Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Ask a hundred people why they own clean energy stocks and you will get a hundred variations on the same answer. Decarbonisation is inevitable. Solar is now the cheapest electricity in history. Governments have committed trillions. The technology curve only points one way.
Every one of those statements is broadly true. None of them has been the dominant driver of returns in the sector for the past five years.
The dominant driver has been the discount rate. Renewable energy equities are, in their cash flow structure and their empirical behaviour, long-duration interest rate instruments. They have been sold as a thematic bet on the future of energy, and bought by people who believe they are expressing a view on climate. What they actually own is a leveraged position on the long end of the curve, wearing a costume.
This distinction is not academic. It determines what these assets do in your portfolio, what they do in a crisis, and what they are about to do over the next two quarters.
What a Solar Farm Actually Is
Strip a utility-scale solar project down to its cash flows and describe it to someone who has never heard of solar power.
There is an enormous outlay at the front. Then there are twenty-five years of highly predictable, low-variance, low-margin receipts, with almost no input cost, minimal operational leverage, and modest maintenance. The revenue is frequently fixed by contract for the first fifteen years. There is no fuel bill to manage, no meaningful commodity exposure, and very little that management can do to change the outcome once the thing is built.
Now ask that person to name the asset class. They will not say growth equity. They will say bond.
That is precisely the point. A solar farm or a wind project is, in cash flow terms, a long-dated fixed income instrument with a construction option attached at the front and an equity-like tail at the back. The duration of those cash flows is enormous. Duration is the sensitivity of present value to changes in the discount rate, and an asset whose cash flows are spread across two and a half decades has a great deal of it.
Contrast this with a gas-fired plant. Lower upfront capital, shorter payback, and the dominant cost is fuel, which is a variable expense passed through in real time. A gas plant’s value is mostly a bet on the spark spread. A solar farm’s value is mostly a bet on the discount rate. They are opposite instruments that happen to occupy the same industry classification.
The Mechanism, Quantified
The theoretical claim would be uninteresting if the magnitude were small. It is not small.
Financing costs account for somewhere between 25% and 50% of the levelised cost of energy for solar PV, and up to half for offshore wind. Academic work on PV cost structures found that raising the nominal WACC from 2% to 10% simply doubles the levelised cost, and concluded that after location, the cost of capital is the single most important input in the entire calculation. Not module price. Not efficiency. Not installation labour. The discount rate.
Sit with that for a moment. An industry that spent fifteen years celebrating a ninety percent decline in module costs can have the entire benefit erased by a move in the ten-year yield, and very nearly did.
The asymmetry against fossil generation is the part that should reframe how you think about sector allocation. Research from Oxford’s Sustainable Finance Group found that in the United States, higher financing costs added 18% to the levelised cost of solar PV without tax credits, while adding only 9% to combined cycle gas turbines. The same monetary tightening damages renewables at roughly twice the rate it damages the incumbent technology they are meant to displace.
The same work noted that the nominal cost of debt for North American solar and wind transactions rose from under 2% in 2022 to nearly 8% by 2024. That is not a marginal repricing. That is a different business.
The Natural Experiment Nobody Wants to Discuss
Here is where the thesis stops being a model and becomes history.
In 2022, Russia invaded Ukraine. European gas prices went vertical. Energy security became the defining political priority of the continent. Every conceivable narrative tailwind arrived at once, and the consensus conclusion was that the renewable transition would now accelerate dramatically.
What followed was the worst three years in the modern history of the sector.
Ørsted, the largest offshore wind developer on earth, took impairments of DKK 28.4 billion on its US portfolio in October 2023 after abandoning Ocean Wind 1 and 2. It then took further charges in January 2025 that brought cumulative writedowns to DKK 40.5 billion, and the company’s own breakdown attributed a specific line item of that impairment directly to higher interest rates. In May 2025 it cancelled the 2.6 GW Hornsea 4 project in the UK, citing supply chain costs, rising interest rates and construction risk. By August 2025 it was raising DKK 60 billion in a rights issue backed by the Danish state, having found that market conditions made a planned partial divestment impossible on acceptable terms.
