ore and gets milled into yellowcake (U₃O₈), a khaki colored powder. Powder can’t be enriched, so it’s converted into uranium hexafluoride (UF₆) the one uranium compound that turns into a gas at modest temperatures, which is exactly what a centrifuge needs.

The gas is then enriched, centrifuges spin it to raise the share of the fissile U-235 isotope . Ordinary reactor fuel (LEU) stops at 3–5% U-235; HALEU is defined as more than 5% and less than 20%, typically around 19.75%.

But you still have a gas. No reactor runs on UF₆. So the enriched gas is deconverted into a usable solid an oxide powder, a metal or an alloy and finally fabricated into the actual fuel form, ceramic pellets, tiny coated TRISO particles or metallic slugs, depending on the reactor. Five steps, five different industrial plants. The shortage isn’t evenly spread across them.

Fuel cycle flow: five HALEU production steps with conversion & deconversion flagged as bottlenecks.
Figure 1. The HALEU fuel cycle as physical steps. Everyone argues about step three; the U.S. squeeze is really at steps two and four.

The Conversion Bottleneck Nobody Mentions

Short answer: Before uranium can be enriched, it has to become UF₆ gas and the United States has exactly one commercial conversion plant. You can build all the centrifuges you want, without domestic conversion feed, they have nothing to spin.

The plant is Metropolis Works in southern Illinois, now run by Solstice Advanced Materials (spun off from Honeywell in October 2025, the trading arm, ConverDyn is a joint venture between Solstice and General Atomics).

It is the only commercial UF₆ conversion facility in the country. It sat idle from 2017 to 2023 when the market was oversupplied, restarted in 2023 and in February 2026 announced it expects to produce more than 10,000 tonnes of uranium as UF₆ in 2026 roughly a 20% jump over its planned 2024 output. Its order backlog now tops $2 billion, and its NRC license runs to 2060.

One plant. That’s the whole domestic conversion industry. Globally the picture isn’t much wider, commercial UF₆ conversion is concentrated in a handful of sites. Metropolis in the U.S., Cameco’s Port Hope in Canada, Orano’s Tricastin in France, and larger capacity in Russia and China.

Metropolis alone has historically represented on the order of a fifth of world capacity. For years no Western converter committed to major new build. Orano’s own leadership has described the required conversion and enrichment investments as massive relative to the size of the companies involved.

U.S. commercial-scale plants per fuel-cycle step (2026): one deep at every step.
Figure 2. Step through the U.S. chain and the count barely changes: one plant deep, everywhere. The orange bars are the choke points.

This is why “just build more centrifuges” misses. The 2026 enrichment expansions, Urenco’s New Mexico groundbreaking, Centrus in Ohio, General Matter in Kentucky, all consume converted UF₆ as their feedstock.

Add enrichment capacity without adding conversion, and you’ve widened the middle of the funnel while leaving its neck exactly as tight. Conversion is the upstream constraint that rarely makes the headline, and it’s the first place the chain runs dry.

The 20% Line Changes the Plant, Not Just the Runtime

Enriching to ~19.75% isn’t a matter of running the same cascade longer. Higher assay raises the risk of an accidental chain reaction inside the plant, which forces different cascade geometry, criticality safe equipment, and a stricter security category. It’s a re-engineering job, not a dial you turn up.

Below about 10% enrichment, the equipment geometry that’s safe for ordinary LEU stays safe. Push toward 20% and criticality safety inside the piping, cascades and storage becomes a live design constraint.

Spacing, batch sizes and container geometry all have to change so the material can never inadvertently go critical during processing. (The security and licensing side of that 10% line and the SWU math behind HALEU’s cost. we cover in the cost article, no need to repeat it here.)

The regulatory shorthand: HALEU is classed by the NRC as special nuclear material of moderate strategic significance, which means a HALEU plant is a Category II facility. The LEU that fuels today’s fleet is of low strategic significance Category III.

Moving up a category means more physical security, tighter material accounting and a heavier licensing basis. So even taking an existing LEU line and “turning it up” to HALEU isn’t a software change; it’s a different plant with a different license. That reality shapes every step that touches enriched HALEU including the two that come after enrichment.

Deconversion – the “Middle Step” That’s Suddenly a Race

After enrichment you have HALEU as UF₆ gas. No reactor runs on gas. Deconversion turns that gas back into a solid, oxide, metal or alloy and until 2026 the U.S. had essentially no commercial HALEU deconversion capacity at all.

Deconversion is the mirror image of conversion, instead of powder to gas, it’s enriched gas to solid. The output can be uranium oxide, uranium metal or a metal alloy depending on what the reactor needs.

It sounds like plumbing and it is but it’s plumbing nobody had built at commercial scale for HALEU, which is why in 2026 it became one of the most contested links in the chain.

