InfrastructureCompute, chips & cloud
Oklo, NuScale and Fermi America shares slide as investors balk at AI power start-ups' spending and non-binding deals
Shares of AI-focused power start-ups Oklo, NuScale and Fermi America have fallen sharply in 2026 as heavy spending continues, revenue stays minimal and headline customer agreements remain non-binding, leaving investors less willing to pay for promised future capacity.
From the Financial Times · AI article of November 11, 2026 and cited sources: Stock prices have dropped well below recent fundraising levels. [1] [2] [3] [9]
SOURCES[1] ‘Fear factor’ sets in as investors question AI energy… ft.com[2] Oklo Inc. Form 10-Q, Q3 2026 (SEC) sec.gov[3] Motley Fool: Should you buy Oklo under $30? fool.com
AT Pulse research deskPublished · Updated · 5 min read · AI-assisted analysisAI-assisted, not human-edited. Methodology Corrections Who owns this site
- Original source
- Financial Times · AI article, independent perspective
- Brief by
- AT Pulse research desk AI-assisted; publisher and contact on the desk page
- When
- November 11, 2026 article published
- Covered
- Nov 11 edition
- Evidence
- 14 sources across 10 sites
- 1‘Fear factor’ sets in as investors question AI energy…ft.com
- 2Oklo Inc. Form 10-Q, Q3 2026 (SEC)sec.gov
- 3Motley Fool: Should you buy Oklo under $30?fool.com
- 4Stocktwits: Fermi America shares plunge after losing $150M…stocktwits.com
- 5Axios: Trump-branded AI data center megaproject stalls…axios.com
- 624/7 Wall St.: Fermi America sinks 11.5% with $375M offering247wallst.com
Reach: not measured yet
Stock prices have dropped well below recent fundraising levels. Oklo sold about 32.6 million new shares in 2026 at an average of roughly $70 each, and the stock has since fallen below $30. NuScale is down about 67% this year. The FT calls the investor mood a 'fear factor'. [1] [2] [3] [9]
Several of the largest customer deals are still unsigned or have lapsed. Oklo's filing describes its Meta agreement for up to 1.2 GW in Ohio as non-binding. NuScale's planned 6 GW programme has no signed power purchase agreement. Fermi lost $150M in tenant construction funding when a letter of intent expired, and its Texas campus still has no binding tenant. [2] [4] [6] [9]
Revenue is years away and current results understate spending. Oklo is targeting first power from its Idaho plant in 2028–2029 and reported $1.86M in Q3 revenue. Its Q3 net loss of $41.7M was well below its $84.3M in operating expenses because interest on more than $3B in cash offset part of the costs. Bank of America expects small modular reactors to reach meaningful scale only in the 2030s–2040s. [2] [8] [10]
Company filings and market reports show power deals that are unsigned or have lapsed, with first output still years away.
Read the full assessment
AI infrastructure planners should therefore treat capacity from power start-ups as optional, and should not count on it this decade until binding contracts are signed.
Executive brief
Oklo sold about 32.6 million new shares in 2026 at an average of $70.19 each, according to its Q3 10-Q. By November the stock was under $30 (Motley Fool). The FT reports that a "fear factor" is hitting start-ups that want to power AI data centres: their shares keep falling while they spend heavily (FT). Oklo, NuScale and Fermi America have big plans but little or no revenue. Several of their headline customer deals are still non-binding, so investors are paying them less for promises of future power.
What changed and event timeline
Fermi loses tenant funding
A tenant's $150M construction-funding deal for Project Matador ended when the letter-of-intent exclusivity expired on 9 December. Shares sank more than 46% ().
Fermi's megaproject stalls
Axios reported that the Texas AI data-centre project was stalling and its CEO had left ().
Fermi raises debt
Fermi announced $375M of 5% convertible notes due 2031. Shares fell about 11.5% to roughly $6.48. The company had no revenue and a $188.7M Q1 net loss ().
Week to Oct 19, 2026: Rate shock hits nuclear stocks
A hotter-than-expected September CPI reading reportedly pushed the 10-year Treasury yield to about 5.3%. Oklo and NuScale each fell about 15% that week ().
Short-lived bounce after earnings
Oklo rose 19% after a narrower-than-expected loss, and NuScale gained 15% ().
FT names the "fear factor"
The FT reports that AI energy start-up shares are still sliding while spending stays heavy ().
Capabilities and access
- Oklo: Aurora advanced fission plant in Idaho, now in structural construction. Management targets 2028–2029 for first power. Q3 revenue was $1.86M (24/7 Wall St.).
- Its planned 6 GW programme with ENTRA1 and TVA has no signed power purchase agreement yet (Motley Fool).
