Cell Energy Density Roadmap to 2035

How much energy will a cell hold in 2030, or in 2035? (If you need a refresher on the two measures, see cell energy density.) Most answers to that question start from a target or a press release. This one starts from the cells themselves. We took every cell in our cell database that has a published energy density and a start-of-production year, 541 cells in all, and plotted them against time by chemistry. The best production cell in each year gives a trend line. We then extended each chemistry to 2035, using that trend, the practical limits of the chemistry, and the targets that governments and cell makers have published.

The short version: conventional NMC and NCA cells are approaching 350 Wh/kg and will struggle to pass 400 Wh/kg. LFP has already made most of its gains. The step to 450 to 500 Wh/kg in an EV-sized cell depends on lithium-metal, silicon-dominant and solid-state cells, and every one of those at that level is still a pre-production sample.

⬤ Graphite / Si-blend

351 → 385 Wh/kg

Best in production today → 2035 midpoint (360–410)

⬤ Next-gen EV-format

381 → 500 Wh/kg

Best EV-format cell in production today → 2035 midpoint (440–560)

⬤ LFP / LMFP

230 → 255 Wh/kg

Best in production today → 2035 midpoint (240–275)

⬤ Sodium-ion

175 → 210 Wh/kg

Best in production today → 2035 midpoint (190–230)

Built from the BatteryDesign.net cell database. Every dot in the charts below is a row in the cell database, which now holds 1,794 cells, from 18650s to storage cells above 600 Ah and from pilot-line solid-state samples to cells in volume production. Each row carries capacity, voltage limits, mass, dimensions, resistance, charge and discharge ratings, cycle life, start of production and sample level. Every value is colour-coded by where it came from: the maker’s own data, a figure calculated from it, or a third-party source. The source link sits with each cell, so you can check any point on these charts yourself.

Gravimetric energy density

Figure 1. Cell Wh/kg by year. Filled dots are production cells (or C and D samples), open circles are pre-production samples and announcements, diamonds are published targets. The dashed orange step in 2026 rests on a single cell whose production status is inferred. Source: BatteryDesign.net cell database v1.8.03.

This post has been built based on the support of and sponsorship from: illumion, Custom Power, Quarto Technical Services, Eatron Technologies and About:Energy

Volumetric energy density

Figure 2. Cell Wh/L by year. The best-cell line only uses cells that publish a mass as well as dimensions. The dashed orange step in 2026 rests on a single cell, ProLogium’s Gen 3.5, whose production status is our inference.

How the charts were built

  • The year for each cell is its start of production. Where a pre-production cell has no production date, we used the year it was announced.
  • A cell counts as production if it is a C or D sample, or has no sample level recorded, and is dated 2026 or earlier. Pre-A, A and B samples and anything dated after 2026 are shown as pre-production. The cell sample maturity page explains the A to D levels.
  • The best-cell line for each chemistry is the highest value reached by any production cell up to that year, so it only ever rises.
  • Four groups: graphite or silicon-blend NMC, NCA, NCMA and LCO; LFP and LMFP; sodium-ion; and next-gen EV-format, which covers semi-solid, solid-state and lithium-metal cells from automotive programmes. Drone and aviation cells using the same chemistries are plotted as a separate dotted line and are not projected.
  • Seven rows were left out because the numbers cannot be right for the stated chemistry, for example LFP cells listed at close to 900 Wh/L. They are being checked in the database.

Chemistry by chemistry

Graphite and silicon-blend NMC and NCA

This group has the longest record, and its best cell has gained roughly 6 Wh/kg and 22 Wh/L each year since 2012. That is slow and steady progress: the best 18650 cells were around 260 Wh/kg in 2014 and reached 280 Wh/kg by 2018, and the best production cells in the database for 2025 and 2026 sit at 320 to 350 Wh/kg and up to about 965 Wh/L.

Extending the same line gives about 350 Wh/kg in 2030 and 380 Wh/kg in 2035. We think that is close to where a graphite-based anode runs out of room, so most of the later gain has to come from adding more silicon to the anode and pushing nickel content in the cathode. Our band for 2035 is 360 to 410 Wh/kg.

