Right now, at least five different companies are telling you their battery has the highest energy density ever achieved. They cannot all be right. The numbers they are quoting range from 260 watt-hours per kilogram to 761, a spread of roughly three times, and every one of them is technically accurate.
That is not a story about anyone lying. It is a story about a measurement that is widely reported and almost never explained. When a company publishes a battery at 604 Wh/kg, the next question should be: at what discharge rate, under what pressure, and for how many cycles? Those three answers move the number more than any chemical breakthrough has.
What Wh/kg Actually Measures, and the Variable Everyone Omits
Energy density is how much energy a battery stores per unit of mass. A watt-hour is the amount of power it takes to run one watt of equipment for one hour. Wh/kg is simply that energy divided by weight, and higher is better because a lighter pack holding the same energy means more range.
There is a second number, Wh/L, which measures energy per unit of volume. It matters enormously and is frequently confused with the first. Two cells can have identical Wh/kg and wildly different Wh/L, because one might be tightly wound and the other loosely assembled. SES AI’s HAPS datasheet makes the gap explicit, listing conventional graphite cells at 260 Wh/kg and 700 Wh/L against its own silicon product at 500 Wh/kg and 1,300 Wh/L, more than a proportional gain.
But neither figure is the trap. The trap is the C-rate.
The C-rate describes how fast a cell is charged or discharged. A cell tested at 0.05C is being drained over roughly twenty hours. A cell at 1C empties in an hour. A cell at 10C empties in six minutes. Every battery is less efficient when hammered than when trickled, and the effect gets much larger as the chemistry gets more aggressive.
SES AI publishes its own data across rates, which is rare and extremely useful. The company’s 100Ah electric vehicle cell reports over 399 Wh/kg at C/20, 390 Wh/kg at C/3, 370 Wh/kg at 1C, and 351 Wh/kg at 3C. One cell, four numbers, a 13% spread. The same company’s 6Ah drone cell is rated at 375 Wh/kg at C/20 and 339 Wh/kg at 1C.
So when a press release says 450 Wh/kg, the honest question is whether that is a trickle measurement of a lab sample or a real-world number from a cell doing actual work. Often it is the former.
The Density Ladder, From Laboratory Record to Commodity Cell
Here is the whole field, sorted.
Tiered by what each number actually represents rather than by who announced it first, the field currently looks like this.
The Laboratory Frontier: 500 Wh/kg and Above
At the very top sits a lithium-sulfur pouch cell reporting 761 Wh/kg, built by Kenji Kakiage and colleagues using sulfurized polyacrylonitrile, published in Communications Engineering. The number is real, and it is a record. It was also measured at 0.05C, which is a discharge taking roughly twenty hours, and later work in the field notes that the cell’s deliberately lightweight design is not suitable for mass production in practical use.
The same chemistry has produced a much more practical 441 Wh/kg cell cycling at roughly 0.5C, according to benchmarking work in Nature Communications, which suggests 500 Wh/kg is genuinely reachable for lithium-sulfur if the rate comes down.
Below that sits a 604 Wh/kg and 1,027 Wh/L result from Tsinghua University, published in Nature in September 2025. It is an impressive result for a fluorinated polymer solid-state cell with no anode material at all. Two caveats belong in the same breath as the number: it was measured at 0.05C, and it required 1 MPa of external clamping pressure to hold the cell together. The paper has since attracted 274 citations and 45,000 reads, and it deserves the attention, but it is not yet a car battery.
The Prototype Tier: 500 to 450 Wh/kg
SES AI announced in September 2026 that it is reviving its anode-free lithium-metal line for drones, with cells available from Q4 2026. Its H6AF cell claims 6.3 Ah, 24.3 Wh and 525 Wh/kg in a 71.5 by 48.5 by 6.2 mm pouch, rated for 7C continuous discharge. The company built the technology for electric vehicles, then returned to it for a market that will pay for density. The 525 figure is a datasheet number for a product about to ship.