The Department of Energy has awarded ten-year contracts to six companies to stand up deconversion services. Nuclear Fuel Services, Centrus’ American Centrifuge Operating, Framatome, Global Nuclear Fuel Americas, Orano Federal Services and Westinghouse.

And in March 2026, Oklo and Centrus announced a joint venture specifically to build HALEU deconversion at Centrus’ Piketon, Ohio, site, co-located with enrichment. The industry’s own framing is telling, they call deconversion the key middle step and a likely bottleneck to large scale deployment. When the people building the reactors start racing to lock down deconversion, that’s the tell that this, not the centrifuges, is where the worry is.

Deconversion fork: one HALEU gas into oxide, TRISO, and metallic fuel for three reactor families.
Figure 3. Deconversion is a fork, not a funnel. One enriched gas splits into three incompatible fuel forms which is why no single plant can serve the whole fleet.

And here’s the part that makes deconversion genuinely hard to scale. There is no single “HALEU fuel.” The enriched gas has to become a different solid for each reactor family. That branching is what stops one flexible plant from feeding everyone and it carries straight into the next step.

Fabrication — Three Fuels, Three Different Plants

HALEU doesn’t become one product. Oxide pellets for light water SMRs, TRISO particles for gas reactors, metallic alloy for fast reactors. Three fundamentally different fabrication routes, each with very few (or zero) licensed facilities in the U.S.

For the high temperature gas cooled reactors (HTGRs), fuel is TRISO poppy seed sized uranium kernels wrapped in layers of carbon and ceramic. In February 2026, X-energy’s subsidiary TRISO-X received the first-ever NRC Part 70 license to fabricate HALEU fuel (license SNM-7007)

The first new fuel fabrication license the NRC has issued in roughly half a century and the country’s first Category II fuel facility. Its TX-1 plant, under construction in Oak Ridge, Tennessee, is designed to turn out about 5 tonnes of uranium (some 700,000 TRISO pebbles) a year enough to fuel up to eleven of X-energy’s Xe-100 reactors.

Two other players are in the TRISO race: Standard Nuclear, which became the first company to physically receive HALEU feedstock for TRISO production (destined for a Radiant microreactor demonstration) and BWXT.

For the sodium fast reactors like Natrium, the fuel is metallic a uranium zirconium alloy (U-10Zr) in steel cladding, which needs metallization, a deconversion route that ends in metal rather than oxide.

Framatome and TerraPower built a metallization pilot line in Richland, Washington and in November 2025 produced their first metallic uranium “pucks”. The production fuel facility for Natrium is being built next to Global Nuclear Fuel Americas’ existing plant in Wilmington, North Carolina. Different chemistry, different plant, different license from the TRISO line even though both start from the same enriched gas.

2026 U.S. HALEU build-out by step: converter, enrichers, deconversion contract holders, fabricators.
Figure 4. The 2026 build-out, step by step. Enrichment finally has four names on the board; conversion still has one and commercial HALEU deconversion has none operating yet.

The pipeline is still filling. As recently as August 2026, a company called FANCO notified the NRC of its intent to build a facility to process HALEU UF₆ into other fuel forms, deconversion plus fabrication with a license application not expected until late 2027. That’s the shape of the problem. Even the projects meant to fix the fabrication gap are years from a license, let alone production.

The Stopgap – Downblending HEU (and Why It Runs Out)

Short answer: While the new plants get built, the U.S. bridges the gap by downblending surplus high enriched uranium (HEU) into HALEU. It works, but it’s finite. Most HEU is reserved for the military and the civilian slice available for blending is small.

Downblending is exactly what it sounds like, take weapons grade or research grade HEU (20% U-235 or higher) and mix it with natural or low enriched uranium until it lands in the HALEU band.

Historically that’s how DOE produced HALEU for its own stockpiles and it’s the fastest way to put fuel in the first advanced reactor cores. Five developers – Kairos Power, Radiant, Westinghouse, TerraPower, and TRISO-X have already been lined up for early HALEU from government reserves.

The catch is arithmetic. Government stockpiles were expected to reach roughly 21 metric tons of HALEU equivalent by mid 2026, of which only a portion is actually usable as supply and a recent Savannah River decision clears a further ~3.1 tonnes of HALEU from about 2.2 tonnes of HEU over two to four years.

Those are small numbers against a fleet’s appetite. The deeper limit, the bulk of U.S. HEU is defense obligated committed to naval reactors, weapons, research reactors and medical isotope production and simply isn’t available for commercial downblending. Downblending is a bridge and a short one.

The finite downblending bridge: DOE stockpile vs usable vs Savannah River tonnage.
Figure 5. The downblending bridge is real and small. For how these tonnages stack up against projected fleet demand, see the supply gap section of the cost article.

Why It’s Still Slow Even With $2.7 Billion on the Table

Short answer: Money finally showed up in 2026. The chain is still slow because licensing takes years, the demand signal is shaky and much of the new capacity lands after the deadline it was meant to beat.

