Beyond the Green Premium: Time to Get Real on Costs

In my last piece, I argued that climate technology is entering a new era.

Climate Tech 1.0 was defined by ambition. Climate Tech 2.0 brought capital and policy support. Climate Tech 3.0 is all about economics.

Technologies need to compete without a green premium. Seems straightforward enough. But there is a problem.

After years of doing diligence on companies, developing technologies myself, and refining my own understanding of capital and operating costs, I’ve realized that almost everyone believes their technology can get there.

And, on paper, they usually can. Assume a 2–5X improvement in performance. Assume nth-of-a-kind costs that follow something like the solar cost reduction curve. Assume cheap, 24/7 green power. Assume utilization is high. Assume construction goes smoothly. Assume you can operate with a skeleton crew. Assume the supply chain scales with you.

Put enough of those assumptions together and almost anything can look economic.

This is where I believe the problem exists. There is a fine line between being cautiously optimistic and being unrealistic. Whether you’re an investor, operator, or technologist, figuring out where you sit on that pendulum is incredibly important.

So, I want to spend some time on the assumptions we make when we say a climate technology is “economic.” While the messaging applies broadly, I have focused my feedback towards fellow entrepreneurs with the hopes it may help in building something great!  

First, a case study.

Take a generic electrochemical technology. From batteries to fuel cells, and from CO₂ conversion to green cement, electrochemical technologies have been a cornerstone of Climate Tech 1.0, 2.0, and likely will continue to be in 3.0.

I’ve developed a number of technologies in this space, invested in them as a VC, and reviewed hundreds more. I have built pilot and demonstration facilities, worked with major EPC companies to optimize for capital costs and operating costs, and formed partnerships with many of the leading cell, electrode, and chemical companies.

While each of these technologies has unique scaling and technical tech challenges, the basic economics of electrochemical systems are similar for each of these and a useful way to illustrate the problem.

At a simplified level, you have current density and voltage.

Current density tells you how much product you can make for a given cell area, so it has a major impact on capital cost. Voltage drives energy consumption, so it has a major impact on operating cost.

The problem is that these variables aren’t independent.

Increase current density and you can reduce the amount of cell area you need. Great for CapEx. But increasing current also increases the voltages (V = IR). Not so great for OpEx.

I can’t tell you how often I’ve seen a model assume a 2–5X increase in current density from Gen 1 to Gen 2 while simultaneously assuming a 25%+ reduction in voltage.

Maybe you get improvements in both. Technology does improve.

But if you haven’t demonstrated those numbers on a small cell under idealized conditions, assuming you’ll achieve both at commercial scale probably isn’t “cautiously optimistic.”

For those that have scaled in this space, you know that as you increase size, mechanical tolerances become difficult to maintain, heat management becomes harder, increased cell height leads to fluid pressure differentials in the cell, resistance loses due to electrical distribution increases, gas coalescence can lead to more inactive area, and so on.

You need to make sure that your assumptions in your models are obtainable and that you are not just solving for the answer you need.

Your capital and operating costs should be built around what is realistically achievable. The technology roadmap should improve the economics from there.

And this goes well beyond just the electrochemistry.

I have regularly seen companies model the costs of the electrode, or cell, and forget what is needed to make that plant work. Things like rectification, balance of plant, installation, redundancy, commissioning, infrastructure, and all the other things that show up when a technology leaves the lab and becomes an actual plant.

That leads to a broader point.

Having a lower cost is not the same as having better economics.

We spend a lot of time in climate tech talking about $/ton, $/kWh, or $/unit.

Those numbers matter. But industrial companies don’t actually make investment decisions based on one isolated number.

They care about the whole investment.

How much capital needs to go in? How long does construction take? What happens to production during integration? What new infrastructure is needed? How reliable is the process? What utilization can you actually achieve? What does maintenance look like? How much working capital gets tied up?

And, of course, what return do you get for taking all of that risk?

This is where I think a lot of climate-tech analysis gets too narrow.

If your process requires enormous amounts of new electricity, you can’t just put the electricity price into the model. Where does the power come from? Do you need new generation? Transmission? Storage? What is the capital implication to put that infrastructure in place?

