Who Is This For?
If you make decisions about batteries, we make those decisions better.
Cell Designers
You're choosing silicon content, electrolyte, and cycling protocol. You need to know which design lasts longest — without waiting 6 months for aging data.
Run 1,000 virtual aging tests in 30 seconds. Change one variable, see the impact instantly.
Pack & Fleet Engineers
You're managing batteries in the field. You need to predict when packs will hit 80% capacity so you can plan replacements and set warranty terms.
Model real-world conditions — temperature swings, fast charging, partial cycling — and get a predicted end-of-life date with confidence bounds.
Finance & Operations
You're setting warranty reserves, planning fleet budgets, or evaluating batteries for second-life resale. You need numbers you can put in a spreadsheet.
Get physics-backed lifetime projections that hold up under changing conditions — not just curves from past data that break when conditions change.
What Scale Prognostics™ delivers
Ten coupled degradation mechanisms in a single engine. Every simulation returns a complete diagnostic picture — not just a capacity curve.
Lifetime Prediction
See exactly how your battery will age, cycle by cycle. Our physics engine runs all ten coupled mechanisms in real time — not a statistical guess based on similar cells, but a live simulation of what's happening inside yours.
Know Why, Not Just When
Which mechanism is killing your cell? SEI film growth? Silicon cracking? Transport blockage? Mode decomposition breaks capacity loss into root causes so you can fix the right problem.
Multi-Seed Consistency Check
When your training data admits multiple equally-good parameter fits, we surface that as a wider uncertainty band rather than a single confident-looking prediction. Cells that show high prediction spread across our calibration runs get flagged for extended analysis instead of being shipped as falsely-precise forecasts.
Real-World Thermal Effects
A battery in Phoenix ages differently than one in Oslo, even at the same average temperature. We model seasonal and daily temperature swings as independent aging drivers.
Honest Uncertainty Bands
Every prediction surfaces the spread between high-degradation and low-degradation parameter perturbations. When your data sits in a regime where the spread is small, you see a tight band; when the data is sparse or sits near a regime transition, you see a wider band. We don’t manufacture confidence we don’t have.
Auto-Calibration to Your Cell
Public benchmarks aren’t your battery. Hand us as few as 3 retention measurements and the engine fits to your specific chemistry. On our 8 published silicon-graphite benchmarks the calibrated RMSE ranges 0.13–1.23%, all at grade A or better. Your cell’s accuracy depends on how representative your training data is — we surface that explicitly in every calibration response.
Warranty Floor Endpoint
Compute the absolute minimum capacity loss a cell will experience under ideal storage — calendar aging only, no cycling stress. The thermodynamic floor below which no warranty claim is physically valid. Useful for warranty reserves, residual value modeling, and second-life pricing. Returns year-by-year retention milestones over an 8-year horizon, ready for an actuarial spreadsheet.
Three steps to a full prediction
No coding required. Use our web dashboard, or connect through our API or Python SDK if your team prefers.
Describe your cell
Tell us the basics: silicon content, particle type, charge rate, temperature, and how many cycles you want to simulate.
Run the simulation
Our engine models all ten coupled mechanisms simultaneously — SEI growth, silicon cracking, lithium plating, cathode LAM, electrolyte depletion, transport loss, and more.
Get the full picture
See your capacity forecast with confidence bands, a breakdown of which mechanisms caused the loss, and an actionable recommendation.
For developers: Everything available in the dashboard is also accessible through our REST API and Python SDK.
What a simulation tells you
Output for a default 20% Si–graphite NMC cell after 1,000 cycles at 1C and 25°C. Every number reproduces from a real engine run (verify with the curl block at /quickstart).
DEFAULT 20% Si-GRAPHITE CELL — CYCLE 1,000Mode decomposition splits total capacity loss into root causes. This is the question every battery engineer asks: “what’s actually killing my cell?”
Tested against the literature
Calibrated against 8 peer-reviewed datasets spanning 2.5%–20% silicon, multiple charge rates, and temperatures from 25°C to 45°C. We show every result — including the hard ones.
| Source | Silicon | Condition | RMSE | Grade |
|---|---|---|---|---|
| Nature Commun. 2021 (LPD) | 20% | 0.5C, 25°C | 0.13% | A+ |
| Kirk et al. 2024 | 10% | 0.5C, 25°C | 0.17% | A+ |
| HPQ GEN3 18650 | 18% | 0.5C, 25°C | 0.18% | A+ |
| Dose et al. 2023 | 8% | 1C, nano-Si | 0.23% | A+ |
| 5% Si Composite | 5% | 1C, 45°C | 0.25% | A+ |
| Nature Commun. 2021 (HPD) | 20% | 0.5C, knee | 1.23% | A |
| LG M50T | 2.5% | 1C, 25°C | 0.32% | A+ |
| Kirk 2024 Fast Charge | 10% | 2C, fast charge | 1.01% | A |
Grades shown are for calibrated presets (model parameters fitted to each dataset). Out-of-sample accuracy depends on your specific chemistry — use the free trial to test against your own data. We publish every grade because transparency matters.
Built for battery teams
Whether you’re designing cells, writing BMS software, setting warranty terms, or choosing a cooling system — we give you the numbers you need.
Cell Design
Sweep silicon content, particle size, and operating conditions to find the sweet spot between energy density and cycle life — before you build a single prototype.
Battery Management
Generate rich degradation trajectories with 30+ channels to train and validate your BMS. Mode decomposition tells you which mechanism to monitor in the field.
Warranty & Finance
Predict the exact cycle at which capacity crosses 80% end-of-life for any operating envelope. Set warranty terms with physics-backed confidence bands.
Thermal Strategy
Compare passive, forced-air, and liquid cooling. Model real-world seasonal and daily temperature swings. Quantify the lifetime cost of underinvesting in thermal management.
Simple, scalable pricing
A single physical aging test takes 4-8 months and costs $10,000-$80,000 — for one set of conditions. One Starter subscription runs 2,500 simulations per month across any conditions you define.
Annual billing saves 2 months (17% off). Volume discounts available for teams of 10+.