BESS Pricing Reset: Buying When Suppliers Have Margin Again

China's listed battery manufacturers closed their interim reporting season on August 28, and the result was not the one most storage buyers had built into their cost models. CATL reported first-half 2026 net profit of RMB 43.28 billion, up 42 percent year on year, with energy storage revenue up 88 percent to RMB 53.3 billion and now accounting for roughly 19 percent of group revenue. EVE Energy shipped 48.0 GWh into the storage market, up 69 percent, for a 10.4 percent global share. The headline numbers belong to the leaders, but the more consequential line sits underneath them: the second tier of Chinese cell manufacturers returned to profitability, in what analysts described as the broadest profit recovery the sector has seen in years.

For a procurement team, that single fact changes the meaning of every quotation in the file. For roughly two years, the lowest number in a battery energy storage tender was being produced by a manufacturer selling at or below its own cost of production. That is no longer true. The market has not simply become more expensive. It has stopped subsidising the buyer.

A Two-Year Window That Has Now Closed

The price trajectory is well documented. Mainstream 314Ah prismatic LFP storage cells bottomed at roughly 0.26 RMB/Wh during the 2025 low. By March 2026, top-tier quotes were approaching 0.4 RMB/Wh and second-tier suppliers were quoting around 0.35, a recovery of more than 30 percent from the January floor. System pricing followed: mainstream LFP storage systems moved from around 0.55 RMB/Wh in December 2025 to above 0.61 RMB/Wh by February, with two-to-four-hour liquid-cooled systems rising 5 to 8 percent month on month.

Three forces are behind it, and only one of them is a commodity. Lithium carbonate rebounded hard from its late-2025 lows, which lifted the raw material floor. The industry is mid-transition from 314Ah to 500Ah-plus large-format cells, which has tightened availability of both formats at once and pushed delivery lead times out to 45 to 60 days, with rush orders carrying a 5 to 10 percent premium. And demand has been structurally stronger than the supply base planned for: global lithium-battery storage shipments exceeded 461 GWh in the first half, 18 GWh of large-scale BESS capacity came online in July alone, and year-to-date deployment is running 27 percent ahead of last year, with data centre-driven load now competing directly with grid and renewables projects for the same production lines.

The commercial consequence is that the capacity overhang which handed buyers a falling price on every successive RFQ has been absorbed. Procurement teams who set their internal benchmark during that window are now comparing live offers against a reference price that describes a market condition that no longer exists.

The Warranty Was Priced Against a Balance Sheet That Could Not Fund It

The deeper problem is not the benchmark. It is what the benchmark was hiding.

A battery energy storage supply agreement is not a delivery contract. It is a delivery contract wrapped around a fifteen to twenty year obligation. The supplier commits to a capacity retention curve, an availability guarantee, performance liquidated damages, a defined augmentation pathway, and in most cases the future supply of compatible replacement cells at a price mechanism agreed today. Every one of those commitments has a cost. That cost is carried on the supplier's balance sheet and funded out of margin.

When a manufacturer is selling below cost, there is no margin to fund it. The warranty is then a contingent liability with no reserve behind it, and its practical value depends entirely on whether the issuer survives long enough to be called on. Buyers who awarded on lowest evaluated price through 2024 and 2025, without testing the issuer, bought exactly that: a twenty-year instrument written by a company that was not covering its own cost of goods in the year it signed. The instrument was never worth what the bid tabulation implied.

A second, quieter failure sits in the escalation clause. Most BESS supply agreements written in the last two years index price adjustment to lithium carbonate. That was a reasonable proxy when the entire price move was raw material driven. It is not a reasonable proxy now. A material share of the current increase is a supply and demand reset, a format transition, and the restoration of manufacturer margin, none of which a lithium index captures. Buyers relying on such a clause are hedged against the smaller of the two variables and fully exposed to the larger one.

How We Read a Storage Offer

At D1R7K0N we treat a BESS package as four separable purchases, and we insist on seeing them separately before any comparison is run.

The cell is named in the purchase order. Integrators substitute. A system quoted against one manufacturer's 314Ah cell and delivered with another's is a different asset with a different degradation curve, a different cycle warranty and a different spare parts future. The cell manufacturer, format and production origin belong in the technical schedule as a fixed term, with any substitution requiring written owner approval and a re-run of the performance model.

The warranty is tested against the issuer, not the document. We request audited financial statements and read gross margin through the downcycle, not just the current period. A supplier that stayed marginally profitable through 2025 is a materially different counterparty from one that has only just returned to profit on the back of a price rally. Where the balance sheet does not support a twenty-year obligation, the gap is closed commercially through a parent company guarantee, a warranty bond, or retention structured against the capacity test schedule rather than against delivery.

The escalation mechanism is defined, capped and shared. An open-ended pass-through is not a price. We specify the index, the reference date, the share of movement passed through, the cap and the collar, and the point at which the price is fixed for good. Where the exposure is a capacity constraint rather than a commodity, we say so and price it as an availability risk instead of pretending an index will cover it.

Augmentation is bought at award, not deferred. The replacement cells a project will need in year seven or ten are the single largest uncontrolled cost in a storage asset's life, and the buyer's leverage over them is at its maximum on the day the original order is placed and at its minimum every day thereafter. Format availability, compatibility commitment and a price mechanism for augmentation volumes belong in the original agreement.

What Changes at the Next Award

The uncomfortable framing of this reset is that storage has become more expensive. The more accurate framing is that it has become legible. A quotation produced by a supplier who has to cover cost carries information that a quotation produced during a price war does not: it tells you roughly what the product actually costs to make, and therefore whether the obligations attached to it can be funded.

Buyers who respond by holding out for last year's number will spend the wait accumulating schedule risk in a market where lead times are extending and rush orders carry a premium. The better response is to change what is being compared. Rebuild the evaluation around delivered cost per usable kilowatt-hour across the guarantee period, with supplier solvency, augmentation price and escalation exposure priced into the tabulation rather than sitting outside it as assumptions.

If you are running a storage procurement into 2027 delivery and want the supplier side of it tested before the award rather than after, that is the work we do.