You can buy a solid-state battery today. It will either be smaller than your fingernail, or it will not be entirely solid.
That one sentence explains most of the noise around this technology. At one end, TDK has been mass-producing all-ceramic solid-state cells for years: surface-mount parts rated in microamp-hours that back up real-time clocks on circuit boards. At the other end, NIO will rent you a 150 kWh "semi-solid" EV pack in China that still contains liquid electrolyte. And in the middle, where the revolution is supposed to happen, an automotive-scale all-solid-state cell you can put on a purchase order does not exist. Not from Toyota. Not from QuantumScape. Not from anyone.
I have been reading "solid-state is three years away" for a decade, and every year the term gets stretched a little further. So this article stays deliberately boring: what verifiably ships in 2026, what the claimed numbers mean at pack level, and what an OEM buyer should do about it.
What Is a Solid-State Battery, and What Counts as One?
A conventional lithium-ion cell moves ions through a liquid electrolyte soaked into a porous separator. A solid-state cell replaces that liquid with a solid ion conductor: an oxide ceramic, a sulfide, a polymer, or a halide. Remove the liquid and you remove the most flammable material in the cell. You also open the door to a lithium metal anode, which is where the dramatic energy density numbers come from.
That is the textbook version. The market version is messier. "Solid-state" now shows up on spec sheets for cells that contain meaningful amounts of liquid or gel. Vendors call these semi-solid, quasi-solid, hybrid, or condensed. Some just call them solid-state and hope nobody asks.
So ask. My first question for any cell pitched as solid-state is simple: what is the liquid electrolyte content, by weight? If the answer is anything other than zero, you are looking at a hybrid. That is not an insult. Some hybrids are genuinely good cells. But a hybrid is a different technology with different failure modes, and it should not borrow the safety story of a technology that has not shipped at scale.
Semi-Solid vs All-Solid-State Batteries: Why the Difference Matters
Semi-solid cells keep a reduced amount of liquid or gel electrolyte, and they are built on production lines that look a lot like conventional lithium-ion lines. That is precisely why they exist: manufacturable now, on mostly proven equipment. The result is evolutionary: higher energy density than mainstream NMC, some improvement in abuse tolerance, familiar formats.
All-solid-state cells contain no liquid at all. This is the version that justifies a decade of hype: pair a true solid electrolyte with a lithium metal anode, and cell-level figures past 400 Wh/kg become plausible on paper, along with faster charging and a smaller fire risk. It is also the version nobody has scaled.
For a pack engineer, the distinction is not academic. A semi-solid cell still ships as Class 9 dangerous goods under UN 38.3, still needs a BMS, still needs real thermal design. You integrate it like the improved lithium-ion cell it is. An all-solid-state cell may demand something conventional cells never asked for: sustained mechanical stack pressure to keep solid interfaces in contact. That pressure has to come from your pack structure, and structure has mass.
If a vendor blurs these two categories in a sales call, treat everything else on the slide with suspicion.
Solid-State Batteries You Can Actually Buy in 2026
Micro ceramic cells. All-solid-state batteries have been in mass production for years. Almost nobody noticed, because they are rated in microamp-hours. TDK's CeraCharge line is the reference example: reflow-solderable ceramic chips, a common part measuring 4.5 by 3.2 mm and storing 100 µAh at 1.5 V. A next-generation material TDK has announced targets around 1,000 Wh/L for wearable coin-cell replacement. If your product runs on milliwatt-hours, solid-state is not the future. It is a catalog item.
The heated veteran. Blue Solutions' lithium metal polymer cells have powered Bolloré's electric buses and car-share fleets in Europe for over a decade. Genuinely solid polymer electrolyte, genuinely on the road. The catch: the polymer conducts poorly at room temperature, so the pack is held at roughly 60 to 80 °C for its entire service life. Real product, real niche, and a standing reminder that "solid-state" by itself guarantees nothing.
Semi-solid EV packs, almost entirely in China. NIO offers a 150 kWh pack built on WeLion semi-solid cells, with a manufacturer-reported figure of about 360 Wh/kg at cell level, backed by 1,000 km-class range demonstrations. Two details tell you where this actually sits: the pack is offered through battery swap and rental rather than outright purchase, and its cost has been reported as comparable to a small car. SAIC's MG, meanwhile, announced an MG4 variant in 2025 that it described as the first mass-produced semi-solid EV. These are halo products proving out a chemistry, not the new mainstream.
