The US Threw Away 20 Million MWh of Clean Energy Last Year. A MIT Startup Wants to Store It in Molten Tin.

The US threw away 20 million megawatt-hours of clean energy last year. Not because the turbines broke — because the grid had no place to put it.

That's curtailment: clean electricity that gets generated, then deliberately discarded. Grid operators do it when supply exceeds what transmission lines can carry or what demand can absorb. California curtailed 3.4 million MWh in 2024, up 29% from the year before. Germany threw away 1,389 GWh of solar power, a 97% increase. In SPP, the regional grid serving parts of the US Great Plains, curtailment increased sixfold since 2020.

The renewable buildout is working. The storage problem isn't solved.

The gap that keeps gas plants running

Here's the basic math that no one likes saying out loud: solar panels peak at noon and go quiet at dusk. Wind is strongest late at night. Electricity demand peaks at 6pm when people get home. Without something to store the midday surplus, you have to keep a gas plant idling somewhere to cover the evening gap.

That's why natural gas hasn't disappeared from the grid despite record renewable installations. The gas plant isn't there to compete with solar — it's there as insurance against the hours when solar isn't running. If you want to actually retire the gas plant, you need storage that runs for eight, twelve, twenty-four hours or more. Lithium-ion batteries typically max out at four hours. They're great for short-term frequency balancing. They can't solve the overnight gap.

Long-duration storage has been the missing piece for a decade. Now one approach is reaching the stage where the physics actually checks out.

A battery that glows white-hot

MIT Professor Asegun Henry and his company, Fourth Power, published an update on March 18, 2026, describing where their thermal battery system stands. The concept is stark: heat carbon blocks to 2,400°C using excess electricity from the grid. Store that heat. When power is needed, pump liquid tin — molten, glowing white — through graphite pipes. The tin heats thermophotovoltaic cells that work exactly like solar panels, except instead of absorbing sunlight, they absorb the intense infrared light pouring off the white-hot metal.

Henry set a Guinness World Record in 2017 for the hottest liquid pump at 1,200°C. The Fourth Power system runs twice that hot. His lab demonstrated thermophotovoltaic cells converting that light to electricity at over 40% efficiency — a record in that category.

The insulation system, made from petroleum coke (a waste product from oil refining), loses only 1% of stored energy per day. That means you can charge it during a surplus on Tuesday and discharge it on Wednesday without significant loss.

Fourth Power can provide anywhere from 10 to over 100 hours of electricity from a single charge. A full commercial unit would deliver 25 megawatts of power with 250 megawatt-hours of storage — the footprint of roughly half a football field.

Why the economics might actually work

The materials are the key. Lithium-ion batteries require lithium, cobalt, manganese, and nickel. The supply chains are long, geopolitically fraught, and don't scale cheaply. Fourth Power's system runs on carbon and tin. Carbon is abundant. Tin is one of the more accessible industrial metals. The graphite bricks are six feet long and 20 inches thick — machined, not synthesized.

"There's not a lot of moving pieces. There's not a lot of players in the supply chain, which means that you can get to cost targets in a fairly straightforward way," CEO Arvin Ganesan told TechCrunch in September 2025.

The target is $25 per kilowatt-hour at scale. Grid-scale lithium-ion currently runs roughly $250/kWh. Natural gas peaker plants — the standard backup — cost between $150 and $300 per kilowatt-hour of capacity installed, plus ongoing fuel costs.

Fourth Power has raised $39 million total across a 2023 Series A ($19 million) and a 2025 Series A Plus ($20 million), backed by Munich Re Ventures, Breakthrough Energy Ventures, and DCVC. The 1 MWh demonstration prototype is under construction for later in 2026. Commercial batteries are targeted for 2028.

The real competition is not other batteries

The framing of "thermal batteries vs lithium-ion" misses the actual competitive target. Fourth Power isn't trying to replace the lithium-ion battery in your phone or your EV. It's trying to replace natural gas peaker plants — the gas turbines that only run during peak demand hours, often just a few hundred hours per year, at some of the highest electricity prices on the grid.

Peaker plants are expensive precisely because they sit idle most of the time. Operators build them anyway because the alternative is rolling blackouts during demand peaks. Replacing a peaker plant with a battery that costs $25/kWh and runs for 100 hours changes the economics of the entire grid, not just the renewables sector.

The Iran war has made this conversation more urgent. The Hormuz closure pushed oil prices above $100/barrel and exposed just how vulnerable every fuel-importing country is to a single chokepoint. The case for accelerating domestic clean energy storage — reducing fossil fuel dependency at the grid level, not just at the generation level — is now a national security argument in Tokyo, Seoul, New Delhi, and Berlin, not just a climate argument.

What has to go right

Fourth Power's demonstration prototype this year will run at 1 MWh — roughly the scale of a large commercial building's daily consumption, not a grid. The jump from a 1 MWh demonstration to 250 MWh commercial units involves solving materials durability at sustained extreme temperatures, graphite pipe performance under thousands of thermal cycles, and tin corrosion behaviour over years of operation.

Henry's team has been testing the components individually. They're now running integrated cycling tests. The word "durable" appears frequently in their public communications — which suggests durability is where the hardest engineering problems still sit.

The physics is confirmed. The materials records are on the books. The company is funded. The competitive case is clear. What's left is the grinding work of cycling the system through enough charge-discharge cycles to know it won't fail after five years in the field.

The renewable energy transition doesn't have a generation problem — it has a storage and transmission problem. Clean power is already being generated faster than it can be used. Twenty million megawatt-hours of evidence landed in 2024 to prove it.

A system that stores electricity as extreme heat in carbon blocks — reliably, cheaply, for days at a time — won't just extend renewable capacity. It'll remove the structural argument for keeping gas plants running. That's the gap Fourth Power is building toward.

The sun in a box isn't a science experiment anymore. It's an engineering project with a 2028 delivery date.