Around 2015, Michael Liebreich told his own analysts that the learning rate for batteries was about 20%. Every doubling of cumulative volume, roughly a fifth off the cost, plus or minus a couple of percent. He offered this, by his own description, with the benefit of no modelling and no research.
The analysts came back with 14%, and they had a defensible reason. They had worked through the bill of materials element by element, and clients' chemists and materials people had signed off on it. Hard to argue with.
He was right and the model was wrong, and his explanation of why is the part worth keeping. The things that took the cost down were not in the bill of materials. Innovation in finance. Machine learning, which at the time nobody in a cell plant thought was relevant. His rule is that when you get that much talent and money and pressure and policymaker push into one place, the shocks are to the upside, and a build-up from components cannot see them coming.
The second half of the explanation is about incentives rather than method. Spotting that a discontinuity is coming is easy. Saying when it arrives is hard, and clients want a date, so the number drifts conservative. What Liebreich says he spent his last few years at Bloomberg doing was pushing analysts the other way: be braver, say what you really think, do not self-censor because a big company came and bounced on you.
He started New Energy Finance in 2004, sold it in 2009, ran it until 2014 and chaired the advisory board until 2017, and is careful to note he has had no executive role at BloombergNEF since.
The 2004 starting conditions explain why almost nobody else was doing it. Solar power was around 50 euro cents per kilowatt hour and the value of the electricity was three or four cents, so the subsidy was effectively the entire business case. One of his first campaigns was against the name. It was called alternative energy, and he could not see what was alternative about something that by 2015 was taking most of the investment going into power generation, and now takes 80 to 90% of it globally.
What green finance is once you take the label off
The difference between green finance and ordinary finance is much smaller than people think.
As soon as private money is involved, most of it has to earn a risk-appropriate return. Multilateral development banks might compromise a little, but the private portion does not bend. Green finance, in his summary, is finance, meaning we need a return appropriate for the risk, and we would like you to be green as well. Which arguably makes it the harder job.
The distinction he thinks people should navigate instead is which bucket of finance fits which stage. Someone spinning a new anode material out of a university lab needs something completely different from someone building their third gigafactory. Selling technology risk to a private equity investor, or country risk to an investor who is not comfortable with that country, wastes everybody's time, whatever colour the label is.
Twenty angel investments and none of them batteries
He volunteers this as a confession. Around twenty angel investments, not one in batteries.
The reasoning is unusually self-aware. He was academically a very strong engineer and describes himself now as a decent engineer and a good generalist. He is not enough of an electrochemist to judge a new cathode material, an electrolyte approach or a flow battery, and his phrase for the position he would be in is that he knows enough to be dangerous. Nor does he have a business process that would let him hand the technical call to a professor and rely on the answer. He does his own diligence, so he has passed on all of it.
His dry summary of the result: he may be a very good battery investor, because he has avoided losing lots of money.
Wind and solar started as venture-type investments and worked their way down to project finance, and he expects battery technology to travel the same road. But batteries differ in one respect. Go back to 1990 or 2000 and there were already large battery companies, while there were no large solar companies at all and in wind there was Vestas and not much else. So there will be a dance between startups and incumbents. The incumbents are reassuring for anyone building something to sell to them. He also thinks they smother, making it harder for a small company to innovate, assemble capital and have a five or seven year crack at the whip.
Every component of a battery also has a stream of data
Asked about the split between technology and finance, Liebreich argues the boundaries are so fluid the question dissolves. There is no hardware any more. Every piece of hardware is also software.
When he started New Energy Finance there was no big data, no machine learning, certainly no large language models and no blockchain. Now every component of a battery carries a data stream: how it was made, how it is performing, how close it might be to failure. Once that data travels with the physical part, the business models stop looking like a single investment followed years later by an exit. Insurance, risk management, battery as a service, power as a service, and what he calls battery material as a service, where an expensive material is leased and still has value at the end of the battery's life.
Cast a wide net, is his instruction, because the industry will absorb what he puts at trillions of dollars and will contain "many, many, many ecological niches", including businesses nobody can currently imagine that will be large in ten or fifteen years.