One motorcycle and a large asterisk. In January 2026, Finnish startup Donut Lab announced what it calls a production-ready all-solid-state battery, launching in Verge Motorcycles' TS Pro. The claims are extraordinary: 400 Wh/kg, a full charge in minutes, a design life of 100,000 cycles, stability from -30 °C to above 100 °C. Production is reportedly limited to a few hundred bikes this year, the materials system is undisclosed, and no independent third-party validation of the cell-level claims has been published as of this writing. I would love it to be real. But extraordinary numbers plus undisclosed chemistry plus no third-party data is a combination every experienced buyer has met before, and it usually ends in a quiet spec revision. Watch it. Do not design around it.
And here is what is not shipping: an automotive-scale all-solid-state cell in customer cars. As of early 2026, that count was zero, industry-wide. B-samples exist. Pilot lines exist. Purchase orders do not.
Solid-State Battery Energy Density: Claims vs Verified Numbers
Ground truth first, using typical cell-level ranges; exact figures always come from the specific cell's datasheet. Mainstream LFP sits around 150 to 200 Wh/kg. The best high-nickel NMC cylindrical and pouch cells reach roughly 250 to 300 Wh/kg. The shipping semi-solid benchmark is that manufacturer-reported 360 Wh/kg from WeLion. All-solid-state programs talk in the 400 to 500 Wh/kg band, and those are targets, not products.
One published figure deserves attention because it cuts against the hype. QuantumScape's company-published specs for its first-generation QSE-5 cell are about 301 Wh/kg and 844 Wh/L. Those are strong numbers. They are also numbers the best conventional cells nearly reach today. The most credible Western all-solid-state developer is not promising double; it is promising a meaningful but incremental step, delivered later.
Then apply two corrections that buyers should always make.
First, packs are not cells. As a rule of thumb, pack-level gravimetric energy density lands somewhere around 60 to 80 percent of the cell-level figure once structure, busbars, BMS, and thermal management are added. A cell that only performs under sustained stack pressure hands part of its headline number straight back to the compression hardware. Fast charging does not eliminate cooling, either; it concentrates the same heat into fewer minutes.
Second, the target moves. Conventional lithium-ion is not standing still: silicon-blended anodes, cell-to-pack LFP architectures, and steady cathode work keep raising the bar. A solid-state cell entering production in 2028 does not need to beat today's cells. It needs to beat 2028's conventional cells, at a competitive price, with a supply chain behind it. That is a much taller order.
Why Solid-State Batteries Are Hard to Manufacture at Scale
The core problem fits in one sentence: liquids wet surfaces and solids do not. A liquid electrolyte flows into contact with every active particle in the electrode. A solid electrolyte touches only where it touches, and every gap is resistance. Cells manage this with pressure, coatings, and interlayers. All of it works in a lab. All of it adds cost and new failure modes on a production line.
Each electrolyte family then brings its own bill:
- Sulfides conduct beautifully and react with moisture, generating hydrogen sulfide. Manufacturing them needs ultra-dry processing well beyond an ordinary lithium-ion dry room, which means managing a toxic gas risk at production throughput.
- Oxides are stable but brittle. Producing thin, defect-free ceramic layers over large areas at automotive yield is a hard ceramics problem stacked on top of a battery problem.
- Lithium metal anodes plate unevenly if anything is imperfect, and dendrites are excellent at finding separator defects. Low temperatures and high charge rates make both worse.
Yield is the quiet killer. A separator defect that costs a liquid cell a little efficiency can short a solid cell outright. That is why every serious roadmap has years of pilot production baked into it, and why early all-solid-state cells will cost multiples of conventional cells. Toyota has reportedly targeted bringing solid-state cost within roughly 1.5 times liquid cells around 2030. Read that carefully: even the optimist's schedule concedes years of cost disadvantage.
When Will Solid-State Batteries Reach OEM Products?
