Catalyst N° 016 of 125 13 Feb 2024

2024 trends: batteries, tax credits, and the cost of capital

with Nat Bullard, independent climate tech analyst and writer, formerly chief content officer at BloombergNEF

In this note

01The question
02The answer
03The argument
04What you need to know first
05Details worth keeping
06Claims worth citing
07Where it’s contested

The question

What do the 2023 data actually say about how decarbonization is going, and where have the binding constraints moved?

The answer

There is no single verdict and the episode does not try for one: it is Kann walking Bullard through about half of thirty slides he picked out of Bullard’s 200-slide annual deck, with a second part promised. The recurring finding is that several things are going far better than most people realize while the aggregate emissions picture has barely moved, and that the constraints on the good news have shifted off technology cost onto finance, feedstock and policy design.

03The argument

The most arresting number sets up everything after it. Measured per person, total greenhouse gas emissions peaked in 1973, the same year the energy intensity of developed-world output peaked, which Bullard treats as one phenomenon rather than two. The catch is what happened next: per-capita emissions have been roughly flat since, only slightly down. Every success story that follows is therefore one side of a race between population growth and the rate of per-capita decline, and the only available move is to push harder on the decline. That makes the optimistic slides more interesting rather than less. Solar installed 440 gigawatts globally in 2023 against something like five when both men started covering it. Light-emitting diodes went from 1% of global residential lighting sales to more than half in twelve years, worldwide and including the global south. Amazon deforestation and land-use-change emissions are both well down from their peaks. All real, none of it yet enough to bend the per-capita line.

Where the episode gets analytically sharp is on money, because that is where the constraint moved. A renewable project is almost entirely upfront capital with small operating costs, which makes it unusually sensitive to the price of money: at a 4% cost of capital roughly three-quarters of the total is the project itself, but at 12% financing becomes 60% of the all-in cost. Bullard calls the preceding decade a holiday from rates history, when the risk-free rate was near zero and nobody had to think about this. What matters is the corollary. The biggest single lever on a project’s cost of capital is the risk-free rate, and no developer can move it, so you are a pure price taker on the variable that dominates your economics, with only currency and sovereign risk spreads layered on top, which is how markets like South Africa or India reach 11% to 13%. That resolves the paradox Kann raises earlier, that module prices kept falling while US solar power purchase agreement prices rose. The module stopped being the story. The brighter finance item is the Inflation Reduction Act’s provision letting tax credits be sold outright, whose first year of pricing behaves as a healthy market should: small transfers take deep discounts and large ones do not, which sorts investors by risk appetite instead of restricting the field to the same specialists who have done tax equity since 2009.

On technology the through-line is that volume manufacturing moves capability faster than analysts expect, and Bullard is careful about which lesson to draw from that. Battery cell manufacturing is entirely Asian in the top ten, and unlike solar it has not consolidated onto one country: China holds first and second while South Korea and Japan remain prominent. Chemistry moved fast too, with lithium iron phosphate going from 5% of electric vehicle batteries in 2019 to 42% in 2023, mostly on the strength of the Chinese vehicle market. Here Bullard explicitly overturns a prior he carried from solar, where one configuration tends to wipe out the others. He expects batteries to separate by use case instead: nickel-based chemistries for hypercars and aviation, lithium iron phosphate for mass-market vehicles, sodium finding its own applications. His test for which laboratory results matter is whether a large manufacturer has committed to volume production, because the commitment is itself the evidence.

The two policy-shaped findings have the most practical bite. Bullard’s recurring slide header is that markets respond to incentives, and his point is that this includes their removal. California’s revision of net metering gutted residential solar volumes, but among the systems still being sold the battery attach rate went from a couple of percent in mid-2023 to nearly 25%, because without the old economics a battery is the only way the math works. He treats that as an intended consequence rather than an accident. Battery recycling cuts the other way. Announced material recovery capacity already exceeds the supply of batteries to recycle and is projected to keep exceeding it through about 2030, by roughly five to one. Kann’s read is blunt: if you are starting a recycling company and are not already sourcing feedstock at scale, that is the only question that matters. Bullard adds the deeper uncertainty, which is that end-of-life is not a fixed category. A battery shipped abroad and kept in service never becomes anyone’s feedstock, and keeping it in use is the more materially efficient outcome anyway.

