SMM and InfoLink both put Chinese 314 Ah ESS cells at approximately RMB 0.365/Wh in the first week of September. The level has held through the summer, against aggregate Chinese cell nameplate capacity that on any straight reading should be pushing it lower. The firmness is real. My assessment is that it is temporary: a congestion premium created by the format transition to 500 Ah+ cells, held in place by certification barriers that block substitution and by a constraint on how fast 314 Ah output can grow. It should erode as large-format qualifications complete, whether or not total ESS demand keeps rising.
314 Ah and 500 Ah+ cells are not substitutable
I described in Issue #11 how Chinese cell prices form discontinuous pools separated by qualification status, certification, and delivery terms, with surplus concentrating where the barriers between pools are lowest. Format is another such barrier, and at the moment one of the hardest to cross.
A 314 Ah lithium iron phosphate cell and a 587 or 588 Ah one are dimensionally incompatible. REPT Battero's published specifications give the gap: 71 × 173 × 206 mm for 314 Ah against 72 × 288 × 216 mm for 588 Ah. The width goes from 173 mm to 288 mm, so the cell will not physically fit a module or rack designed around the smaller format. Support structure, compression arrangement, series-parallel topology, busbar configuration and cooling-plate contact area all change with it. A Far East Battery production-line agreement specifies different cell dimensions, series configurations and module arrangements for its 314 Ah and 587 Ah products, and requires fixture and control-recipe changes between them, on a line expressly designed to handle both.
The engineering difference becomes a pricing barrier through certification. China's CQC certification rules group prefabricated ESS products by enclosure, topology, cell and module model, series-parallel configuration, voltage and thermal-management method. Change the cell model and you trigger configuration review, potentially new type testing. CQC's stated processing clock is 60 working days for type testing plus 30 for the certification decision, before any corrections or retests.
An integrator with a certified 314 Ah container shipping into utility-scale projects cannot swap in 500 Ah+ cells without redesigning the module, requalifying the system and recertifying the configuration. For the months that takes, its demand is locked to 314 Ah supply.
Large-format penetration is real but unpriced
SMM estimated 500 Ah+ ESS cells at about 42 GWh in H1 2026, or 8.6% of total ESS-cell shipments. InfoLink put the figure above 10% of utility-scale ESS shipments specifically. Different denominators, so the two percentages should not be averaged. The direction is the same either way: a minority share large enough to absorb engineering and manufacturing attention, not large enough to relieve demand on 314 Ah lines.
GGII estimated more than 400 GWh of new 500 Ah+ capacity entering operation during 2026. Penghui Energy told investors in February that its existing 314 Ah, 100 Ah and 50 Ah lines were running at full production while it planned to begin 587 Ah production during the year. CATL reported 94.86% utilization across its battery-system capacity in H1 2026, but as I noted in Issue #11, the filing does not separate ESS from power batteries or allocate capacity among formats.
Neither SMM nor InfoLink publishes a separate spot price series for 500 Ah+ cells. CATL's online store offered 587 Ah product as of late August but carried no 587 Ah price in its September promotion. InfoLink noted in April that leading large-format quotations were converging toward 314 Ah levels, without publishing a numerical assessment alongside that observation. The absence of a standardized benchmark says the transaction population is still too thin for routine price discovery. It also means the 314 Ah to 500 Ah+ spread, which would be the most direct measure of the congestion premium, cannot be calculated from public assessments.
Output growth is decelerating, not contracting
SMM estimated August ESS-cell output at 90.76 GWh, up 5.6% MoM, and forecast September at 94.47 GWh, up 4.09% MoM. Output did not fall; the rate of increase did. Contraction would point to demand destruction or deliberate curtailment. A slowing growth rate points to a constraint on how fast supply can expand, which in a market where demand is still rising supports price firmness without requiring any reduction in volume.
SMM attributed the slower growth partly to copper-foil allocation, reporting that leading foil producers were directing new capacity toward higher-margin electronic copper foil while cell producers renegotiated foil agreements and reserved production at smaller suppliers. SMM did not name the producers or quantify diverted tonnage.
Issuer filings corroborate the mechanism without sizing it. DeFu Technology reported progress on a 50,000-tonne-per-year high-end electronic circuit-foil project. Tongguan Copper Foil completed 25,000 tonnes per year of new electronic copper-foil capacity in June. Nord disclosed that some of its lines can switch between lithium-battery foil and standard electronic foil, which makes the output split a zero-sum allocation decision, and that processing fees had risen for both high-end ultra-thin battery foil and high-end electronic foil. The filings do not supply a clean margin comparison between the two grades.