Vattenfall halted Norfolk Boreas. Avangrid paid a penalty to walk away from Commonwealth Wind after concluding it was unfinanceable. RWE cancelled a 2 GW Australian project as commercially unviable.
Then came the cleanest data point in the entire argument. In August 2025, Germany’s Federal Network Agency auctioned 10.1 GW of North Sea offshore wind capacity and received not a single bid, the first time this had ever happened. The advisory analysis pointed to a widening gap between project risk and expected return, driven by higher capital costs alongside capture price forecasts falling short of earlier assumptions. Industry bodies attributed the failure to increased power market risk and project financing costs. Germany subsequently cancelled its remaining 2026 auctions entirely and pushed them into 2027 pending a redesign.
Note carefully what this evidence is not. The American offshore wind collapse of 2025 and 2026 is substantially political, driven by an administration hostile to the sector, and it would be dishonest to attribute it to monetary policy. That is exactly why the German auction matters so much. Germany had no hostile administration, no policy reversal, and no permitting collapse. It had a spreadsheet that no longer worked at prevailing rates, and developers who could count.
The Double Hit
There is a second mechanism operating simultaneously, and it compounds the first in a way that project finance models handle badly.
Renewable project valuations depend on assumed power prices in the merchant tail, typically years fifteen through twenty-five. Those assumptions are being revised downward at the same time discount rates are rising, because the more solar you build, the less each incremental unit earns at the moment it generates. Spain recorded a record 397 hours of zero or negative power prices in a single quarter this year, more than a third of all daylight hours in the period.
So the numerator shrinks while the denominator grows. Cannibalisation reduces the terminal value at the same moment that the discount rate applied to it increases. This is why the German bidders walked away rather than bidding low: the problem was not that returns were thin, it was that the model no longer produced a defensible number at all.
Readers who followed the argument about why solar cannot rescue anyone from the current oil shock will recognise the shape of this. The binding constraints on the energy transition are almost never the ones being discussed in public.
Why the Costume Matters
If renewables are rate instruments, why does nobody price them that way?
Because the shareholder base was assembled on a different premise. Thematic and sustainability-mandated flows built a holder population that selected these assets for what they represent rather than for their factor exposure. That population reads adoption statistics, policy announcements and technology cost curves. It does not read the two-year note.
This creates a specific and repeating pathology. When rates move against the sector, the fundamental news flow stays good, because installation records keep being broken and costs keep falling. Holders see confirming evidence everywhere and conclude the selling is irrational. They add. The rate factor grinds on regardless, because it was never responding to the news they were reading. The drawdowns are therefore deeper and longer than they would be in a sector whose owners understood what moved it.
The portfolio construction consequence is the one most likely to hurt people. An investor holding long-duration government bonds alongside a clean energy allocation, believing they have diversified across asset classes, has in fact concentrated a single factor bet. Those positions do not offset. They are the same trade expressed twice.
The Live Test, Right Now
This is not a retrospective argument, which is what makes it worth publishing this week rather than at leisure.
Clean energy had an excellent twelve months. The recovery was driven by two things: a cumulative 75 basis points of Federal Reserve easing from late 2025, and a genuinely new demand story from AI data centres contracting for long-duration carbon-free power. Sector strategists surveyed by Morningstar were divided on how much of the rebound was fundamental versus a valuation reset, with one explicitly identifying stabilising interest rates as a key catalyst.
That backdrop has now inverted, and the cause is the oil shock.
Inflation is running near 3.6%. The Federal Reserve under Kevin Warsh has held at 3.50% to 3.75% for five consecutive meetings, and as crude pushed back above $100 this month, futures markets moved to price a meaningful probability of a hike rather than a cut. Nearly half of policymakers indicated at the June meeting that they would support tightening later this year, and market consensus has shifted toward two hikes in 2026, potentially beginning in September or October.
Assemble the chain. A Middle East war raises oil prices. Oil prices raise headline inflation. Inflation forces a hawkish central bank. A hawkish central bank raises the discount rate. The discount rate is the primary determinant of renewable project value.
