Donut Lab, the Finnish company behind the CES-January solid-state battery billed as first for production vehicles, published a VTT report on 21 October on a −30 °C test. The headline: 93% energy retention. The figure is real; the document — confidential per VTT, released by Donut Lab — is complete, signed and legible. Page 10: during cold discharge, internal temperature reached about +38 °C, surface near −6 °C. After five hours’ stabilisation it began at −30 °C, then self-heated while delivering 1C for an hour—not while at −30 °C.
- Genuine retention: VTT measured real energy from a self-heating module.
- Thermal split: +38 °C inside on discharge; +63 °C on charge.
- No control: no conventional lithium-ion parallel run.
- No life test: Ten reports, none cycle-life despite 100,000-cycle claim.
Cold start, not cold operation
Donut Lab does not conceal the warming. Its statement says:
“the cells inside were at −30 degrees Celsius throughout before the dismantling process began”
That “before” shifts the meaning; two paragraphs later it concedes:
“a functioning battery heats itself; which can be seen directly in VTT’s results, where the cells heat up during discharge.”
The report states:
“The internal module temperature reached approximately +38 °C during discharge and +63 °C during charge”.
The disclosure sits below the headline.
Fairness cuts both ways: the +25 °C run self-heated to +57 °C. Frozen versus warm start makes 93% less deceptive than it appears. Any module this size warms under load. The missing lithium-ion control prevents separating chemistry from thermal mass and full initial charge.
The 93%, 97% and 99% figures all come from one PDF. At −30 °C the module gave 26.821 Ah and 1,253.9 Wh; at +25 °C, 27.770 Ah and 1,353.9 Wh. 96.6% in ampere-hours, 92.6% in watt-hours. Voltage sag explains it: average discharge fell 48.75 V to 46.75 V, about 0.14 V/cell. Donut Lab leads with 97%; much English-language press with 93%. CEO Marko Lehtimäki recalls January’s pledge:
“our technology retains 99 % of its capacity in minus-30-degree conditions”.
That 99% is unmeasured. CTO Ville Piippo’s claim is testable:
“A typical lithium-ion battery would struggle to provide power, even for a brief moment, in such cold temperatures”.
It fails. Energies peer review at −20 °C, 1C found 74.8% LFP, 68% LMO, 45.2% LCO retention. They deliver less, but deliver. An EV battery on sale charges 10–97% in nine minutes and works at −20 °C. Caveat: no open-access peer-reviewed −30 °C figure exists; −20 to −30 °C is nonlinear, not extrapolable.
What the protocol left out
VTT tested discharge only, not cold charging; one current rate; two cycles, reporting the second. No low-temperature cycling degradation, vehicle pack or real range. Donut Lab claims a 100,000-cycle design, yet none of ten reports — seven VTT, three Intertek — includes cycle life. No price, only lower cost than lithium-ion.
Chemistry is unnamed. VTT calls the sample:
“energy storage device supplied by the customer”
“solid-state” appears only in the client-chosen project name, “Solid-State Battery Performance Test 3”. Voltage window: 2.70–4.15 V per cell for V1, 2.3–4.25 V for V1.5. My inference: high-nickel lithium-ion; sodium-ion stays near 3.5 V at 50% charge, LFP tops at 3.65 V — but VTT does not say so.
A client report, not an audit
A state technological centre report, signed by Ari Hentunen as author, reviewed by Mikko Pihlatie and approved by Antti Arasto, Energy vice-president, exceeds most battery announcements — but remains a client report.
“Testing was performed according to the customer’s test plan”; the sample was “supplied by the customer”.
VTT measures what it is asked; it does not audit chemistry, confirm shipping product or choose protocol. The cold cell was V1, Gen1’s “final testing” unit; 409.3 Wh/kg came from V1.5 at 0.1C. Lehtimäki told a Finnish outlet, reported by Electrek, that the tested cell
“ni siquiera es la que se va a enviar a los clientes”.
Three cells in one conversation.
This follows a difficult summer. In June, a Ziroth investigation with more than twenty specialists argued VTT’s voltage curves match lithium-based chemistry. The VTT report is Donut Lab’s best and most revealing material: real measurement by a serious lab, full PDF visible, yet data undercutting the headline. No contradiction, only hierarchy — choosing 97% over 92.6% is the same decision. Cycle life remains unanswered. Until tested, the result is a module with good thermal mass delivering most energy when charged warm and discharged quickly from cold. Honest, not headline.