The Financial Times supplied an interesting prompt this week. Short sellers have made an estimated $2.1 billion betting against Oklo, NuScale and Nano Nuclear, while $30.3 billion of their combined market value has disappeared since the three small modular reactor developers peaked in October 2025. I have been writing for years that the underlying SMR economics did not support the valuations investors were assigning to the sector, so I went back through my own published work and asked a broader question: how much loss could investors have avoided by treating those technology assessments as investment diligence rather than merely arguments about energy technology? I matched substantive public warnings against subsequent valuations and tried to make the exercise hostile to my own thesis. If a loss had already happened before I published the relevant warning, I did not count it. If a company had reached a much higher speculative peak before the warning, I did not reach backward and claim that value. On that basis, about $95 billion of roughly $119 billion in shareholder value exposed in the core hydrogen, eVTOL and SMR set was subsequently erased, close to 80%. Add X-energy, which listed years after my SMR critique was already public and has since lost another $5.8 billion from its post-IPO surge, and the broader number is about $100 billion. That is market capitalization, not $100 billion of cash physically fed into companies and burned. Nor could every shareholder have sold at the reference values simultaneously. The useful counterfactual is simpler and more personal. An investor considering one of these companies could have looked at the system economics, decided the valuation required too many optimistic assumptions, and put the money somewhere else. Avoiding an 80% loss is capital preservation regardless of whether the alternative investment went up. Hydrogen is the largest part of the bill. In December 2020 I wrote that the emerging “hydrogen economy” narrative was mostly hype outside a narrower set of industrial uses. The argument was not that fuel cells or electrolyzers did not work. It was that delivered hydrogen for energy faced conversion losses, compression and distribution costs, new infrastructure requirements, low utilization during market formation, and direct competition from electrification. After that warning was public, Plug Power, Ballard Power and FuelCell Energy reached combined valuations measured in the tens of billions. Together with the value that remained in Nikola after my warning, the hydrogen portion of the exercise accounts for roughly $59 billion of subsequent shareholder-value destruction. My 2020 article was particularly explicit that hydrogen for ground transportation had already lost to direct electrification. Electric air taxis provide an unusually clean dated receipt. On November 24, 2021, I calculated that the listed urban-air-mobility pure plays had already fallen from $27.92 billion at their peaks to $11.82 billion. I called the sector vastly overvalued and told investors to cut their losses. I do not count the $16.1 billion that had vanished before that article. Four months later, when I checked again, cumulative destruction had increased by another roughly $5 billion. The underlying problem was again a system problem: a vehicle that can fly is not the same thing as a profitable service once certification, maintenance, vertiports, noise, airspace, utilization and the actual addressable passenger market are included. SMRs bring the exercise back to this week’s FT story. In 2021 I argued that making reactors smaller surrendered important economies of vertical scale while assuming factory learning that had not yet been demonstrated. In 2023, after NuScale’s UAMPS project collapsed, I wrote about the investment and SPAC dynamics sustaining the story. The sector then entered another speculative cycle around AI electricity demand. Oklo, NuScale and Nano Nuclear have since lost $30.3 billion from their October 2025 peaks, according to the FT. X-energy adds another $5.8 billion of post-IPO decline if it is included in the broader tally. There is a further wrinkle that makes this more useful than a simple chart of falling prices. Speculative valuations can be extremely valuable to the companies themselves. A developer that sells new shares while its stock is expensive converts narrative enthusiasm into real cash, even if later shareholders suffer badly when the valuation falls. Hydrogen companies have repeated versions of that recapitalization cycle for years, and SMR developers used the latest enthusiasm to strengthen their balance sheets. A collapsing share price can therefore coexist with a company that is financially better equipped to keep spending. That is one reason market-value destruction and actual capital consumed have to be measured separately. The repeated mistake across these technologies is more interesting than the individual stock charts. Investors tend to value the interesting component and assume away the system around it. A fuel cell is not a hydrogen transportation economy. An eVTOL aircraft is not a high-utilization urban aviation network. A reactor design is not a licensed, financed, serially manufactured power plant delivering electricity at a competitive cost. The missing parts are often where capital requirements, delays and weak economics accumulate. This is not a market-timing system. NuScale is a useful warning against pretending otherwise. Its shares went through another enormous speculative rise after I had already criticized the economics. A weak long-run business case can coexist with a rising stock price for a long time, especially when a new narrative such as AI power demand arrives. The useful signal is not “short this stock tomorrow.” It is “do not accept a valuation that requires the whole system to become cheaper, faster and easier than the evidence supports.” There is another category I deliberately left out of the $100 billion figure: fusion. I have been publicly skeptical of commercial fusion electricity since at least 2021, but most of the major companies remain private and their latest financing rounds have generally marked valuations upward rather than downward. That is capital at risk, not capital I can responsibly call destroyed. The exclusion is part of the point. A diligence framework should be willing to count the losses, refuse unresolved cases, and distinguish a weak system thesis from a tradable short-term prediction. The full TFIE Strategy Briefing analysis contains the dated receipts, the denominator rules, the detailed fusion exclusion and a separate calculation of how much actual operating cash and physical capital was consumed rather than merely repriced by public markets. The larger lesson is straightforward: technology diligence is financial diligence. Investors do not need to predict the exact day a narrative breaks to save themselves a great deal of money.
+55 others