Amprius of Fremont, California, has pushed further still. Its own investor materials list 515 Wh/kg as the highest energy density delivered to customers, in a 25 Ah silicon cell measuring 12.7 by 57 by 95 mm, developed for a high-altitude platform customer. The company won a 2026 CES Innovation Award for a 520 Wh/kg design, roughly twice a standard graphite cell at about 260 Wh/kg. In September 2026 it announced a second-generation SiCore cell that reaches 500 Wh/kg at 1C and, crucially, can be built on conventional lithium-ion manufacturing equipment already installed across its contract manufacturing network. Amprius notes it has exceeded 500 Wh/kg before but says those earlier cells relied on specialized processes and were never produced at commercial volume. Its broadly available SiMaxx product ships at 450 Wh/kg and 1,150 Wh/L and has been on sale since early 2022. The gap between the 515 Wh/kg cell and the 450 Wh/kg product is the same lesson this article is about, told by a single company.
The Verified Commercial Tier: 390 to 410 Wh/kg
Factorial Energy reported cells delivered to a global automotive original equipment manufacturer at 390 Wh/kg, with a target of roughly 450 Wh/kg for drone applications by the end of 2026. Its FEST platform has reached 85% manufacturing yield, up from 10% when production began. The company booked its first commercial drone order in July 2026, a first supercar order in January, and expects first revenue in 2027. Earlier, 77Ah FEST cells at 375 Wh/kg over more than 600 cycles powered a Dodge Charger Daytona development vehicle, charging from 15% to 90% in 18 minutes.
SES AI’s 100Ah automotive cell, meanwhile, sits at 399 Wh/kg and 862 Wh/L at C/10, dropping to 351 Wh/kg at 3C. That is a company publishing underperforming numbers for a larger format, which is a good sign for the industry’s honesty problem.
One entry belongs here as a warning. Finnish startup Donut Lab published an independently measured 409 Wh/kg in September 2026, verified by VTT, Finland’s national research center, at a 0.1°C discharge rate. It is a legitimately measured number, and it sits in this tier. But the company also missed its own promise to have a solid-state cell powering a production electric vehicle by the end of Q1 2026, and its chief executive has said the cell that was tested is not the one customers will receive. A verified laboratory figure and a shipping product are not the same claim, and Donut Lab is a clean example of both arriving in the same press release.
The In-Production Car Tier: 350 to 360 Wh/kg
This is the number that matters if you care about cars, and it is far lower than the headlines suggest.
WeLion supplies 360 Wh/kg semi-solid cells to NIO in a 150 kWh pack, delivering over 1,000 km of range in ET7 and ES6 sedans. Once fully packaged, the pack is 260 Wh/kg, which is the figure that represents the metal a car actually carries around. NIO also sells the pack as a battery subscription, an arrangement that exists precisely because these cells are far too expensive to put in every vehicle.
CATL’s Qilin condensed battery reaches 350 Wh/kg and, unusually, that number comes with genuine mass production behind it. It is built on high-nickel cathodes and silicon anodes, and CATL has applied it to passenger vehicles. BYD’s second-generation Blade Battery manages 162 Wh/kg at the system level, a 5% gain, and sells in enormous volume.
The Commodity Baseline: 150 to 190 Wh/kg
Lithium iron phosphate cells still account for a large share of the global market, and they cluster around 150 to 190 Wh/kg. These are the cells that will power the majority of electric vehicles built over the next several years, and they are cheap, safe, endlessly producible, and good enough.
For context, the International Energy Agency projected in 2020 that conventional lithium-ion could reach about 325 Wh/kg at the cell level, near its practical ceiling. The Tesla 4680 cylindrical cell sits at roughly 246 Wh/kg. A widely cited figure of 500 Wh/kg for solid-state designs is a target rather than a shipping product, and the gap between that target and what ships is where most of the confusion lives.
Why the Gap Exists: Three Hidden Variables
Distance on the ladder is not mostly chemistry. It is mostly testing conditions.
