VALAR WATCHTake action

Operations

What is physically happening: construction, fuel, testing, power levels, staffing, deliveries.

recordconfirmed Valar's public safety agreement still lists 4.95% fuel, unchanged in every copy from Nov. 2025 to Sept. 2026
2026-09 · dir-009 · Valar Atomics, U.S. Department of Energy

Valar publishes one version of the Ward 250 Nuclear Safety Design Agreement (No. 100403, Revision 02, dated October 2025). The Internet Archive holds seven captures of that PDF (7 November 2025; 1 January, 3 March, 5 July, 6 July, 6 August and 18 September 2026); with the live file as of late September 2026, all eight are the same file, and each carries the fuel line 'Fuel Enrichment 4.95% U-235 LEU limits'. So the public copy was not revised after DOE's packaging review approved shipping HALEU fuel to Ward 250 (20 May 2026, safe-010) or after the reactor's first criticality (18 June 2026, orig-010). The agreement itself says that if the fuel differs from its specification, the safety assumptions will be updated in the later Preliminary Documented Safety Analysis, of which no public copy was found (safe-007); an updated analysis may exist. The companion Quality Assurance Program Description (No. 100474, Revision 02) is likewise unchanged across its six copies.

In plain termsThe only safety agreement the public can read still says the reactor's fuel is 4.95% enriched. It has not been changed since at least November 2025, even after DOE approved shipping richer HALEU fuel to the reactor and the reactor started up. The company may have updated its fuel numbers in a later safety analysis, but no public copy of that document was found in the records read.
verified 2026-10-02: Every archived copy of the PDF has the same text (Internet Archive captures from 7 November 2025 to 18 September 2026, plus the live file); the 4.95% line is in each. Copies are kept; the PDF files themselves are also identical (rechecked 2 Oct 2026). QAPD: 6 copies, identical text. Revised before publication (late September 2026): the archive index lists a 6 July 2026 capture the first count missed; seven captures plus the live file, all identical. Revised before publication (2026-10-01): wording made more exact against the cited records.
recordconfirmed Valar's two public safety papers were never listed in its own website library in any copy since they went up
2026-09 · dir-010 · Valar Atomics

Valar posted its Ward 250 safety agreement (NSDA) and Quality Assurance Program Description on its website on 13 October 2025, each on its own page linking the PDF. Valar Watch compared every saved copy of Valar's library page (valaratomics.com/docs): Internet Archive captures of 16 October and 7 and 19 November 2025, and 5 July, 6 August and 25 August 2026, and the live page in late September 2026. In none is either safety-paper page among the library's listed posts; the library lists the NRC-lawsuit post, the Project NOVA release and Valar's other announcements. So since they went up, the only safety papers Valar publishes have been reachable only by a direct link (for example from the Ward 250 page's data or from outside reporting), not by browsing Valar's library. The record does not show why they were left unlisted.

In plain termsThe two safety papers Valar has made public were not listed in any saved copy of its website library from October 2025 to September 2026. They are online, but a visitor browsing Valar's library would not find them; you need the exact link.
verified 2026-10-02: Saved copies of /docs (10 versions, 2025-04-08 to late September 2026): the links to listed posts were read from each copy; 'Ward250-Nuclear-Safety-Design-Agreement' and the QA post never among them. Captures before October 2025 predate the posts.
recordconfirmed Incidents: none found in public records read, but DOE keeps occurrence reports closed to the public department-wide, 'for security purposes'
2026-09 · safe-033 · U.S. Department of Energy, Valar Atomics

Valar's NSDA commits Ward 250 to DOE's occurrence reporting order (NE O 232.1), accident investigations (NE O 225.1), ES&H reporting, independent oversight, and DOE's employee concerns and differing professional opinion programs (which DOE-STD-1271 says allow no alternative). DOE's Occurrence Reporting and Processing System page says that, 'for security purposes', access to occurrence reports is restricted to approved DOE and DOE contractor personnel, a restriction covering every report in the system, not one set for Ward 250; that page describes the department-wide order (DOE O 232.2A), while the NSDA commits to the Office of Nuclear Energy's own order, and whether Ward 250's reports go into the same system is not stated in any record read. No public report of an incident, occurrence, violation, enforcement action or worker injury at Ward 250, NOVA or the fuel line was found in DOE, state or news sources read as of late September 2026. OSHA's inspection database was not searched. Absence of a public record is not evidence that nothing happened; DOE's occurrence reports for Ward 250 can be requested under FOIA. Valar's quality program (p.20), for reporting confirmed suspect or counterfeit items, cites a different occurrence-reporting order, NE O 232.2. The NE orders were not found in public records, so which is current could not be checked.