04What you need to know first

Cost of capital and the risk-free rate: What it costs a project to borrow: a baseline rate set by the wider economy, usually government debt, plus spreads for project and country risk. The baseline dominates and nobody in the industry controls it.
Transferable tax credits: Clean energy projects earn federal tax credits many developers cannot use, having no tax bill to offset, and used to need elaborate ownership structures with an investor who could. The Inflation Reduction Act lets the credit be sold outright at a discount to face value, which is what the pricing data here measures.
LFP versus NMC: Two lithium battery chemistries. Lithium iron phosphate is cheaper and longer-cycling but less energy-dense; nickel manganese cobalt is denser and dearer. That tradeoff is why Bullard expects both to survive.
Net metering: The rule crediting rooftop solar owners for exported power. How generously it is set decides whether a system pays for itself alone or needs a battery alongside it.

05Details worth keeping

The module cost collapse broke the old rule of thumb that modules were half of system cost. Bullard notes it would now imply a 24-cent-per-watt installed utility-scale system, which does not exist. Kann’s version: module cost matters less to the price of solar power than at any point in history, because no other component matched that decline.
Bullard’s lesson from a white paper he wrote roughly a decade earlier: divesting five trillion dollars from fossil fuel equities does not put five trillion into clean energy equities, because that much clean energy stock does not exist and the money goes wherever its yield, liquidity or diversification requirements send it. Going long one thing does not imply going short another.
His LED illustration of why lighting turned over so fast: it is a credit-card decision with automobile-length consequences. He installed about 50 bulbs in a house and moved out five years later without replacing one.
Lighting’s falling share is partly efficiency and partly that we now use electricity for far more things, a point Kann makes and Bullard accepts.

06Claims worth citing

All figures as stated on 2024-02-13, from Bullard’s annual decarbonization deck unless another source is named. This is a 2023 snapshot and prices, shares and deployment figures move fast.

Per-capita greenhouse gas emissions across all gases peaked in 1973 and have been roughly flat, slightly down, since. Methane and nitrous oxide peaked in the 1970s; per-capita carbon dioxide did not peak until well into this century. Developed-world energy intensity of GDP peaked the same year. Bullard
Global solar installations of about 440 gigawatts in 2023, up from 252 in 2022, against roughly five gigawatts globally in 2008. China alone installed north of 200 gigawatts in 2023. Kann and Bullard
Mid-market module efficiency about 15% in 2012 and about 21% now; module price from $1.09 to 12 cents. Units are unstated but the context is dollars per watt, and the low print reflects utility-scale buying in western China rather than a California rooftop. Bullard
Light-emitting diodes were more than 50% of global residential lighting sales in 2022, against 1% in 2010. Bullard
Lighting has fallen from more than a quarter of electricity demand at the start of the twentieth century to barely over 10%, now below both cooling and low-temperature heating. The baseline is 1900, and 10% of today’s demand exceeds 25% of 1900’s in absolute terms. Bullard
Amazon deforestation of about 5,000 square kilometers last year, roughly 60% below 2021 and perhaps a fifth or less of 1990s levels. Bullard
Land-use-change emissions around four gigatons now, down from eight in the late 1950s with spikes above seven in the late 1990s, with researchers’ caveat that these data get revised. Bullard
Global grid storage of about 100 gigawatt-hours installed in 2023, up from around 30 in 2022, after BloombergNEF raised its projection by “almost 30 or 40%,” quoted from memory. BloombergNEF, cited by Bullard
Tax credit transfers under one million dollars price at 84 to 86 cents on the dollar for a 16% to 19% return; transfers around 50 million dollars price at 94 to 96 cents for 4% to 6%, competing with Treasury bills. Bullard says IRR first and corrects himself to return on investment. Basis Climate, late 2023, cited by Bullard; Crux cited as similar
At a 4% cost of capital about 75% of a renewable project’s cost is capital and operating expense and 25% financing; at 12%, financing is 60% of the all-in cost. Kann reading Bullard’s slide, confirmed by Bullard
Costs of capital of 11% to 13% occur in markets such as South Africa or India once currency and sovereign risk spreads are added. Bullard
Long-short clean-versus-traditional excess returns of +17%, +13% and +160% for 2018 to 2020, then -77% in 2021, then -29% and -40% through September 2023; the last two are not clearly assigned to years. Bullard attributes the reversal to clean valuations normalizing alongside oil and gas deleveraging and buybacks. Bernstein research, cited by Bullard
All ten of the largest battery cell manufacturers are Asian, with everyone else totaling 6%. Kann restates that as roughly 94% and Bullard answers “I believe so,” so treat the 94% as inferred. Bullard, restated by Kann
2023 was probably the first terawatt-hour year of battery production, hedged with “I believe.” Bullard
Lithium iron phosphate went from 5% of electric vehicle batteries in 2019 to 42% in 2023, largely on the Chinese vehicle market. Kann and Bullard
Lithium prices rose roughly tenfold from January 2020 to mid-to-late 2022, then collapsed. Bullard
California residential solar battery attach rates went from a couple of percent in mid-2023 to almost 25%, and Kann adds that some installers are well over 50%. Bullard; Kann
By 2030, about two million tons per year of end-of-life batteries plus production scrap against roughly ten million tons of announced recovery capacity, a fivefold gap implying 20% average utilization. Kann notes many announced facilities will not be built. Circular Energy Storage, cited by Kann and Bullard