The funding is real. In January 2026 the DOE awarded $2.7 billion across three enrichment task orders $900 million each to Centrus’ American Centrifuge Operating (HALEU, Piketon), General Matter (HALEU, at the former Paducah site in Kentucky) and Orano Federal Services (LEU) plus a smaller award to Global Laser Enrichment.

Centrus signed its contract on June 30, 2026, committing to deliver one metric ton of HALEU as UF₆ by March 2032. On August 18, 2026, Urenco USA broke ground in New Mexico on a roughly 50% capacity expansion, with the first new cascades slated for 2032. The DOE has even floated “Nuclear Lifecycle Innovation Campuses” inviting states to host full fuel cycle sites end to end. So why isn’t the shortage over?

Three reasons. First, licensing time.TRISO-X filed its application in April 2022 and wasn’t licensed until February 2026 close to four years for a single fabrication plant and that’s a success story.

Second, the chicken and egg demand signal. Fuel cycle companies won’t sink billions into conversion or deconversion without firm orders and reactor developers can’t place firm orders until they’re sure fuel exists. Each side waits for the other.

Third, timing.The Prohibiting Russian Uranium Imports Act phases out the last import waivers on January 1, 2028 yet most of the new domestic capacity comes online in 2029-2034.

Capacity timeline 2023–2037 vs the January 2028 Russian import ban.
Figure 6. The awkward truth in one chart: the deadline is 2028, and most of the new domestic capacity arrives well after it.

And none of this counts the logistics between steps. Moving enriched HALEU between conversion, enrichment, deconversion and fabrication sites is its own criticality problem, the shipping packages themselves have to be licensed and criticality safe, a constraint we covered separately. Every hand-off in this chain has a licensing tail.

The Bottom Line

Centrifuges are the part of the HALEU story that photographs well, so they get the attention and, lately, the money. But a supply chain is only as fast as its slowest licensed step and in the U.S. that step keeps turning out to be one of the unglamorous ones on either side of enrichment.

Conversion is a single plant feeding the entire front end. Deconversion, until this year, barely existed at commercial scale and now has six companies scrambling to build it. If you want an early read on whether the advanced reactor buildout is on track, don’t watch the centrifuge announcements. Watch whether a second U.S. converter breaks ground and whether the first commercial HALEU deconversion line actually starts up. Those two boring links are the real schedule.

Frequently Asked Questions

How is HALEU made?

HALEU is made in five steps. Uranium ore is mined and milled into yellowcake (U₃O₈); the yellowcake is converted into uranium hexafluoride (UF₆) gas; the gas is enriched in centrifuges to between 5% and 20% U-235 (usually ~19.75%); the enriched gas is deconverted back into a solid oxide or metal and that solid is fabricated into fuel. In the U.S. today, limited HALEU is also produced by downblending surplus high enriched uranium.

What is uranium conversion, and why is it a bottleneck?

Conversion turns solid uranium concentrate (yellowcake) into UF₆ gas, the only practical feedstock for centrifuge enrichment. It’s a bottleneck because the United States has just one commercial conversion plant, Solstice’s Metropolis Works in Illinois. Expanding enrichment without expanding conversion simply moves the shortage upstream.

What is deconversion?

Deconversion is the reverse of conversion: it turns enriched HALEU UF₆ gas back into a usable solid like uranium oxide, metal or alloy that can be fabricated into fuel. No reactor runs on gas, so deconversion is an unavoidable step. Commercial HALEU deconversion capacity in the U.S. is only now being built, under DOE contracts awarded to six companies.

Why can’t the U.S. make enough HALEU yet?

Because every link is thin at once: one commercial converter, one commercial enrichment plant (plus a HALEU demonstration cascade), essentially no commercial HALEU deconversion until 2026 and only a few licensed fabrication facilities. Licensing each new plant takes years and a “chicken and egg” problem, no fuel without firm orders, no firm orders without fuel, slowed private investment until federal funding arrived in 2026.

What is HALEU downblending?

Downblending mixes high enriched uranium (20%+ U-235) with natural or low enriched uranium to bring it down into the HALEU range. It’s the fastest way to supply the first advanced reactor cores, but it’s a finite bridge. Most U.S. HEU is reserved for defense uses and isn’t available for commercial blending.

Is building more centrifuges enough to fix the HALEU shortage?

No. Centrifuges only enrich a gas that conversion has to supply and that deconversion has to turn back into a solid. Without matching conversion feed upstream and deconversion plus fabrication downstream, extra enrichment capacity just widens the middle of a funnel whose neck stays tight.

Sources and Further Reading

All figures and claims above are drawn from primary and industry sources. Key references:

About The Author

Elliot Marsh is a working reactor engineer with hands on experience in reactor physics, core management, reactivity control and nuclear fuel cycle planning at an operating power station. He writes about nuclear energy for readers who want the engineering reality, not the press release.

Similar Posts