In plain termsIf something goes wrong at a DOE-authorized reactor, the operator must report it to DOE in a formal 'occurrence report'. DOE's occurrence database is open only to approved DOE and contractor staff, for every facility in it (DOE cites security), and it is not stated whether Ward 250's reports go there. We found no public report of any incident at Ward 250, but because the database is closed, that tells us little. A FOIA request is the way to find out.
verified 2026-10-02: Re-read NSDA p. 45-46, the ORPS page and DOE-STD-1271 Attachment 1 in late September 2026. Holds. Fixed: the ORPS page covers DOE O 232.2A; the NSDA's NE O 232.1 reports are not shown to go into ORPS, so the plain text no longer asserts it. No public incident report was found. Corrected in late September 2026: added that the quality program cites NE O 232.2.
official statementreported As reported, Carbon commissioners called Ward 250 a 'commercial reactor' and the planned data center waterless; Valar calls Ward 250 a research reactor and has called the data center both waterless and low-water
2026-07-01 · loc-005 · Jared Haddock, Tony Martines, Carbon County Commission, Valar Atomics, NVIDIA

ETV News (3 July 2026) reported the Carbon County Commission meeting of Wednesday 1 July 2026 (ETV's text gives 'June 2' and 'June 1', but its publication date, the fire restrictions ahead of the Fourth of July and the NVIDIA announcement place it on 1 July; the commission's 1 July minutes record the same celebration (csite-026) and record Commissioner Haddock saying Valar had announced a data center partnership, one 'that will run without any water'). As ETV reported it, Commissioner Jared Haddock described Ward 250 as having recently become 'the first government-approved commercial reactor' to achieve a self-sustaining reaction outside a federal lab, and Commissioner Tony Martines said that because Valar's reactor is gas-cooled and NVIDIA's chips are air-cooled, the planned data center won't need water. Set beside the record: Valar's own February 2026 release says Ward 250 is a research reactor and 'not a commercial power plant'; DOE's pilot is for research and 'not' to demonstrate commercial suitability; DOE's 'first DOE authorized reactor built outside of a national laboratory' holds (orig-010). On water, Valar's account as ETV reported it was that the data center would use a closed-loop cooling system 'designed for low water consumption' and that it already had the water it needs, while seeking up to 1,000 more shares for other projects (ETV, 28 July). The part about being first outside a federal lab holds; 'commercial' goes beyond what Valar and DOE say. The commissioners' no-water description matches Valar's own 4 August 2026 release, which calls the planned data center 'a waterless 30 MW AI factory'.

In plain termsTwo Carbon County commissioners, speaking at a public meeting as the local paper reported it, described Valar's test reactor as a 'commercial' reactor and its planned data center as needing no water. The officials' description, as reported, differs from Valar's own on the reactor: Valar calls Ward 250 a research reactor that is not a commercial power plant. On water, Valar has described the data center both as waterless (in its own August 2026 announcement) and as designed for low water use (as the local paper reported in late July).
verified 2026-10-02: Re-read ETV 3 July, the Feb 2026 Valar release and ETV 28 July: holds as ETV's paraphrase. Plain field now says 'as reported'. Revised before publication (late September 2026): the comparison is restated as a plain difference.
recordconfirmed DOE shipping review: fuel for Ward 250 is HALEU (under 20%), not the 4.95% its safety agreement lists
2026-05-20 · safe-010 · U.S. Department of Energy, Valar Atomics, National Criticality Experiments Research Center