Discharge rate. As established, the same cell can read 399 or 351 Wh/kg depending on how hard it is worked. High-density chemistries suffer more because their advantages rely on exotic materials that are also more fragile.
Applied pressure. Solid-state and lithium-metal cells physically need to be squeezed. The Tsinghua cell used 1 MPa, roughly ten atmospheres, to maintain electrode contact. Amprius datasheets recommend 15 psi of external clamping on certain designs. A car pack has to survive vibration for a decade, which is a different engineering problem from holding a cell together on a bench.
Cycle life. This is the number almost nobody mentions. Energy density at 0.05C on the tenth charge is not energy density at 1C on the thousandth. A car battery needs thousands of cycles. A drone needs perhaps a few hundred. That single requirement is the largest filter between the top of the ladder and the bottom.
The Frontier Has Quietly Left the Car
Look again at which products are actually shipping the highest numbers, and a pattern emerges that contradicts most of the coverage in this field.
Amprius 450 Wh/kg: drones and electric aviation. SES AI 525 Wh/kg: drones. SES AI’s 100Ah 399 Wh/kg cell: electric aviation. Factorial’s 450 Wh/kg target: drone programs. SES AI’s own 6Ah drone cell: drones. Amprius cells have flown in the stratosphere, powering AALTO’s Zephyr high-altitude pseudo-satellite during a 67-day record flight that ended on 28 April 2025.
Meanwhile the highest numbers in a production car are 360 Wh/kg, and the highest volume cell in a car is 162.
This is not a coincidence. Aviation customers buy energy density the way Formula 1 teams buy downforce. Weight is the enemy, endurance determines whether the vehicle is viable at all, and cycle life is secondary because aircraft are replaced or re-certified rather than driven for a decade. A drone that needs a battery swap after 300 cycles has lost nothing. A car that does would be useless.
The same divergence explains why the same companies keep reporting different numbers for different products. A drone cell and an automotive cell from one manufacturer are not competing on the same axis. Anyone comparing them directly is reading the datasheet instead of the use case.
The Fluorine Catch Nobody Wants to Discuss
There is a chemical tension running underneath this entire race, and it is almost entirely absent from English-language coverage.
Higher energy density in lithium-ion chemistry has historically come from adding more fluorine. Polyvinylidene fluoride remains the standard electrode binder, and lithium bis(trifluoromethanesulfonyl)imide is a common electrolyte salt. The fluorinated electrolyte that produced the Tsinghua result is what allowed a lithium-rich manganese cathode, a chemistry too reactive for ordinary liquid electrolytes, to function at all. Fluorine is not an incidental ingredient in this field; it is load-bearing.
That matters because in 2023 the European Chemicals Agency proposed restricting roughly 10,000 per- and polyfluoroalkyl substances, and the proposal explicitly names LiTFSI, polyvinylidene fluoride and polytetrafluoroethylene, the exact compounds in widespread battery use. A review in Communications Chemistry catalogues how fluorinated electrolytes generate corrosive hydrogen fluoride on contact with trace moisture, and how fluorine-bearing materials complicate recycling by contaminating recovered product.
Nowhere in this race has anyone claimed to have solved the fluorine problem. The highest-density silicon cells are conventional lithium-ion and use a liquid electrolyte like any other, so they are no more fluorine-free than what came before. What silicon does solve is the anode. Silicon can hold roughly ten times as many lithium ions per gram as graphite, and because it is used in place of the graphite rather than alongside fluorine-based chemistry, the expansion-driven cycle-life penalty that has limited silicon anodes for two decades no longer blocks the density gain. As IL has explained, silicon’s theoretical capacity of 3,579 mAh per gram is nearly ten times graphite’s 372, and it is the route most likely to bring high density to mass-market vehicles. Amprius has delivered cells to Airbus through 2025.