The public roadmaps cluster tightly: limited automotive launches in 2027 to 2028, meaningful volume closer to 2030. Here is the field as of mid-2026, with everything treated as a company statement rather than a law of physics. These dates have slipped before.
Toyota is the one most people cite. On the sulfide route with Idemitsu and Sumitomo, it has pointed at first BEVs in the 2027 to 2028 window, Lexus first. The electrolyte pilot plant is under construction; there is no production vehicle yet. QuantumScape, licensing to Volkswagen's PowerCo, takes the oxide-separator plus lithium-metal path: B1 samples shipping, pilot line running, vehicle field tests underway, with volume meant to come through licensing rather than its own factories. Samsung SDI is on sulfide too, targeting mass production around 2027, with pilot-line samples out at automakers. None of the three has a cell you can buy.
The semi-solid camp is the one actually shipping. NIO, on WeLion cells, offers the 150 kWh swap pack today at a manufacturer-reported figure near 360 Wh/kg, hybrid chemistry reaching customers through swap and rental rather than the parts counter.
Then the companies keeping their options open. CATL and BYD are developing multiple routes, sulfide among them, with demonstrations floated around 2027 and volume around 2030, while their advanced liquid and semi-solid cells ship meanwhile. Mercedes-Benz and BMW are road-testing prototype packs from partners like Factorial and Solid Power, aiming at series production late this decade. Read the group together and the pattern holds: pilot lines, samples, and field tests everywhere, purchase orders nowhere.
Now the part that matters for most readers here, who are not building cars. Automotive programs get first allocation of every promising new cell. Industrial, medical, and portable-equipment OEMs get access years later, once formats standardize, documentation matures, and MOQs and lead times drop to sane levels. Watching how LFP, silicon-blend anodes, and the 21700 format rolled out, my working rule of thumb is three to five years between first automotive series production and a cell an industrial OEM can buy with mature paperwork at reasonable volume. A 2027-2028 automotive start therefore lands in the early 2030s for the products this blog cares about.
Should OEM Buyers Wait for Solid-State Batteries?
No. If your product ships before 2029, your realistic chemistry shortlist is LFP or NMC in proven formats (we covered why LFP keeps winning these design-ins earlier), with sodium-ion worth a look at the cold-weather and cost-floor margins. Waiting for a better battery is how a product misses two market cycles while competitors ship on cells that exist.
What I would do instead costs almost nothing. Specify the pack interface, not the chemistry: with connector, communication protocol, and mounting cleanly defined, a future cell change is a pack revision, not a product redesign. Give the BMS headroom on voltage range and firmware room for new chemistry profiles, and keep chemistry names out of marketing claims and regulatory filings unless they must be there.
Then, when a vendor pitches you a solid-state or semi-solid cell (this is starting to happen), run these seven questions:
- What is the liquid electrolyte content by weight? Zero, or it is a hybrid.
- What sample stage is this: A, B, or C? Who outside your company has cells right now?
- Energy density was measured at what C-rate, what temperature, and what stack pressure?
- Cycle-life data: how many cells in the dataset, what depth of discharge, what temperature, and can I see a third-party report?
- Is UN 38.3 complete for this exact cell revision? Where does IEC 62133, or the standard for my end application, stand?
- What price and MOQ will you put in writing for delivery this year, not the roadmap year?
- Who is the second source? If the honest answer is nobody, price that supply risk into the program.
A vendor with a real product answers from test reports. A vendor with a roadmap answers from a slide deck. Telling them apart takes about ten minutes and can save you a year. It is the same logic we apply to evaluating any custom battery pack manufacturer: evidence over narrative.
At PackForge, we build custom lithium battery packs with cells that ship: LFP and NMC in 18650, 21700, prismatic, and pouch formats, from cell makers whose datasheets we can hold them to. We track solid-state closely, partly because customers ask and partly because one day it will change our BOM. Today, it changes our reading list. When an all-solid-state cell arrives with a real datasheet, real UN 38.3 paperwork, and a real price, we will happily design a pack around it.
Now I am curious about your side of the table. Has "solid-state" started appearing in your RFQs, roadmap meetings, or vendor pitches? And when you asked the liquid-content question, what did the answer turn out to be? Share it in the comments. The specifics help everyone calibrate.