A DOE Safety Evaluation Report, prepared by the manager of DOE's Packaging Certification Program and approved by DOE's Headquarters Certifying Official on 20 May 2026, reviews and recommends an amendment to the certificate of the Model 9979 shipping package to allow 'limited shipments' of high-assay low-enriched uranium (HALEU) TRISO fuel compacts from the National Criticality Experiments Research Center (NCERC) in Nevada to the Valar Ward 250 reactor site in Utah. Each package holds one hexagonal graphite 'WARD250 FUEL BLOCK' (Valar drawing V-002782) with 170 compacts; the U-235 in a block is 234.94 grams 'at less than 20% enrichment', the application describes the uranium in the particles as 'enriched to a maximum of 19.9 wt.% U-235', and the stated condition is 'less than 20 weight percent'. The report says the block's end use is as a fuel element in the Ward 250 microreactor. Valar's NSDA (October 2025, Table 6) lists Ward 250's fuel enrichment as '4.95% U-235' on the basis of 'LEU limits', and says that if the fuel differs from the specification its safety assumptions will be updated in the PDSA. The later PDSA and DSA were not found in public records, so whether DOE's approved safety basis was updated to HALEU could not be checked. This is the DOE record that disc-001 lacked: it matches the NOVA release's 'HALEU TRISO' wording (orig-008) and a Valar speaker's statement that part of the core went critical in Nevada, and conflicts with the NSDA's 4.95% figure. Per element, the NSDA's own figures imply 162.83 g of U-235 in 3.29 kg of heavy metal; DOE's block holds 234.94 g of U-235, about 1.18 kg of uranium at the 19.9% maximum enrichment. No single enrichment fits both (76 such blocks holding 250 kg would need 7.14%), so Table 6's equal per-element loading does not match the block DOE cleared, unless some of the 76 elements are unfueled or loaded differently, which the paper does not say.

In plain termsUranium fuel is graded by the share of U-235, the kind that splits. Ordinary power-plant fuel is below 5%; HALEU runs from 5% to just under 20%. Valar's public safety agreement says Ward 250 burns 4.95% fuel. DOE's own shipping review says the fuel approved for shipment to Ward 250 from Nevada is HALEU, enriched up to 19.9%, which is up to about four times richer. The fuel is the same kind of tiny coated bead (TRISO); what differs is how much of the splitting kind of uranium is in each bead. Richer fuel changes the reactor physics, and the safety agreement says fuel-design changes will be carried into the later safety analysis. The only public safety agreement describes different fuel from the fuel DOE reviewed for shipment; the safety analysis DOE approved, which may have been updated, was not found in public records.
verified 2026-10-02: Re-read the SER (all 9 pages) and NSDA Table 6 and pp. 11, 26 in late September 2026. Conflict holds. Fixed: the SER reviews and recommends the letter amendment (the amendment letter itself was not read); added the 19.9% maximum the SER gives; plain now says 'approved for shipment' and 'up to about four times', not 'sent' and 'roughly four times'. Corrected in late September 2026: added the per-element U-235 and heavy-metal comparison with DOE's block. Corrected 2026-09-30: reworded the plain-language text.
recordconfirmed On paper Ward 250 is temporary: 12 months of running, then cooldown and teardown; its two papers differ on length
2026-04-22 · plan-003 · Valar Atomics, U.S. Department of Energy

Valar's Nuclear Safety Design Agreement (October 2025) describes a lifecycle of about 18 months from authorization through decommissioning: a pre-operational phase, 12 months of nuclear operations in test campaigns, then defueling and decommissioning that returns the site to unrestricted use. DOE's environmental determination (22 April 2026) describes a 24-month research program: about 12 months of reactor operation, 6 months of in-vessel cooldown and 6 months of decommissioning, after which the single core of spent fuel goes in dry casks to an unnamed 'DOE-approved facility'. DOE's agreement with Valar runs from October 2025 to 4 July 2027. If operations are counted from criticality in June 2026, the 12 months end around June 2027, and DOE's schedule puts the end of decommissioning around mid-2028; no record read gives the actual planned end date. Set beside this, Valar now describes the Utah site as home, not just a test site: its head of projects told the state trust-lands board the lab is 'no longer just a test facility', it is buying the neighboring 112-acre parcel (gov-022), it was approved for a state tax credit that requires it to stay 10 years, and its Castle Country page says 'we're ready to build more'. At the 1 July 2026 event Taylor said the fuel inside Ward 250 would let it keep running for 165 years (KSL/Deseret, 6 July); that describes the fuel's capacity at very low power, not the approved plan, which is about 12 months of operation with no refueling (DOE). The two are not necessarily in conflict (the test reactor can be removed while the company stays), but no document found in public records as of late September 2026 says what replaces Ward 250 or whether Ward 250 will actually be torn down on the stated schedule.

