Sila Moses Lake: An Operating Plant Still Has to Qualify Every Customer

Commercial manufacturing moves silicon-anode material beyond pilot batches. It does not collapse the remaining gates: repeatable plant yield, accepted customer lots, cell integration and vehicle-program release.

Sila says its Moses Lake, Washington, plant began operations in autumn 2025 and was ramping production in 2026. That is meaningful deployment evidence: a silicon-anode supplier has moved from development-scale output into a purpose-built commercial facility. The disciplined description stops there. “Operating” does not by itself disclose saleable yield, sustained tonnes per year, customer acceptance or use in a production vehicle.

The product is an input, not a complete battery

Sila sells Titan Silicon, an anode material intended to replace some or all graphite in a lithium-ion cell design. The company advertises higher energy density, fast charging and long cycle life. Those are supplier performance claims, not universal properties of every cell that uses the material. Electrode loading, electrolyte, formation, thermal control and the cell maker’s recipe all affect the realised result.

This boundary defines the business. Sila does not need to assemble every pack or sell a vehicle, but its material must fit a customer’s manufacturing process and performance target. A better powder can still create a poor product if expansion, coating behaviour, first-cycle losses or lot consistency make integration difficult. Qualification is therefore part of the commercial product, not paperwork after the science.

Commissioning, operations and ramp are different milestones

In April 2025, Sila described the plant as commissioning and expected production in the second half of that year. Its later accounts say operations began in autumn 2025; a July 2026 update says production was ramping. The sequence is coherent, but each verb answers a different question. Commissioning shows equipment being brought into service. Operations show material being made. Ramp should eventually show rising, repeatable saleable output.

Public reporting still lacks a dated series for actual tonnes produced, first-pass yield, utilisation, rejected lots and customer-accepted volume. That absence is not evidence of failure. It is why the current stage should be called early commercial ramp rather than steady-state production.

Two capacity units cannot be compared without assumptions

The US Department of Energy’s environmental assessment describes infrastructure for up to 300 tonnes per year in an initial phase and up to 2,300 tonnes per year in a later phase. Sila describes initial Phase 1 capacity as enough material for roughly two gigawatt-hours of cells and presents much larger long-term site ambitions.

Tonnes of anode material and gigawatt-hours of finished cells are not interchangeable units. Converting between them requires an assumed material loading per kilowatt-hour, the share of silicon material in the anode, usable manufacturing yield and the cell design. Reporting both figures is useful; treating them as independent proof of each other is not.

The sale begins before the revenue is repeatable

The commercial model is material supply to battery manufacturers and vehicle programs. Sila and Panasonic Energy announced a procurement agreement, while Sila has also named Mercedes-Benz as a customer. These relationships demonstrate demand and qualification work. They do not disclose how much Moses Lake material has been accepted, when it enters customer production, the realised selling price or the gross margin after ramp costs.

A supply agreement can reserve capacity and coordinate engineering years before a vehicle launch. For analysis, the evidence chain is: nominated program, sampled lots, validated cell recipe, customer acceptance, production release, shipment and reorder. Skipping directly from agreement to scaled revenue hides the part of the business most exposed to delay.

Stage	Evidence available	Evidence still needed
Technology	Supplier publishes material and cell-level performance claims	Customer-specific, independently comparable cell results
Plant	Commissioning in 2025; supplier says operations began and ramp is under way	Saleable tonnes, yield, uptime and energy per tonne
Qualification	Named procurement and vehicle relationships	Accepted lots, production release and launch timing by program
Business	Material-supply model and stated initial capacity	Shipped volume, realised price, ramp cost and reorder evidence

The next proof is boring—and decisive

A new materials plant earns headlines at opening. Its durable advantage appears in quieter records: stable lots, fewer process excursions, customers that qualify the material without redesigning the whole cell, and later lines that reuse the first line’s operating knowledge. Moses Lake matters because those records can now exist at commercial scale.

Watch for customer-confirmed production use, a consistent output series and capacity expressed in both physical material and explicitly modelled cell equivalents. Also watch whether future expansions reproduce yield and qualification speed. Those data would separate a large building with promising chemistry from a manufacturing platform.

Offaxis conclusion: Sila has crossed a real threshold from pilot material into an operating commercial facility. The procurement-relevant question is no longer whether silicon can improve a laboratory cell. It is whether Moses Lake can make accepted material repeatedly, at the cost and pace required by customer programs. Factory ramp and customer qualification are the two linked products.

Primary sources & further reading

Published and checked 22 October 2026. Industry analysis, not independent material testing. Supplier claims and Offaxis interpretation are distinguished above.