I want to be precise about where the causation stops. SMM is the only source I have found directly connecting copper-foil allocation to September ESS-cell production schedules. The issuer disclosures establish that named producers are expanding into electronic foil, and that flexible lines exist where battery-grade output competes with electronic-grade output for the same equipment time. They do not quantify how many tonnes shifted away from battery-grade production in any given month. Copper-foil allocation as a contributing constraint is supported but not measured.
Aggregate surplus is not format-fungible supply
Mysteel forecast Chinese electrolytic copper-foil equipment capacity at 2.294 million tonnes in 2026, including approximately 1.47 million tonnes of lithium-battery foil. GGII estimated domestic ESS-cell nameplate capacity exceeding 1.2 TWh by year-end. Read on their own, those numbers imply continued price compression.
They imply it because they treat aggregate nameplate capacity as format-fungible supply. A buyer with a certified 314 Ah system delivering into a utility-scale project in Q4 cannot reach surplus sitting in 500 Ah+ nameplate that has not achieved stable yield, cleared customer qualification, or passed system-level certification. That buyer's marginal price is set inside the 314 Ah balance, which is tighter than the aggregate figures suggest.
The qualification barrier is already thinning
The congestion premium persists as long as large-format yields remain immature, customer qualifications remain incomplete and system-level certifications remain pending. All three conditions are time-limited. How time-limited is harder to establish than I would like.
On yield, no public source I have found gives format-level yield rates for 500 Ah+ lines. GGII's 400-plus GWh of new large-format capacity entering operation during 2026 is a nameplate figure, not evidence that those lines have achieved stable, qualified output. The Sunwoda executive who identified heat dissipation, vent pressure and cell-to-cell consistency as the main engineering challenges above 500 Ah was describing problems that bear directly on yield, but neither Sunwoda nor any other producer has disclosed a yield figure for the format. This is a data gap, not an assumption.
On qualification timelines the public evidence comes in fragments. Hithium announced 587 Ah sampling in April 2025 and reported mass production and initial delivery by late August 2025, roughly four and a half months from sampling to first manufacturer-reported delivery. That timeline is now over a year old; I cite it because it remains one of the few publicly documented sampling-to-delivery intervals for the format. CQC's stated processing clock adds 60 plus 30 working days on top. Custom independent testing can take up to six months. These clocks run in partial sequence, and no public source collapses them into a single typical duration.
A June 2026 procurement by NARI, the State Grid equipment unit, accepted interim rather than final GB/T 36276-2023 test reports for cells at or above 587 Ah, while imposing completed documentation and delivery-history requirements on established formats. A state-owned utility buyer accepting partial documentation from large-format suppliers is evidence of a market where the new format is commercially active and not yet qualified to the standard 314 Ah supply meets.
SMM projected the large-format share at about 15% in H2, up from 8.6% in H1, and noted that expansion remained subject to production ramp-up and customer validation. The penetration percentage is a trailing indicator rather than the trigger. What erodes the premium is qualification: enough integrators completing system redesign, recertification and customer acceptance to make 500 Ah+ supply executable against the same project pipeline that currently demands 314 Ah cells. If the qualification and certification pipeline moves at the pace the capacity buildout implies, bankable large-format supply should start arriving in volume during 2027. As it does, the 314 Ah demand pool shrinks and the congestion premium with it.
Treat RMB 0.365/Wh as the price of a format-specific bottleneck with an expiry, not as a floor to extrapolate into 2027 contracting. The aggregate surplus that should be compressing ESS cell prices has not gone anywhere. It is sitting behind a qualification barrier that is thinning at a rate the public data does not let me measure precisely.
-
Consumption tax channel effects: EVE Energy issued a July 24 notice seeking to pass the full 2% consumption tax to buyers from September 1, but SMM reported that actual pass-through remained under negotiation — whether system-level prices absorb or transmit the cell increase will reveal relative bargaining power across the supply chain.
-
CATL's finished-goods buildup: CATL's net finished goods rose 110.7% to RMB 47.6B in H1 while contract liabilities fell 25.9%, but the filing does not identify the product, format, or geography composition of that balance — whether it reflects 314 Ah or large-format inventory matters for assessing format-specific supply tightness.
-
Export-control suspension expiry: The November 10 deadline for reimposing controls on high-energy-density cells, battery equipment, and specified materials could distort H2 export volumes if buyers pull shipments forward, complicating any read of ESS absorption trends from trade data alone.
-
Format-specific price discovery: Neither SMM nor InfoLink publishes a separate 500 Ah+ spot assessment, and InfoLink noted in April that leading large-format quotations were converging toward 314 Ah levels — the emergence of a standardized large-format benchmark would be the clearest signal that the congestion premium is collapsing.