In plain termsThe official paperwork treats Ward 250 like a science experiment with an end date: run it for about a year, let it cool, take it apart, ship the used fuel away and leave clean ground. Valar's own safety document says the whole thing takes about 18 months; DOE's environmental paper says 24. Meanwhile the company tells local officials it is here to stay, and at a public event its CEO said the fuel could last 165 years; the approved plan is about one year of running. Both can be true, but no plan for what comes after the test reactor, and no firm date for taking it down, was found in public records as of late September 2026. The safety agreement's 18 months does not fit its own schedule: from the end of its review, the Table 7 pre-operational period plus 12 months of operation already runs about 19.75 months, before any decommissioning, unless the 18 months is counted from startup. DOE's 24 months fits that startup reading.
verified 2026-10-02: Checked in late September 2026: NSDA p. 8 (18 months; 12 months of operations) and DOE CX (24 months: 12+6+6; spent fuel to a 'DOE-approved facility') re-read and hold. 'Took' a tax credit changed to 'was approved for' (the record shows board approval). Taylor's 165-year remark added from KSL. Corrected in late September 2026: the plain-language text adds that the NSDA's 18 months does not fit its own schedule.
recordconfirmed Utility district minutes: Valar 'tied into the county sewer system without prior authorization'; state permit needed
2026-03-19 · loc-015 · Castle Valley Special Service District, Valar Atomics, Utah Division of Water Quality, Jess Housekeeper, Willem Kriel

Castle Valley Special Service District (CVSSD) runs water and sewer for Castle Valley towns including Orangeville. Its approved minutes of 19 February 2026 record Valar's Jess Housekeeper presenting Valar's plans (by video), the board approving a $20,000 outside-district water connection for Valar contingent on all water shares being submitted to Orangeville, and the manager reporting that Valar 'will be required to apply to' the state Division of Water Quality (DWQ) for a sewer discharge permit, which DWQ would administer for four years, and that no sewer connection would be approved until DWQ approved the permit. The approved minutes of 19 March 2026 (whose opening line misdates the meeting to 19 February) record the manager's update: sewer plans 'were not submitted to the State as required', water service work was paused pending DWQ approval, 'Valar tied into the county sewer system without prior authorization', Valar was asking for ordinary sanitary sewer service for restrooms only with no process water, a site visit with the regional DEQ engineer was planned, and water service would not be provided until all sewer permits and approvals were in place; no action was taken. The district's May, July and August 2026 minutes (read; the June minutes, approved on 16 July, were not found on the state public notice website as of 30 September 2026) do not return to the sewer issue, though the May minutes discuss annexing property near Orangeville 'that may be developed by Valor', and its 17 September agenda lists no Valar item. DOE's categorical exclusion puts water use under 1,000 gallons a day for domestic use, with no process-water discharges, which matches the restrooms-only request. As of 30 September 2026, no record read shows whether DWQ issued a permit, or how the tie-in the minutes report was resolved.

In plain termsThe local water-and-sewer district recorded in March 2026 a staff report that Valar had 'tied into the county sewer system without prior authorization', and had not sent its sewer plans to the state as required. The request itself was ordinary, toilets and sinks only, not reactor water. It matters less for danger than for process: the district's minutes are the only record of it found, and no Valar response or later resolution appears in the records read. Whether the state permit was later issued was not found in the records read as of 30 September 2026; a GRAMA request to the Division of Water Quality and to the district would show it.
verified 2026-10-02: Re-read the Feb and Mar minutes (text) and the May and July minutes (scanned): holds. Corrected 'June minutes read' (June's are not posted). Plain-language text: 'habit' implied a pattern from one record, now 'process'. Revised before publication (late September 2026): the plain-language text attributes the tie-in to the staff report and drops a conclusion. Revised before publication (2026-09-30): reworded the detail.
analysisconfirmed Ward 250's safety papers lean on Ward Zero tests and reuse its hardware; no Ward Zero test data was found in public records as of late September 2026
2025-10 · orig-059 · Valar Atomics, U.S. Department of Energy, Ward Zero, Ward 250