If you need hydrogen, where is it produced? How does it get to you? How is it stored?

If you capture CO₂, what happens after it leaves your process? Do you have to purify it? Do you have to compress it? Do you have to transport it?

If you depend on a new feedstock, is it actually abundant when there are 100 plants consuming it instead of one? Have you taken into account the infrastructure and storage needed for those feedstocks?

None of these questions are particularly novel. Operators think about them every day.

But they can get lost surprisingly quickly when we’re evaluating a new technology.

So instead of asking, “What does this technology cost?”, I think the more useful question is “What has to be true for this technology to work economically at industrial scale?”

That framing changes the conversation. It also reflects how I think about scale.

Almost every company has an nth-of-a-kind cost curve. And there should be one. Engineering improves. Manufacturing gets better. Equipment becomes standardized. Teams learn how to build plants faster.

But “we get cheaper at scale” isn’t really an answer. I wish I had a dollar for every time someone called out Swanson’s Law or Wright’s Law as their justification. You should know which costs will get cheaper and why. Is it due to benefits from economies of scale? Are you using standard scaling factors? Is it due to tech improvements and do you have data that validates that pathway? Is it due to modularization and saving on recurring engineering costs?

Manufacturing labor might fall. Installation time might improve. Equipment might become standardized.

But electricity doesn’t magically follow your learning curve. Neither does a commodity feedstock. Transportation still costs money. Financing a billion-dollar plant is still financing a billion-dollar plant.

There is a huge difference between identifying specific costs that should decline with replication and assuming the entire system eventually becomes cheap enough because solar did.

Think holistically when putting together your technoeconomic analysis and make sure you stress test it.

Do your models still work if your capacity factors drop to 75%? If feedstock or power costs go up by 10%, does it still work?

If you are lucky enough to secure project debt or financing on a first-of-a-kind facility, do the project economics work if you are paying 15+% for the capital?

If you do tornado plots on your economic models, and truly stress test them, you will get a better understanding of where the risks and opportunities lie.

And this is also why I think retrofittability is underrated.

A cement plant, steel mill, refinery, or chemical facility isn’t a blank sheet of paper.

There are billions of dollars of existing assets sitting there: kilns, mills, utilities, power connections, logistics networks, permits, labor, customer relationships, and decades of operating knowledge.

If one technology requires replacing most of that infrastructure and another can plug into it, those two technologies do not start from the same economic position.

Avoided CapEx is real value.

More broadly, I think we should spend less time trying to prove that technologies can work and more time trying to break the economics.

Build the downside case. It is OK if you think it is pessimistic.

Take away the cheap electricity. Lower the utilization. Increase the construction costs. Increase the timelines. Constrain the feedstock.

In the end, if the economics only work when everything goes right, the economics probably don’t work.

That isn’t pessimism. It’s what industrial companies do before putting billions of dollars into the ground.

And a last bit of venting.

Plants cost money. I can’t tell you how many times I have seen a capital cost projection that puts a small 20% adder on equipment cost, or just assumes some capital scaling off of materials costs.

If we tie this back to my case study, accept the fact that electrochemistry is capital intensive. And no, the electrode raw materials are not the bulk of the capital cost. The electrodes, membranes, cell housing, cell support structure, and building are just a small portion of the cost. You also have the electrical rectification, bussing, instrumentation and controls, salt and/or water treatment, gas separations or compression, tank farms, and so on. In a traditional chlor-alkali plant, if memory serves me correctly, the electrolyzers were less than 10% of the overall capital cost.

So, get used to Lang factors. Unfortunately, when you add in engineering, installation, piping, instrumentation, utilities, and buildings, you are typically spending at least 3X the equipment costs to build out a plant. (Note: Someone will correctly point out that this benchmark is specific to high labor cost regions. More to come on this in a future blog post).  

My final words for now.

Climate Tech 3.0 can’t just eliminate the green premium in a spreadsheet. It has to eliminate it in the real world.

The technologies that scale will have to survive the factory, infrastructure, financing, integration, supply chain, and all the annoying realities that come with operating physical assets.

That’s a much higher bar than demonstrating a breakthrough in the lab.

The best technologies shouldn’t become less convincing as you dig into the economics. They should become more convincing.