LFP and LMFP

LFP moved fastest between 2016 and 2020, going from about 130 to 200 Wh/kg as large prismatic and blade cells arrived, and has crept up since to about 230 Wh/kg. A straight line through that history would predict 300 Wh/kg by 2035. LFP will not get there: the cathode’s capacity and its 3.2 V working voltage set a ceiling well below that. We have capped the band at 240 to 275 Wh/kg for 2035, with LMFP providing the upper part of the range. Volumetric density follows the same shape, from 500 Wh/L today to around 560 Wh/L.

Sodium-ion

The sodium-ion cells in the database only go back to 2023, but the improvement is quick, from 139 Wh/kg to 175 Wh/kg in about three years. Even so, sodium-ion is catching up with early LFP rather than with NMC. We expect roughly 195 Wh/kg by 2030 and 210 Wh/kg by 2035.

Next-gen: silicon-dominant, lithium-metal and solid-state

This is where the chart gets hard to read, because format matters as much as chemistry. Small drone and aviation cells in this group already exceed 500 Wh/kg. SES lists an anode-free lithium-metal cell, the H6AF, at 6.3 Ah and 525 Wh/kg [1], and Amprius expects commercial availability of a 500 Wh/kg SiCore cell in Q4 2026 [2]. These are cells of a few amp-hours built for aircraft that care about every gram.

EV-sized cells in the same chemistries are a separate story, and the charts give them their own line. The first are already in production. WeLion’s semi-solid cells reached 347 Wh/kg in 2023, and ProLogium’s 185 Ah Gen 3.5 cell is quoted at 381 Wh/kg and 903 Wh/L. That already beats the best graphite cell on Wh/kg, but not on Wh/L, where tightly wound 21700 cells still lead at about 965 Wh/L. The projection starts from those production cells and rises to about 430 Wh/kg and 1,000 Wh/L in 2030, then 500 Wh/kg and 1,150 Wh/L in 2035, passing graphite on volumetric density around 2030.

One caution on that line. Its 2026 step rests on a single cell, ProLogium’s Gen 3.5, and counting it as production is our judgement. ProLogium says manufacturing has started, but it has not said how many of the 185 Ah EV cells it is making. The charts draw that step as a dashed line for this reason.

The drone and aviation cells are plotted for reference as a dotted line but not projected. Joining them to the EV-format line would make next-gen appear to fall back in 2027 before recovering, which is an artefact of mixing formats rather than anything the chemistry is doing.

Those numbers line up with China’s Energy-saving and New Energy Vehicle Technology Roadmap 2.0, which sets cell targets of 350 Wh/kg for 2025, 400 Wh/kg for 2030 and 500 Wh/kg for 2035 for high-end vehicles [3]. The same roadmap sets a much lower bar for mass-market vehicles, 200, 250 and 300 Wh/kg, which falls close to the LFP band. Chinese cell makers are now talking about solid-state trial production in 2027 and volume production around 2030 [4], and CATL’s chairman described the technology in June 2026 as level 4 on a 9-point maturity scale [5].

Roadmap values

Chemistry groupBest in production, 20262030 (range)2035 (range)
Graphite / silicon-blend NMC, NCA351 Wh/kg, 965 Wh/L360 (340 to 385) Wh/kg
1,040 (950 to 1,100) Wh/L
385 (360 to 410) Wh/kg
1,100 (1,000 to 1,180) Wh/L
LFP / LMFP230 Wh/kg, 500 Wh/L245 (230 to 260) Wh/kg
530 (490 to 570) Wh/L
255 (240 to 275) Wh/kg
560 (510 to 600) Wh/L
Sodium-ion175 Wh/kg, 350 Wh/L195 (180 to 210) Wh/kg
390 (360 to 420) Wh/L
210 (190 to 230) Wh/kg
420 (380 to 450) Wh/L
Next-gen, EV format381 Wh/kg, 903 Wh/L430 (390 to 470) Wh/kg
1,000 (930 to 1,080) Wh/L
500 (440 to 560) Wh/kg
1,150 (1,000 to 1,300) Wh/L
Projected cell-level energy density by chemistry group.