Valar's NSDA (October 2025) says Ward 250 directly uses hardware that has undergone full-scale testing, that design work focuses on nuclear-specific modifications to the proven WardZero configuration, and it credits Ward Zero testing for key safety points: natural circulation during a loss of coolant flow, and the capacity of the Reactor Cavity Cooling System. It also asks DOE to accept the final safety analysis at 90% design completion because the WardZero program has completed design, installation and testing of major safety-significant systems. Valar's QA Program Description, which Valar's October 13, 2025 post says DOE approved, has a section on using existing WardZero hardware: Ward Zero components may go into Ward 250 if they stay traceable and are evaluated, and Ward Zero operating data may count as design verification by comparison to 'similar proven systems'. Ward Zero was completed in February 2025, before that QA program was approved. Valar's June 2026 draft says the reactor underwent non-nuclear testing at its Hawthorne headquarters above 750 C and 6 MPa before shipment. No Ward Zero test report, data set or DOE review of it was found in public records as of late September 2026, and the page titled 'Ward Zero Whitepaper' holds no Ward Zero data (orig-058). Whether DOE examined the Ward Zero data would be shown by the PDSA and DSA, of which no public copy was found (safe-007).

In plain termsWard 250 was not built from scratch in Utah. By Valar's own safety papers it reuses parts of Ward Zero, the electric-heated prototype tested in California, and Ward Zero's tests are offered as proof that key safety features work, such as the reactor cooling itself by natural air flow. Some of that hardware was built before the quality program was approved by DOE, as Valar says it was; the program allows that if the parts are traced and checked. No Ward Zero test results were found in public records, so the public cannot check the evidence this part of the safety case rests on.
verified 2026-10-02: Second check, late September 2026: NSDA (full-scale tested hardware; proven WardZero configuration; natural circulation and RCCS credited to WardZero; DSA at 90% design because WardZero tested major safety systems), QA plan (WardZero hardware section; 'similar proven systems'), QA post and draft page re-read. Match; approval wording attributed to Valar's post. Revised before publication (late September 2026): whitepaper line now 'holds no Ward Zero data'; source title parenthetical now 'not listed in Valar's library'. Revised before publication (2026-09-30): reworded the plain-language text.
recordconfirmed The accident analysis assumed 3 megawatt-days of fuel use; the same document's limit allows 30, ten times more
2025-10 · safe-009 · Valar Atomics, U.S. Department of Energy

In the NSDA's Maximum Hypothetical Accident analysis (pages 34-36), the radioactive inventory is based on 'OpenMC neutronic analysis with 3 MWd total burnup' and 'Conservative 30 EFPD operation assumed for inventory buildup' (30 effective full-power days at 100 kW is 3 megawatt-days). Earlier (page 10) the NSDA says the analysis assumes operation at maximum licensed power long enough to reach equilibrium inventories. Its operating limits (page 25) set 'Fuel burnup limit: 30 megawatt-days', and its objectives call for 12 months of operation with availability above 80% and steady operation up to 250 kWth (page 8). The limit is ten times the burnup the accident inventory was built on. Valar's July 2025 county presentation (orig-045) and its June 2026 draft post both describe 30 effective full-power days, which matches the accident analysis rather than the 30 MWd limit. The lifecycle also differs: the NSDA gives about 18 months 'from authorization through decommissioning', DOE's categorical exclusion a 24-month program (12 months of operation, 6 of cooldown, 6 of decommissioning). No public copy of the approved DSA, which would fix the inventory and limits, was found as of late September 2026 (safe-007). Our check using standard half-lives: going from 3 to 30 MWd at the same power would raise iodine-131 by about 8% and the iodine and noble-gas dose by about 4%; cesium-137 and strontium-90 would grow about tenfold but stay about 4% of the iodine-131 activity. The NSDA's 18 months is also shorter than its own schedule: Table 7's dates plus 12 months of operation already exceed 18 months before any decommissioning, unless counted from startup.

In plain termsThe longer and harder a reactor runs, the more radioactive by-products build up in its fuel, and an accident analysis has to assume the worst amount. Valar's worst-case analysis assumed about 30 days at full power. The same document's own limit would allow roughly ten times as much fuel use, and its goals speak of a year of operation. The company's later statements point to 30 days, which would fit the analysis. No public record found as of late September 2026 shows which limit DOE approved.
verified 2026-10-02: Re-read NSDA pp. 8-10, 25 and 34-37 and CX in late September 2026. Holds: '3 MWd total burnup' and '30 EFPD' (p. 34) against a 30 MWd limit (p. 25); I-131 named 'primary dose contributor'. Arithmetic checked (30 days x 0.1 MW = 3 MWd). Corrected in late September 2026: last sentence replaced with our half-life check (I-131 +8%, iodine and noble-gas dose +4%); added the 18-month schedule point.