Clean Energy and Nuclear Have Split Five Ways
A subsidy cliff meets AI power demand
① "Clean energy and nuclear" is no longer one trade. Policy (subsidies) and demand (AI power) pulled it in opposite directions — solar and wind lost their tax credits, while nuclear was left untouched.
② Nuclear itself has split too. Equipment/infrastructure suppliers (GEV, +50.7%) and diversified utilities (NEE, +13.2%) rose, while the former poster children of the AI-power trade — Constellation Energy (-23.0%) and Vistra (-33.4%) — actually fell, not because of earnings, but because their valuations reset.
③ The single picture our June energy-transition map drew now needs to be redrawn as five separate stories.
📜 Why they can't be bundled — policy split them first
The One Big Beautiful Bill Act (OBBBA), signed July 4, 2025, rolled back a large share of the IRA's clean-energy tax credits. But it treated each technology completely differently.
| Technology | Under OBBBA | Deadline |
|---|---|---|
| Solar & wind | Investment tax credit (48E) and production tax credit (45Y) terminated | Construction must start by 2026-07-04, or the facility must be placed in service by 2027-12-31 — the construction deadline has already passed |
| Nuclear | This bill didn't touch nuclear's tax-credit provisions — effectively preserved | N/A |
This isn't a matter of market sentiment — it's a difference written into the statute itself. The bar for what counts as "started" is strict, too: the IRS applies a test that "preliminary planning, a contract, or spending alone" isn't enough — there must be physical work of a significant nature. In practice, any new solar or wind project starting from here on effectively can't claim the federal credit.
Our June 11 energy-transition map argued that "large-scale cuts would be politically difficult, since many solar/wind plants sit in Republican-leaning districts." In reality, that cut was already law, and the construction deadline has now passed. The lesson: when assessing policy risk, check what the statute already says rather than what "could" happen.
📊 The data — one "power theme" that split five ways
Line up the one-year returns of the related stocks and ETFs marketbrief tracks, and it's immediately obvious this isn't one story.
| Group | Ticker | 1-year return | P/E | Vs. analyst target |
|---|---|---|---|---|
| Grid/power equipment | GE Vernova (GEV) | +50.7% | 27.0x | +31.6% upside |
| Diversified utility | NextEra (NEE) | +13.2% | 18.1x | +22.3% upside |
| Solar single-stock | First Solar (FSLR) | -7.8% | 12.1x | +39.3% upside |
| AI-power IPP poster child | Constellation (CEG) | -23.0% | 24.9x | +36.7% upside |
| AI-power IPP poster child | Vistra (VST) | -33.4% | 23.7x | +54.7% upside |
| Nuclear/uranium basket | NLR ETF | -21.6% | — | MDD -32.8% |
| Uranium basket | URA ETF | -15.7% | — | MDD -32.1% |
Grouped into five buckets:
| Group | Representative | Character |
|---|---|---|
| ① Equipment/infrastructure suppliers | GEV | Sells gas turbines, grid equipment, even nuclear services broadly — wins regardless of which generation technology comes out ahead. The best performer |
| ② Diversified utilities | NEE | A mix of renewables, nuclear, and regulated utility — steady gains |
| ③ AI-power IPP poster children | CEG, VST | Overheated on 2024–2025 hyperscaler-PPA expectations; now working through a valuation reset |
| ④ Nuclear/uranium theme baskets | NLR, URA | Heavy small-cap and development-stage exposure makes this the most speculative group, with the deepest drawdowns |
| ⑤ Solar single-stock | FSLR | Took the policy cliff head-on, but has been pressed down to a 12x P/E that's now under debate as undervalued |
📉 Why did solid earnings still come with falling stocks — CEG and VST's valuation reset
Group ③ is the most confusing part of this story. Vistra's Q2 2026 adjusted EBITDA jumped more than 30% year-over-year and kept beating consensus, yet its stock drifted lower every single quarter of 2026. Constellation disappointed the market when its 2026 guidance came in below expectations back in March — but the business itself didn't break.
This mirrors the Marvell case from our Jackson Hole analysis. Marvell beat consensus and still fell more than 9% — because how much was already priced in mattered more than the beat itself. CEG and VST built up excessive expectations during the 2024–2025 AI-power hype cycle, and what's happening now isn't a fundamentals problem — it's that inflated bar normalizing.
The September rate hike is accelerating that reset. Stocks like CEG and VST are valued largely on far-future cash flows from long-term PPA contracts, discounted back to present value — and when the discount rate (interest rates) rises, the more distant that cash flow sits, the harder its present value gets cut. Earnings staying flat is enough for the stock to still fall, purely on the rate backdrop.