Classical economists, he thinks, are wrong-footed by this, because so much of it is co-evolution of technologies playing on each other. His example is heat pumps for cars. Not a battery, and not the battery industry's work, but it gives the vehicle more range and makes the pack last longer, so it acts like a battery extension and puts the car into a new market.
That sets up a tension he states rather than resolves. He ran New Energy Finance with ruthless focus: prepaid subscription information services, one event a year, no drift into consulting. But when outcomes are genuinely unclear, whether solid-state arrives, whether aviation becomes a market large enough to sustain a supply chain, optionality matters too. A gigafactory should be able to take different chemistries. Requiring lots of things outside a company's control boundary to all go right is dangerous.
ESG is risk and fairness, and mostly it is risk
Liebreich says he is now old enough to have seen ESG come and to be watching it go, and treats that as a cycle rather than a verdict. His complaint is what it turned into: box-ticking and methodology arguments about scope three.
Stripped back, he says it is about two things, risk and fairness, and both really reduce to risk. The E is whether investors are best served by ignoring environmental issues in their supply chain or their markets. The G is entirely risk management. The S looks like fairness, but unfairness is a risk too: in recruiting, in promotion, in the way a chief executive treats his personal assistant.
On the American pushback, where some states and some people are threatening to sue anyone employing ESG, his answer is to drop the acronym and keep the work. He does not do ESG. He does 360-degree risk, and justice and fairness, for the good of the business, to build it fast while taking as little risk as possible. His advice to anyone whose ESG job is purely box-ticking is to resign today, because they are too talented to waste their lives on something that eliminates no real risk.
In batteries the lens is the supply chain. A great deal more mineral will be mined and refined, and it has to hold a social licence: consultation before a mine opens, processing that does not create environmental hazards, and respect for the decision-making of countries desperate for their own development but with fragile institutions. As the first major consumers of that output, he argues, we carry a responsibility to help them build those institutions.
His objection to the alternative is blunt. It is too easy to do a deal, bring the minerals back and process them in China, where in his assessment nobody cares about the environmental conditions, or does not care as much as we do, and either way leave very little of the value add in the countries the minerals came from.
The row of doom
The Hydrogen Ladder was not his idea. Someone at the Clean Cities campaign posted a sketchy version with five or six categories, and Liebreich fleshed it out. When the criticism came in he chose not to push back, because he was learning too much from it, and crowdsourced further versions instead. It is now on version five.
It is an inevitability scale, not a sequence. It asks what we will actually be doing with clean hydrogen once the transients have settled, the stupid subsidies have gone and experience curves have pushed some costs down, call it 2035 or 2040. At the top sits what we already use hydrogen for: fertilisers and petrochemicals. At the bottom is what he calls the row of doom.
Space heating of any sort is down there, and so is most mobility, because it competes with batteries. His example is the hydrogen scooter. Battery scooters exist, hundreds of millions of people ride them, and his verdict on building a worse one is that it is a stupid idea. The same, he says, goes for taxis, buses and trains. There may be some long-distance freight, but once the easy freight is electrified the residue is small, and the part that is hard to do electrically is easy to do with biofuels, or with compressed biogas today. The case he leaves genuinely open is off-road machinery working where there is no plug.
His summary borrows an old Heineken advertisement. Hydrogen will decarbonise the parts no other technology can reach, because it is expensive to produce, transport, store, distribute and use, so it takes a really good reason to reach for it. Its use, he says, will be a lot less than the current zeitgeist suggests.
That has a career consequence, and he does not dress it up. He feels sorry for the people who have rushed into hydrogen on the strength of the hype and the very large targets. Some will spend ten, fifteen or twenty years in it and never find the exit. Hydrogen businesses will be built, but far fewer than the noise implies.
Batteries he regards as the safer bet, on the grounds that he cannot see the industry going away. His practical advice is to spend six months to a year networking before committing, because a misstep costs a year or two, and to know whether you are a big-company person or a startup person. Startup world is harsh. It is easy to get into and hard to survive, and after a few years it is difficult to get out, because your CV has started to look different.
This piece draws on the full conversation, which is available with a complete transcript on the episode page.