Data from S3 Partners, as cited in a Tuesday media report, reveals that short sellers have accumulated more than $2 billion in paper profits over the past year by betting against three leading small modular reactor (SMR) companies—NuScale Power (SMR.US), Nano Nuclear Energy (NNE.US), and Oklo Inc. (OKLO.US). These stocks, once red-hot amid the unprecedented AI infrastructure boom, have plummeted, wiping out over $30 billion in combined market capitalization across the trio.
The hefty gains for bearish traders cast a clear short-term shadow on sentiment and valuations for emerging nuclear leaders like Oklo Inc., but they do not negate the long-term growth narrative tied to nuclear power within the AI infrastructure surge. With short sellers pocketing roughly $2.1 billion, and both Oklo Inc. and NuScale Power still having about 18% of their float shares out on loan, the market is shifting from a mindset of "AI needs efficient, clean nuclear power, so all nuclear startups deserve high valuations" to a more critical review of licensing, delivery timelines, and the unit economics of SMRs.
From an underlying energy engineering perspective, nuclear supply—especially the SMR route—stands out as one of the most strategically valuable long-term power sources for the data centers now being built at scale. High-density AI GPU clusters demand round-the-clock, high-utilization, stable power with stringent quality and low emissions, and nuclear energy delivers with a typical capacity factor above 90%, exceptional fuel density, minimal operational carbon output, and decades-long asset life, reducing reliance on weather-dependent sources, gas pipelines, and long-distance transmission. Compared to gigawatt-scale traditional reactors, the SMR pathway—via factory fabrication, passive safety features, and modular expansion—theoretically offers lower per-project capital hurdles, allowing deployment to scale from tens of megawatts to hundreds as data centers expand, all while sited near load centers. However, these cost advantages only materialize once "Nth-of-a-kind" replication kicks in; first-of-a-kind builds are often the priciest.
Where the short thesis gains traction
Adam Stein, director of nuclear innovation at the Breakthrough Institute, remarked Tuesday that the industry experienced a "textbook hyper-speculation cycle" last year, noting that "stock prices and valuations were inflated by speculation." Beyond S3 Partners' data showing over $2 billion in short profits against the three SMR leaders, S&P Global Market Intelligence indicates that NuScale Power and Oklo Inc. each have 18% of their float out on loan—a typical proxy for short-selling intensity—while Nano Nuclear Energy has nearly 30% of its float in that state.
A key test of investor risk appetite for the nuclear sector is expected in the coming weeks, as two U.S. nuclear firms with SMR operations—Holtec International and Westinghouse—are slated to debut on U.S. stock exchanges. Tech giants like Meta, Google, Microsoft, and Amazon are increasingly turning to these still-nascent SMR technologies to meet future clean, efficient power demands for data centers, and the Trump administration has publicly backed the nuclear industry, pledging to cut red tape and invest tens of billions in new reactors while reviving old ones. Amid global low-carbon and full-decarbonization trends, nuclear power has become the preferred energy source for Amazon, Google, and Microsoft, offering 24/7 reliable support for their massive data centers. Political and corporate support for nuclear is now stronger than at any time since the 1970s.
Yet, uncertainty persists over when these unproven reactor technologies will achieve large-scale delivery, and some analysts flag concerns about a shortage of high-assay low-enriched uranium (HALEU), a specialized fuel required for SMRs. In response to the media report, a representative from Nano Nuclear Energy said in an email that stock price swings do not fully reflect underlying business success and disputed the "hype cycle" characterization, while Oklo Inc. stated it has made "multiple concrete and positive strides" over the past year, projecting its Aurora reactor to begin commercial operations efficiently in 2028.
From concept chasing to delivery verification
Meta, parent of Facebook and Instagram, has signed long-term agreements for up to 6.6 gigawatts of nuclear capacity, including a partnership with Oklo Inc. to develop up to 1.2 gigawatts in Ohio, with first power targeted for 2030. This underscores that tech demand is real, but a demand contract does not equal an approved, on-time reactor. The short-selling pressure stems from a severe mismatch between nuclear power's demand certainty and SMR firms' timeline for profitability. No SMR is yet commercially operating in the U.S.; Oklo Inc. has made design and testing progress, but its Aurora reactor still needs full NRC licensing, construction, fuel qualification, and grid connection, leaving its 2028 target with little room for delay. Moreover, these three SMR leaders have yet to prove first-of-a-kind costs, construction schedules, and capacity factors at scale.
In Q2, Oklo Inc. generated only about $1.21 million in revenue and a net loss of roughly $48.54 million, though it boosted cash and marketable securities to around $3.01 billion via share issuance, which also brought dilution. More critically, most advanced reactors depend on HALEU, and the U.S. Department of Energy acknowledges that commercial domestic supply remains insufficient. With constraints on fuel, nuclear-grade components, regulatory talent, and construction capacity, a one-year project delay can sharply reduce the present value of future cash flows. Thus, the short thesis focuses on "the market prematurely pricing extreme valuations for dozens of SMRs," rather than questioning nuclear power's fundamental outlook.
SMRs represent a vital future technology path for nuclear energy, but the pre-crash valuations of upstarts like Oklo Inc. were closer to volatile venture-capital options than mature utility assets. Oklo Inc.'s $3 billion in short-term liquidity and its SMR deployment deal with Meta reduce near-term financing risk, yet they do not eliminate technology, regulatory, or construction risks. Judging Oklo Inc.'s true worth requires more than tracking tech-giant orders and AI infrastructure momentum; it demands continuous verification of five milestones—a commercial license, deliverable HALEU fuel, binding power purchase agreements with prepayments or credit support, first-of-a-kind capital costs and schedule, and actual grid connection of the first Aurora unit.
Comments