China's solar supply chain has spent three years in a margin-destroying price war, and every attempt to formally end it has been struck down by Beijing's own antitrust regulator. So when eight polysilicon producers walked out of a room in Shanghai on the evening of August 6 with a pledge not to sell below cost, the market's reaction was not scepticism. It was a rally. Tongwei rose 6.26% the next session, GCL Technology gained 8.96% in Hong Kong, and Xinte Energy jumped 14.04% 1 2. For a sector that has lost money on nearly every ton of polysilicon shipped since 2023, that is the clearest signal yet that investors think this attempt at discipline, unlike the last one, might actually stick.

A Handshake Where a Merger Failed

The eight signatories, Tongwei, GCL Technology, Daqo New Energy, Xinte Energy, Asia Silicon, Xinjiang East Hope New Energy, Qinghai Lihao Clean Energy and Xinjiang Goens Energy Technology, together control more than 90% of China's effective polysilicon capacity 3. Under the initiative, every participant pledged that sales prices, including bids submitted in competitive tenders, would not fall below the cost calculated under China's newly introduced General Rules for the Cost Accounting Model of the Photovoltaic Industry 3. That is a materially different structure from what these same companies tried in mid-2025, when six of them proposed raising roughly 50 billion yuan to buy out and permanently idle about a third of the country's polysilicon capacity 4. This time there is no joint venture, no capital pool, and no shared ownership of idled plants. There is only a shared number nobody is supposed to underprice.

Why the Regulator Killed the Cartel and Not the Handshake

China's State Administration for Market Regulation convened a closed-door meeting on January 6 and suspended the buyout plan three days later, citing monopoly concerns after a wave of complaints that producers were using industry coordination as cover to engineer price increases 4 5. That episode is the reason the August pact was built to look nothing like a cartel on paper. A capital pool that pays weaker rivals to shut down invites an antitrust review because it is explicitly about reducing supply and competitors. A cost-accounting standard that simply defines a shared floor price is dressed as compliance with fair-pricing rules rather than collusion on output. Whether regulators ultimately treat the distinction as real or cosmetic is the open question hanging over the entire structure, but for now Beijing has let it stand, and each producer is left to self-certify that its own tender bids clear the floor.

The Cost Math Behind the Floor

The number itself is aggressive. Full production cost across the industry is generally accepted at 45,000 to 48,000 yuan per ton, more than 40% above the roughly 32,000 yuan average transaction price that prevailed before the pact 1 2. That gap is the entire story: it means every ton sold in recent months was sold at a loss, and it means module makers further down the chain, who have already absorbed a run of cost shocks this year, are staring at another one. China ended the VAT export rebate on PV products in stages through 2026, a policy shift that PV Tech reported triggered panic buying and a module price surge as exporters rushed shipments ahead of the change 6. J.P. Morgan had separately forecast polysilicon prices would bottom and rebound in the second half of the year as buyers worked through existing stock and returned to the market 7, a call the Shanghai pact now appears to be pulling forward and steepening.

Who Wins When Silicon Gets More Expensive

The immediate winners are the polysilicon producers themselves, whose balance sheets have been bleeding for three straight years; a floor near full cost is the difference between survival and forced consolidation for the weakest of the eight. The clearer losers, at least in the near term, are wafer, cell and module manufacturers who buy polysilicon as an input and have far less pricing power over their own downstream customers, largely utility developers locked into fixed-price power contracts. Chinese module prices were already expected to firm toward 0.12 dollars per watt in the second half of the year even before this pact 8; a durable 40% jump in the upstream input pushes that trajectory higher still, and the pressure lands hardest on the thinnest-margin module assemblers rather than the vertically integrated giants who signed the agreement.

The Skeptic's Case

The August pact is a pledge, not a regulation, and pledges among fierce competitors racing for market share have a specific failure mode: someone always has a reason to be the one who quietly cuts a tender bid to keep a plant running. The 2025 buyout scheme collapsed at the regulatory stage rather than the market stage, so this is the first real test of whether the underlying incentive to undercut still wins even when producers pledge in writing not to give in to it. There is also a second-order variable outside Beijing's control entirely: Washington is separately weighing its own polysilicon price floor and new tariffs aimed at Chinese solar and chip supply chains, a move that, if it lands, would reshape export economics for these same eight companies regardless of what price they agree to charge each other at home. For investors, the trade is now less about whether China's solar sector has bottomed, and more about whether a cost-accounting standard can do what a monopoly regulator would not let a merger do.