Forward-looking statements, recorded as predictions rather than findings, with the conditions their speakers attached:

Solar could plausibly double again from 440 gigawatts within a couple of decades, manufacturing headroom not being the constraint. Offered as a suspension of disbelief, with a warning that analysts imposing an asymptotic limit on solar have repeatedly been proven wrong. Bullard
A one-terawatt solar year is “totally plausible,” said while noting the market is in dramatic oversupply with crashed module prices. Kann
Light-emitting diodes will probably saturate lighting, leaving only niche applications. Bullard
Grid storage reaches a first terawatt-hour year, after which uses broaden beyond four-hour duration and frequency and voltage services. No date. Bullard
Lithium iron phosphate will not stay ring-fenced to China, and chemistries separate by application rather than one winning outright. Flagged as a break from his solar-derived prior. Bullard
Battery prices fall toward cost of goods sold as the lithium collapse flows through, conditioned on contract stickiness and on the market staying competitive. He calls the collapse a complex signal for new lithium mines. Bullard
Recycling capacity outruns feedstock through about 2030, conditioned on end-of-life volumes and on how end-of-life is defined. Bullard, on Circular Energy Storage data
Battery attachment becomes the norm in California residential solar, being the only configuration that makes economic sense. Kann
The tax credit transfer market broadens beyond specialists toward exchange-like transparency and liquidity, which Bullard says it has not reached. Bullard

07Where it’s contested

The episode makes no unified claim, by design. It is a curated tour of half of thirty slides, it ends mid-list, and Kann frames the selection as personal favorites. Reading a thesis into it would be reading one in.
Bullard repeatedly refuses to extrapolate. He warns against imposing limits on solar growth, calls his own optimism a suspension of disbelief, and on battery chemistry explicitly interrogates and then rejects the prior he carries from solar.
Several figures carry their own caveats. Land-use emissions data get revised. Deforestation rates fall partly because less forest remains to cut. The 12-cent module price is a specific print many buyers never see. The storage revision is quoted from memory.
Kann goes further than Bullard will on the long-short trade. Kann guesses a long-only clean energy position would have outperformed; Bullard says he could not speak to it and only agrees it would be simpler.
Kann’s recycling framing comes from the investing side, as he says. He evaluates companies pitching battery recycling, so his conclusion that feedstock is the first question is a view from that seat.
”An intended consequence” is Bullard’s inference. He argues a policymaker should have known California’s net metering change would impair the market while pushing batteries into it, but offers no evidence of stated intent.