The case for sodium-ion

The premise Peak Energy was founded on has already broken, and Mossburg says so. Two years before recording, the pitch for sodium-ion was that it would be fundamentally cheaper at the level of raw atoms than lithium iron phosphate, a continuation of the earlier shift from nickel manganese cobalt, which traded energy density for cost and turned out to be worth making. Then the price of lithium iron phosphate fell very fast, to roughly half what it had been two years earlier, so the bar rose sharply exactly as the company started. He says the original trajectory remains possible given comparable investment, but that the reason to be interested changed from the economics of the materials to what the chemistry lets you do at the system level.

Start with the gap he has to close. Sodium-ion is not one thing; like lithium-ion it is differentiated mostly by cathode. Layered oxides approach lithium iron phosphate on energy density but cost more because they contain a transition metal, cycle less well, and are harder to design safely, and much of the existing deployment sits there, in small applications like twelve-volt replacements and scooters, where high power and cold-weather performance matter. Peak chose the other branch, a sodium pyrophosphate cathode paired with a hard-carbon anode, which he calls NFPP and never spells out. Its materials are dirt cheap and substantially less energy dense, which he names as the chemistry’s primary problem. On a cost-per-kilowatt-hour basis those cells currently run $15 to $30 above an equivalent lithium iron phosphate cell, against a Chinese cell at $50 to $60, and supplier quotes point to about $20 per kilowatt-hour coming out of the sodium cell price over two to three years. That erodes the premium but not all of it: he expects to be about $10 more expensive in 2028.

The reversal comes at the system level, and it runs against the usual intuition. Lower energy density normally makes everything else worse, since fewer kilowatt-hours per container means more steel, wiring and containers per unit of storage, for the same reason panel efficiency matters in solar. Two chemistry properties invert it. NFPP is comfortable between 45 and 60 degrees Celsius, degrading about as fast there as lithium iron phosphate does at 25, and managing heat is the hardest problem in a grid battery, which sits in a desert for twenty years while power is pushed in and out of it. And the cell’s failure mode is gentler: it begins self-heating at a lower temperature, burns cooler in thermal runaway so propagation is easier to stop, and vents a less explosive gas. Together those let Peak remove the thermal management system outright rather than engineer it better. No fans, no pumps, no chillers, no external auxiliary power, no moving parts at all, on a design philosophy he attributes to a team drawn from Tesla and SpaceX: the best part is no part. Because cooling is the most complex and expensive part of a lithium iron phosphate system, deleting it recovers enough cost and volume to put balance-of-system roughly at parity despite the density penalty, leaving total installed cost within $20 to $30 per kilowatt-hour of a good Chinese lithium system today.

And Mossburg is direct that this is still not a reason to buy. If the story ended with being $10 more expensive in 2028, he says, there would be no reason for his company to exist. The argument rests on the two-thirds of project cost that is not hardware. By his accounting hardware is now about a third of the total; operations and maintenance, including degradation, auxiliary power and round-trip efficiency losses (the share of stored electricity you get back out), is about another third on a net present value basis at a 10% discount rate, and undiscounted he says it is by far the largest bucket; installation and commissioning is the remainder. The industry has pushed on the hardware third since grid storage began, with most of the fall in the last three or four years, and left the other two largely alone. Peak’s claims sit there: roughly 50 times less auxiliary power than an equivalent lithium system, and removal of something close to 90% of the components that need routine maintenance or fail, which also happen to be the components behind most storage fires, which he notes are still fairly rare. That nets about $75 per kilowatt-hour of net present value benefit on a lifetime-cost basis in a hot climate. Cycle life supports it, with fewer degradation mechanisms than lithium iron phosphate and cells near 10,000 cycles still well above 80% state of health, meaning the share of original capacity they still hold. What he does not claim is calendar life. His cells have been on test for a little over a year against a twenty-year product, and while he argues the lithium industry cannot answer that question either, that is an observation about shared ignorance rather than an answer. He puts that limit there himself.