⚡ Why nuclear captured AI demand and solar didn't
The answer is simple — AI data centers need round-the-clock, uninterrupted baseload power, while solar and wind are intermittent. Battery storage infrastructure isn't built out at the scale needed to close that gap yet.
As of May 2026, announced nuclear deals tied to AI data centers had already exceeded 13 projects and 9.8GW of committed capacity, with every major hyperscaler signing at least one nuclear agreement. AI-optimized facilities also draw roughly 80MW of power, more than double a standard data center's ~32MW — reinforcing the need for large, stable baseload supply.
Policy (the subsidy cliff) and demand (AI power gravitating toward nuclear) both turned against solar and wind at the same time. That doesn't mean solar is finished, though — a stock like FSLR trading down to a 12x P/E is also a sign that much of the bad news is already priced in, shifting the conversation to an undervaluation debate. The +39.3% gap to the analyst target price supports that read.
☢️ Not all nuclear is created equal — new-build vs. existing fleet
Within nuclear, two groups with very different financing structures react differently to the rate environment.
| Type | Example | Effect of higher rates |
|---|---|---|
| Existing operating reactors | Constellation's and Vistra's existing nuclear fleets | Monetize already-built capacity via PPAs — little new capital needed, relatively less rate-sensitive |
| New-build / SMR | Small modular reactor developers | Years to completion, dependent on financing the whole way — the exact zone where investment typically shrinks 40–60% as rates rise |
Projects that take years to complete and require continuous borrowing along the way — new nuclear builds, large SMRs, utility-scale renewables — see their financing costs rise noticeably in a high-rate, hawkish-dot-plot environment like this one. Assets that simply monetize already-built capacity through contracts face much less of that burden. The deeper drawdowns in theme ETFs like NLR and URA versus individual large-caps likely reflect their heavier weighting toward exactly this financing-dependent, small-cap, development-stage cohort.
🛡️ A practical checklist
- Sort each holding into one of the five groups first — equipment suppliers, diversified utilities, AI-power IPPs, nuclear theme baskets, and single-stock solar all carry the same "clean energy" label, yet their performance has run in opposite directions.
- For new solar/wind project names, check the policy deadline first — the construction-start deadline (2026-07-04) has already passed, and that keeps weighing on new pipeline.
- Separate earnings from valuation for AI-power IPPs — buying on "earnings keep beating" alone can mean walking into already-priced-in expectations, as with CEG and VST. Our guide to reading behind price targets helps check what consensus has already baked in.
- Weigh the volatility in theme ETFs heavy on new-build/SMR names — NLR and URA's -32%-range max drawdowns show just how hard this cohort swings during a reset.
You can check how your own portfolio is split across these five groups with our portfolio checkup. For more on how the current rate environment weighs on long-duration cash-flow assets broadly, see Inside the September FOMC.
📌 What to watch next
| Date/event | Why it matters |
|---|---|
| 2026-10-27–28 FOMC | Further hikes would add more valuation pressure on long-duration cash-flow assets — new nuclear, SMR, solar |
| CEG/VST Q3 earnings (mid-October) | Whether continued EBITDA beats finally move the stock — the signal that the valuation reset has run its course |
| 2027-12-31 placed-in-service deadline | Whether projects that missed the construction deadline try to still qualify by hitting this in-service date instead |
| New AI-datacenter/nuclear deal announcements | Whether the 9.8GW committed figure keeps growing and hyperscalers' nuclear tilt solidifies further |
※ Written on September 20, 2026, based on the publicly available text of the OBBBA (signed 2025-07-04) and IRS guidance, market data marketbrief tracks for the stocks and ETFs mentioned (as of 2026-09-18), and public market analysis. The names, returns, P/E ratios, and price targets cited are illustrative for the data discussion and change over time. Tax-credit content is general information, not tax or legal advice. This content is not a recommendation to buy or sell any security; investment decisions and their outcomes are the investor's own responsibility.
- Funding Landscape — IRA Clean Energy Tax Credits After the One Big Beautiful Bill
- Utility Dive — Wind, solar must show 'significant' work under new Treasury rules
- The Motley Fool — Nuclear Stock Face-Off: Constellation Energy or Vistra
- Yahoo Finance — Vistra Won't Stop Declining in 2026
- IAEA — Data Centres, AI and Cryptocurrencies Eye Advanced Nuclear
※ This report is provided for informational and educational purposes only and does not constitute a recommendation to buy or sell any security.
New research, when it lands
Subscribe and the next deep dive comes to you, along with the daily market brief — free, unsubscribe anytime.