- Fermi: Project Matador, a roughly 7,500-acre Texas campus targeting up to 17 GW. It has no binding tenant (24/7 Wall St.; Stocktwits).
Read the full section
- Oklo: Aurora advanced fission plant in Idaho, now in structural construction. Management targets 2028–2029 for first power. Q3 revenue was $1.86M (24/7 Wall St.).
- NuScale: Small modular reactors (SMRs). Its planned 6 GW programme with ENTRA1 and TVA has no signed power purchase agreement yet (Motley Fool).
- Fermi: Project Matador, a roughly 7,500-acre Texas campus targeting up to 17 GW. It has no binding tenant (24/7 Wall St.; Stocktwits).
Technical analysis for researchers and developers
- Oklo has a DOE-approved Preliminary Documented Safety Analysis (June 2026), but its 10-Q still lists completing the NRC regulatory pathway as a major risk (10-Q).
- Oklo's Q3 operating expenses were $84.3M, but its net loss was only $41.7M because interest on $3.13B of cash and securities offset part of the spending (10-Q).
- Bank of America expects SMR adoption to scale meaningfully only in the 2030s–2040s (Motley Fool).
Read the full section
- Approval paths differ. Oklo has a DOE-approved Preliminary Documented Safety Analysis (June 2026), but its 10-Q still lists completing the NRC regulatory pathway as a major risk (10-Q). DOE approval and commercial NRC licensing are separate gates.
- Interest income hides cash burn. Oklo's Q3 operating expenses were $84.3M, but its net loss was only $41.7M because interest on $3.13B of cash and securities offset part of the spending (10-Q).
- Long adoption timeline. Bank of America expects SMR adoption to scale meaningfully only in the 2030s–2040s (Motley Fool). Data-centre planners should not treat SMR capacity as available within this decade.
Claims and evidence
- Oklo's Meta deal (up to 1.2 GW in Ohio) is non-binding.
- The reported size of Oklo's 2026 fall varies. Sources give 46.25% (Motley Fool), "more than 60%" (Motley Fool) and 68% (24/7 Wall St.).
- NuScale is down 67% in 2026, to about $5.22 (Motley Fool).
Read the full section
- Oklo's Meta deal (up to 1.2 GW in Ohio) is non-binding. This comes from the company's own filing, so it is the strongest available evidence (10-Q).
- The reported size of Oklo's 2026 fall varies. Sources give 46.25% (Motley Fool), "more than 60%" (Motley Fool) and 68% (24/7 Wall St.). Differences in measurement dates and start prices may explain the gap.
- NuScale is down 67% in 2026, to about $5.22 (Motley Fool).
- Fermi faces a class action alleging it overstated tenant demand. This is an allegation, not a finding (Stocktwits).
Context and prior work
In October 2025 NuScale peaked at $57.42 and Oklo at about $194 (Startup Fortune). Wider doubts about AI capex are part of the backdrop. Gartner estimates that AI revenue from 2024 to 2030 would need to reach $8.5T to justify data-centre investment, but forecasts only about $3.6T (Digital Today). Local opposition is also slowing data-centre rollouts (FT interactive).
Limitations, safety and contested findings
- Only the FT summary line was available.
- The cause of the slide is disputed. Other analysts point to weaker data-centre build-out and capital constraints (Motley Fool).
- Sell-side optimism remains. One Wall Street analyst projects 240% upside for Oklo (Yahoo Finance).
Read the full section
- Only the FT summary line was available. Its specific claims, named companies and quotes cannot be confirmed, and no other outlet's coverage of this FT story was found.
- The cause of the slide is disputed. Startup Fortune blames rising rates, noting Oklo fell days after a DOE programme selection on 13 October (Startup Fortune). Other analysts point to weaker data-centre build-out and capital constraints (Motley Fool).
- Sell-side optimism remains. One Wall Street analyst projects 240% upside for Oklo (Yahoo Finance).
Business and practitioner implications
- Power deals are not delivered power.
- Raising money is getting more expensive. Oklo issued equity well above today's price, and Fermi is paying a 5% coupon on convertible debt.
- Established utilities look stronger. Incumbents like Constellation are framed as the nearer-term beneficiaries of AI power demand (Motley Fool).
Read the full section
- Power deals are not delivered power. Non-binding agreements and expired letters of intent show that AI infrastructure teams should treat start-up power capacity as optional until contracts are signed.
- Raising money is getting more expensive. Oklo issued equity well above today's price, and Fermi is paying a 5% coupon on convertible debt. Both mean dilution or interest costs for shareholders.
- Established utilities look stronger. Incumbents like Constellation are framed as the nearer-term beneficiaries of AI power demand (Motley Fool).
The source trail.
Sources (14)
‘Fear factor’ sets in as investors question AI energy start-ups
Feed excerpt only.
ft.com