Why the headline numbers run ahead of production

Plotting sample level alongside energy density makes one pattern obvious. The highest numbers on the chart are almost all open circles. Every EV-format claim above about 450 Wh/kg is at pre-A or A sample, and some are single announcements with no voltage, capacity or cycle life attached. Tailan’s claim of a 120 Ah cell at 720 Wh/kg in 2024 is the extreme case and sits off the top of Figure 1 [6].

It has typically taken three years or more for an announced figure to reach a production cell. Amprius announced its 500 Wh/kg platform in 2023 and shipped prototypes that year [7]. The production cell at that level is due in late 2026. Announced production dates also slip. StoreDot said its extreme fast charging cells would be in mass production in 2024 [8], and EHang and Inx aimed for volume production of certified solid-state cells by the end of 2025 [9]. Neither date has been confirmed as met. Our solid-state production timelines post follows the same pattern for the larger developers.

The database also shows how much a figure tends to fall between sample and product. SES quoted 417 Wh/kg for its 107 Ah Apollo sample in 2021; its 100 Ah EV cell at B-sample in 2025 is 399 Wh/kg. Amprius quoted 500 Wh/kg for samples in 2023, while its production SiCore cells reached about 450 Wh/kg in 2025. As a rule of thumb, read the open circles as 5 to 10% higher, and two to three years earlier, than the matching production cell.

What this means for pack design

  • For a vehicle launching before 2030, plan around the production lines in Figures 1 and 2, not the open circles. An NMC pack designed today can reasonably assume 330 to 360 Wh/kg cells; an LFP pack 220 to 245 Wh/kg.
  • Check the sample level and the test conditions behind any cell figure above 400 Wh/kg. A C/10 capacity from a two-cycle test is a different thing from a C/3 capacity after formation and 500 cycles.
  • Watch volumetric density as closely as gravimetric. EV-format next-gen cells lead graphite cells on Wh/kg today but still trail the best cylindrical cells on Wh/L, and some need stack pressure that adds mass and volume at module level. Our post on solid-state packaging covers that.

Run your own version

These charts use four columns from the database: Wh/kg, Wh/L, start of production and sample level. The same file lets you cut the data other ways, such as by format, by maker, by C-rate or against cycle life, or filter out anything below C sample before you set a target for your own pack. The cell database is included in the yearly download pass, and there is a corporate download pass for use inside a business.

References

  1. SES AI supplier profile, Unmanned Systems Technology, accessed September 2026.
  2. Amprius Introduces SiCore 500, Amprius Technologies, 2 September 2026.
  3. Interpretation of the 14th Five-Year power battery technology roadmap (in Chinese), xnyauto.com, summarising the Energy-saving and New Energy Vehicle Technology Roadmap 2.0.
  4. China’s battery makers target 2027 solid-state cell trial production, CarNewsChina, 20 August 2026.
  5. CATL chairman flags long road to solid-state battery mass production, CnEVPost, 24 June 2026.
  6. This new solid-state battery cell claims to set industry records, Electrek, 3 April 2024.
  7. The All-New Amprius 500 Wh/kg Battery Platform is Here, Amprius Technologies, 23 March 2023.
  8. StoreDot ships EV battery cell samples to global car makers, StoreDot, 7 September 2022.
  9. EHang and Inx achieve breakthrough in solid-state battery technology, EHang, 13 November 2024.

See also: Cell Database, Cell Sample Maturity, Cell Energy Density, Silicon Anode, Solid-State Production Timelines, Solid-State Fast Charging: What the Test Data Actually Shows, Solid State Battery, Semi-Solid State Battery, Anode Architectures for Solid-State Batteries, Sodium-Ion Battery, Cell Benchmarking, 2025 Battery Roadmaps, Battery Cell Design

Leave a Comment