TEA breakdown: green ammonia and synthetic methane

Start with ammonia, and start with operating cost, because that is what sets the floor. Ball anchors everything to a long-term US selling price of $500 to $600 per ton. Today about 75% of the hydrogen feeding an ammonia plant comes from steam methane reforming, and that hydrogen step accounts for roughly 80% of ammonia’s greenhouse gas emissions, so the simplest version of green ammonia is to swap in electrolytic hydrogen and leave the century-old Haber-Bosch process alone. Now price the swap. Take 50 kilowatt-hours of electricity per kilogram of hydrogen, which is roughly what a good electrolyzer does and which Kann notes only a few machines beat, and take two cents per kilowatt-hour, which is cheap power. That lands at about 20 cents per kilogram of ammonia, or around $210 per ton: almost half the entire budget, spent on electricity, before a dollar of capital is amortized. The reason this matters more than it first appears is that it is not an engineering shortfall anyone can design away. Ball’s point is that the thermodynamics of water electrolysis are what they are: getting the energy cost down far enough already requires running electrolyzers near the thermodynamic limit for hydrogen, so there is little headroom left and the burden shifts onto capital cost, which now has to fit inside the remaining half.

That pressure is what pushes startups toward redesigning Haber-Bosch rather than just feeding it differently, and the redesigns run into a second constraint. The two pitches are lower-temperature, lower-pressure synthesis reactors, so the loop can ramp up and down with renewable output instead of running flat out, and decentralization, so you stop paying to ship a corrosive, hazardous chemical around the world. Both answer real problems. Intermittency genuinely costs something: an electrolyzer running below full capacity forces you to buffer hydrogen on site, and hydrogen storage runs 30 cents to about $1.20 per kilogram of hydrogen as compressed gas, which works out to 5 to 20 cents per kilogram of ammonia. That is large against a $500 to $600 per ton target, and Ball’s point is that it can eclipse the very transport saving decentralization is chasing. The decentralization prize is real but bounded: transport is roughly 20% to 25% of the delivered price, and that share is the entire budget a small plant has to work within. Against it sits the six-tenths rule, the observation that equipment cost scales as capacity raised to about the 0.6 power. Run that in reverse and half the capacity costs well more than half as much. Electrolysis cells and stacks are modular enough to partly escape it; the ammonia synthesis loop and the air separation unit that supplies the nitrogen are not. So a decentralized plant has to pay a scaling penalty out of a transport budget that is at most a quarter of the price.

Synthetic methane is the same arithmetic, harsher. Nothing technical is in the way: the methanation reaction is over a century old, commercial today in coal-to-gas plants, with high conversion, high selectivity and mild reactor conditions. The problem is the bill of materials. Best case you need about half a kilogram of hydrogen per kilogram of methane. Grant the aspirational $1 per kilogram hydrogen on the Department of Energy’s roadmap, and the hydrogen alone costs $10 per million British thermal units of gas, against a Henry Hub price that has spent the past decade mostly in the $2 to $5 range, as low as $1.50 and spiking near $10. So the cheapest input, priced at a level nobody has reached, is already two to five times the whole benchmark commodity, and it scales roughly linearly, so $2 hydrogen means $20 per million BTU. Carbon dioxide matters less but is not free: best case 2.75 kilograms of carbon dioxide per kilogram of methane, so even hitting the hoped-for $100 per ton for direct air capture adds about $6 per million BTU, and today’s direct air capture is in the high hundreds to $1,000 per ton. Before any capital, the product is outside the market.

There is one real lever left, and Thiel is careful about how much it buys. The whole chain is only around 50% efficient, with losses split roughly evenly between electrolysis and the methanation step, which throws off a lot of heat. That split is a hint: high-temperature electrolyzers fed by the methanation heat, with very tight heat recovery and integration, could recover part of it. But his verdict on the base case is that it is really hard, and the scope of that verdict matters. He limits it to run-of-the-mill, high-capacity-factor power-to-gas with no special markets and no edge cases, and leaves the edge cases explicitly open: a buyer who will pay in the $20s per million BTU, where Kann notes some very low-carbon renewable natural gas already sells; an unusually cheap carbon source such as biogenic point-source capture or biogas upgrading where the carbon dioxide and methane already arrive mixed; or capital cheap and flexible enough to grab hours of near-zero or negative-priced power and do something else the rest of the time. Thiel’s summary covers both halves of the episode: what matters is the hydrogen, then the hydrogen, then the hydrogen. Ball’s version for ammonia is that there is no single miracle. You need cheap electrons and cheap electrolyzer capital together, and ideally a reactor that could take air directly and let you delete the nitrogen separation step altogether.