An oil crisis is therefore transmitted to renewable energy equities as a negative shock, through the rates channel, with a lag of roughly two to three quarters. The intuitive trade, long solar because oil is expensive, has the sign backwards for the assets most people use to express it.
The Strongest Counterargument
Intellectual honesty requires taking the other side seriously, and there is a real one this cycle that did not exist in the last.
Hyperscaler power procurement is a genuinely new variable. When an investment-grade technology company signs a fifteen to twenty year power purchase agreement, it does more than provide revenue. It de-risks the revenue, and de-risked revenue attracts cheaper debt at higher leverage. In other words, AI demand does not merely improve the numerator. It can lower the project’s own cost of capital, partially offsetting the very channel this article describes.
That is a serious argument and it may prove to be the thing that breaks the historical correlation. Two caveats are worth holding alongside it. First, the benefit accrues to projects that secure such contracts, which is a minority of the sector and disproportionately the largest and best-capitalised developers, so it is a differentiator within the sector rather than a lift for all of it. Second, a lower risk premium on a project is not the same as a lower risk-free rate underneath it. Counterparty quality compresses the spread. It does not move the base.
The honest formulation is that AI demand weakens the rate sensitivity of a subset of renewable assets without eliminating it for any of them.
Positioning
The framework generates a fairly clean sorting of the sector by rate exposure rather than by technology.
Highest sensitivity sits with independent power producers and yieldcos: entities that own long-lived contracted generation financed with leverage, whose entire equity value is a duration calculation. These are the purest expression of the trade and should be treated as such in both directions.
Developers carrying large uncontracted pipelines are next. Pipeline value is almost entirely terminal value, which is the most discount-rate-sensitive component of any valuation, and it is simultaneously exposed to the capture price erosion described above.
Equipment manufacturers behave differently and are frequently misclassified alongside the asset owners. Their cash conversion cycle is measured in quarters rather than decades, so their intrinsic duration is short. Their rate exposure is indirect, arriving through customer demand rather than through their own discount rate, and where they hold multi-year backlogs with pricing power, that indirect channel is buffered. This is the same conclusion the transformer and switchgear analysis produced from an entirely different direction, which is usually a sign an argument is load-bearing.
Regulated utilities with renewable programmes sit apart again, because allowed returns on rate base are periodically reset in line with prevailing capital costs. The regulatory compact is, in effect, a partial inflation and rate hedge that merchant developers do not have.
The general principle: in this sector, capital structure and contract duration tell you more about the next twelve months of returns than the technology does.
What Would Prove This Wrong
The thesis is falsifiable, which is the only kind worth publishing.
If the Federal Reserve delivers the hikes markets are now pricing and clean energy equities hold their gains, the rate factor has genuinely weakened and the AI demand story is doing more work than this analysis credits. That would be strong evidence against the argument, and it is directly observable within two quarters.
If hyperscaler PPAs broaden from a handful of marquee deals into a standard financing structure across the sector, the cost of capital advantage stops being a differentiator and becomes a sector-wide reset. Watch the debt terms on new project financings rather than the headlines announcing the contracts.
And if European auction participation recovers in 2027 under redesigned contract-for-difference frameworks, that would demonstrate the problem was auction design rather than the underlying cost of money. WindEurope has argued precisely this, and they may be right. The counter-observation is that a CfD is largely a mechanism for transferring price risk to the state, which lowers the project’s risk premium without changing the rate environment. It treats the symptom well.
The Bottom Line
Buy renewable energy equities if you want. There are good reasons to, and the sector contains genuinely excellent businesses trading on defensible multiples.
But be clear about the position you are taking. You are not primarily expressing a view on decarbonisation, technology adoption, or policy commitment. You are expressing a view on the path of long-term real interest rates, with roughly twice the sensitivity of the fossil generation you are betting against, into a monetary environment that an oil war has just turned hostile.
If that is the trade you meant to put on, size it accordingly. If it is not, you are wearing the costume too.