The open question is therefore narrower than it first appears. It is not whether fluorine-free high-density cells exist, but whether regulators restricting fluorinated compounds will tighten the supply chain for the electrolytes that most of the industry, silicon anodes included, still relies on. Nobody has written that trade-off down clearly, and the answer will shape which chemistries dominate the next decade of aviation and automotive hardware.
How to Read the Next Battery Press Release
Five questions, in order, will tell you more than any headline number.
- Is this a cell or a pack? Pack numbers are always lower. NIO’s 360 Wh/kg cells are 260 Wh/kg once assembled. A pack figure means someone already solved the hard part.
- What is the discharge rate? If the number is not stated, assume it was measured at C/10 or slower. Anything at 0.05C is a laboratory curiosity.
- Is external pressure required? Clamping requirements do not survive road vibration. If a datasheet recommends 15 psi, ask how a pack holds that for ten years.
- How many cycles to 80% capacity? This is the number that determines whether a car exists. Under 500 cycles means no car.
- Is it in a product, a pilot line, or a lab? Amprius has shipped silicon anodes since 2022 and has a 500 Wh/kg cell due for commercial availability in Q4 2026. Factorial expects first revenue in 2027. CATL’s Qilin is in mass production now. These are not the same claim.
For a detailed technical breakdown of where the industry actually stands, see our solid-state battery scoreboard.
Apply those five and the ladder stops being confusing. The 761 Wh/kg cell becomes a 0.05C laboratory result in a design unsuitable for production. The 604 Wh/kg cell becomes a 1 MPa clamped pouch at trickle discharge. The 450 Wh/kg cell becomes a real product on conventional production lines, available now, for drones. And the 162 Wh/kg becomes the number that will actually power most of the cars sold this decade.
All four are true. Only one of them is a car battery you could buy.
Frequently Asked Questions About Battery Energy Density
What is the highest battery energy density ever achieved?
A lithium-sulfur pouch cell reporting 761 Wh/kg holds the record for any chemistry, measured at a 0.05C discharge rate. The highest commercially available cell is around 450 Wh/kg from Amprius, and the highest shipping car cell is 360 Wh/kg from WeLion in NIO’s semi-solid packs.
Is a higher Wh/kg always better?
No. Energy density alone ignores volumetric density, cycle life, power delivery, cost, safety and manufacturing difficulty. A cell optimised for maximum Wh/kg is often deliberately poor at everything else, which is why the same technology gets different numbers for drone and automotive versions.
What is the difference between Wh/kg and Wh/L?
Wh/kg is energy per unit of mass, Wh/L is energy per unit of volume. A cell can lead on one and lag on the other. For phones and drones, where space is the binding constraint, Wh/L often matters more than the mass-based figure that dominates car marketing.
Why do solid-state battery claims keep changing?
Because the term covers several unrelated chemistries. Oxide, sulfide and polymer electrolytes behave differently, and “semi-solid” or “condensed” products are not all-solid-state at all. The 500 Wh/kg figure quoted for solid-state is a target for a specific polymer and sulfide roadmap, not an achieved product, and the gap between that target and commercial reality is where most confusion lives.
Will 500 Wh/kg cells reach cars?
Probably, but not from the direction headlines suggest. The cells already above 450 Wh/kg are going into drones and electric aviation, where weight dominates value and short cycle lives are acceptable. The binding constraints for cars are cost, cycle life and pack-level performance, none of which are solved by density alone. Realistically the 350 to 400 Wh/kg tier is the near-term ceiling for production vehicles.
The Number That Actually Matters
Battery energy density is one of the few figures in technology reporting that gets quoted constantly and understood rarely. It is not a single measure, it is a family of measurements taken under incompatible conditions, and the gap between a laboratory record and a car on the road is roughly a factor of two.
None of that means the progress is fake. The 162 Wh/kg cell in a budget EV today and the 450 Wh/kg cell in a drone on sale now are both real advances on the 260 Wh/kg cells of a few years back. But the next time a headline claims a record, the interesting question is not how high the number went. It is what had to be true to measure it, and whether any of that survives contact with a production line.
