Holtec Pulled Its $900M Nuclear IPO One Day Before Pricing — The SMR-to-Data-Center Story Just Met the Public Markets

Holtec Nuclear Corporation told the SEC on September 16, 2026 that it was suspending the roadshow for the Nasdaq listing it had launched eight days earlier. The company was marketing 50 million Class A shares at $15 to $18, targeting up to $900 million in proceeds and a fully diluted valuation of $10.2 billion at the top of the range. Pricing had been scheduled for Thursday, September 18. Founder and chairman Dr. Krishna \"Kris\" Singh told the Financial Times the company would not proceed in a \"perfect storm\" in AI-sector valuations, and management gave no new date.

The pattern matters. None of the three 2026 pure-play nuclear listings traded up after debut; two of them are down materially. Holtec's decision to walk away rather than reprice tells the underwriter syndicate — J.P. Morgan, Guggenheim and a roster of co-managers — that even with a $1.52 billion DOE loan guarantee already in place for the Palisades restart, the equity market was not going to underwrite a $10B-plus valuation for a business with a 2025 operating margin of 6.4% and one 805-MW reactor under refurbishment.

The prospectus makes the gap explicit. Holtec International reported 2025 revenue of $576.6 million, down 25% from $765.5 million in 2024, and operating income of $37.1 million, down 87% from $291 million. The 2025 headline net income of $386.6 million was almost entirely an investment-gain line. Pro forma for the corporate reorganization, 2025 revenue rises to $893.4 million and operating income to $140.2 million, but pro-forma net income actually falls to $217 million because of new minority-interest charges. At a $10.2 billion valuation, the deal was being priced at roughly 17.7x trailing revenue and 73x pro-forma operating income — for a company whose entire SMR-300 design is still pre-NRC.

The VanEck Uranium and Nuclear Energy ETF (NLR) peaked on January 28, 2026 and has since given back roughly 35%, finishing the first half down 6.27% on a NAV basis through June 30, 2026. VanEck's own diagnosis is that the pullback is a positioning and valuation reset, not a thesis break: spot uranium held in the mid-$80s, the long-term contract price climbed to about $95 per pound — an 18-year high — and the U.S. Department of Energy committed $17.5 billion in conditional loan guarantees for up to 10 large Westinghouse AP1000 reactors in June.

Within that NLR cohort, the de-rating has been brutally concentrated. NuScale Power and Oklo — the only two publicly traded pure-play SMR developers in the U.S. — are down 85% and 82% from their 52-week highs, and both are pre-revenue on a TTM basis: $10.7M for NuScale and just $1.2M for Oklo as of Q2 2026. Operating generators have re-rated too, but less violently — Constellation Energy is off 36.7% from $410.41, Cameco is off 32.9%, GE Vernova is off just 22.6%. The market is paying up for cash flow and writing down the option value of pre-revenue SMR designs.

The structural story is intact and arguably accelerating. The IEA projects data-center electricity demand will roughly triple from 460 TWh in 2024 to more than 1,000 TWh in 2030 and 1,300 TWh in 2035 in its base case. Goldman Sachs Research has consistently framed 85-90 GW of new nuclear capacity as the swing variable for meeting that load growth. The contract announcements in 2026 alone make the demand side concrete: Microsoft holds a 20-year PPA for the full 835 MW of the restarted Three Mile Island Unit 1 (Crane Clean Energy Center), with the Trump DOE layering a $1B federal loan onto the deal in November 2025; Meta signed an agreement in January 2026 with TerraPower for up to 8 Natrium reactors (initial two units sized at 690 MW) plus separate deals with Oklo and Vistra; and Walmart signed its first long-term nuclear PPA in June 2026, taking 176 MW from Constellation's Dresden plant in two 15-year tranches starting in 2029.

Uranium is a tell. Spot prices have stayed range-bound in the mid-$80s since the early-year spike, but long-term contract prices have pushed to roughly $95/lb, an 18-year high, while Cameco shares have sold off 33% from the January peak. Industry forecasts call for uncovered uranium requirements of more than 3 billion pounds through 2045 and roughly a doubling of primary production. The commodity is signaling the same long-cycle tightness the data-center off-takers are. The equity market is the one short-circuiting the transmission.

Holtec's pullback has a clearly stratified read across the supply chain. Upstream — the uranium miners and the fuel fabricators — has been the most exposed to the sentiment unwind. Cameco, the world's largest pure-play uranium producer, trades at $90.60 against a $135.24 high and a still-rich 160x trailing P/E that prices in long-cycle tightness the equity market is no longer rewarding with multiple expansion. Standard Nuclear, the TRISO fuel pure-play that went public in July at $15, sits at $11.91 — a 20.6% post-IPO drop that effectively marks the new nuclear-fuel template.

Downstream, the off-takers have been the least damaged. Constellation Energy — owner of the largest U.S. nuclear fleet and the only U.S. operator with a hyperscale restart already financed — is at $259.53, off 36.7% from $410.41, but still up materially on a 12-month operating basis. Talen Energy, operator of the Susquehanna nuclear plant that has been a test case for behind-the-meter data-center power, trades at $287.60, with the 50-day moving average at $340.44 reflecting the recent cooling. The asymmetry is informative: the market is paying for megawatt-hours under contract and discounting megawatt-hours not yet built.

Holtec's cancellation is the warm-up act. The actual stress test for the nuclear-IPO thesis is the upcoming listing of Westinghouse Electric Company, which the Wall Street Journal reported in early September is targeting a minimum $30 billion valuation, against the roughly $8 billion Brookfield and Cameco paid for the business in 2023. Westinghouse is the only nuclear franchise with an installed base, a working AP1000 design, a $17.5B DOE loan commitment for up to 10 reactors, and a Cameco strategic partnership underpinning a U.S./UK build program. If Westinghouse can price at $30B-plus in a market that just refused Holtec at $10.2B, the entire nuclear-IPO thesis resets higher. If it has to walk down to the high $20s or pull entirely, the SMR-to-data-center pipeline as a public-equity asset class goes back to a private market for at least another cycle.

Three near-term catalysts decide this. First, Westinghouse's S-1 terms: the headline multiple, the lockup structure and the cornerstone investor list will reveal whether institutions are still willing to anchor a multi-decade nuclear build cycle. Second, NuScale and Oklo quarterly updates: both report Q3 2026 numbers in early November, and any new offtake or revenue line is the most credible signal that the option value is real. Third, the next hyperscaler capex update: Meta's July guide-up was met with a stock drop, and the 2027 capex deceleration UBS is forecasting is the single biggest demand-side risk.

The next 30 days settle the immediate read. Holtec Nuclear can either quietly file an amended S-1 and re-launch once AI-related volatility subsides, or shelve the listing for 2027; the company gave itself the room to do either by suspending rather than withdrawing. A re-launch at a meaningfully lower valuation — say $11-$13 per share, or roughly $7.2B-$8.5B fully diluted — would re-anchor the SMR-300 narrative and take pressure off NuScale and Oklo. A re-launch at the same range would be a worse signal than the suspension, because it would mean the company negotiated harder on optics than on price.