Daqo Tripled Volume, Sold Below Cash Cost, and Bet the Balance Sheet on 1,000-Volt DC
The volume unlock nobody wanted to need
Daqo's third quarter is, on the surface, a violent operational snap-back. Sales volume went from 15.2 thousand metric tons in Q2 to 46.9 thousand in Q3 — revenue jumped from $62.7 million to $194.2 million on essentially flat prices. That is what happens when a company that had been deliberately starving the market finally decides to sell.
The problem is the arithmetic underneath. Average selling price rose 2% sequentially to $4.12 per kilogram, while average cash cost fell 7.4% to a record-low $4.23 per kilogram. Daqo sold its product for less than its cash cost of making it. Gross loss was $60.6 million, gross margin negative 31% — a huge improvement on Q2's negative 132%, but still a company paying customers to take its output. Inventory provision of $30.9 million (down from $55.7 million) did much of the margin work.
Why the capitulation? Management frames it as strategy: “Following our shift in June from a wait and see stance to a more market oriented sales and pricing strategy.” — Xu Xiaoyu, Management Representative (speaking on behalf of CEO) · 2026-10-27 But the prior calls show exactly how much reluctance preceded it. In April, the CFO described stopping sales outright once price fell under cost: “once price fell below production cost, and then we stopped selling to the market.” — Ming Yang, Chief Financial Officer · 2026-04-29 And in the same call, he laid out precisely the branch the company has now taken:
if the government, despite all the rhetoric, nothing happens... then obviously, we would lower our utilization and then start to sell at close to market pricing, right? whatever it takes to move volume.
They didn't even wait for the utilization cut — they just started moving volume. That is not a bullish tell. It is a company choosing cash preservation over price discipline because the policy it was waiting for didn't arrive on schedule.
The enforcement that keeps being two months away
The anti-involution policy remains the load-bearing beam of this equity. It is also the least reliable. In February, management anchored on a new cost determination "around midyear" and a price floor of at least RMB 53-54 per kilogram. In April: “I understand June is the time line of the new government policy.” — Ming Yang, Chief Financial Officer · 2026-04-29 It is now late October, and the language has softened to a "gradual process" with the target reshaped around Daqo's own all-in cost of roughly RMB 43 per kilogram — with a hoped-for RMB 45-50 range if enforcement bites. Note the drift: the aspiration has fallen roughly RMB 10 per kilogram in eight months. What is on offer instead is rhetoric plus a mechanism. Xu Xiaoyu described punitive actions as fines for sub-floor sellers, and, at the extreme: “the utmost extreme case would be when they just cut off the power for these facilities.” — Xu Xiaoyu, Management Representative (speaking on behalf of CEO) · 2026-10-27 The harder-edged lever is energy consumption standards that phase out outdated capacity from January 1. Meanwhile the thing that actually governs the price — inventory — is worse, not better: “the industry level inventory is really accumulating to a relatively high level to around 700 thousand metric tons overall.” — Xu Xiaoyu, Management Representative (speaking on behalf of CEO) · 2026-10-27The sell below cost problem, in other words, is currently being solved by Daqo selling below cost. That is the tension the sell-side keeps circling and management keeps deflecting.
Three calls, one answer on the buyback
Here is the number that should dominate the conversation: Daqo holds $2.06 billion of readily convertible assets — cash, short-term investments, bank notes receivable, held-to-maturity investments and a $1 billion fixed-term deposit — against a market capitalization of roughly $744 million. Zero debt. 70% nameplate utilization. The equity trades at a fraction of the cash, which management acknowledges: “At the current share price, we are undervalued relative to our balance sheet and also our long term earnings power.” — Xu Xiaoyu, Management Representative (speaking on behalf of CEO) · 2026-10-27 And yet: “before deciding to launch a program, we want to have a more clear view of how the anti involution policy is rolled out and enforced.” — Xu Xiaoyu, Management Representative (speaking on behalf of CEO) · 2026-10-27 This is now the third consecutive call with the same conditional. In October 2025 they were waiting on the size of the consolidation vehicle's initial investment. In February 2026: “we believe it is essential to wait for more clarity on the policy implementation and the outcomes before proceeding.” — Anita Xu, Senior Management (speaking on behalf of Chairman) · 2026-02-26 The $100 million program announced in August 2025 remains, functionally, a press release. Each deferral is individually reasonable; stacked together they read as a board that does not want to commit capital while the thesis is unresolved — which is precisely when a buyback is worth the most.An escape hatch built from bus duct
The genuinely new thing in this report is not polysilicon at all. Daqo is standing up a manufacturing base for AI data center power infrastructure — energy storage, solid state transformers, solid state circuit breakers — targeting the 1,000-volt DC architecture being pushed by NVIDIA and other AI infrastructure builders. AIDC-related is a keyword that did not exist in this company's vocabulary a year ago; it appeared in force last quarter and is now the centerpiece of the forward story.we are also diversifying beyond our core polysilicon business to reduce our exposure to solar PV sector's cyclicality by targeting the fast growing AI data center power infrastructure market.
The financial shape: RMB 1 billion of CapEx expected next year against only about $20 million committed so far, construction complete in the first half of 2027, first sales next year, and at operating cash flow breakeven by end-2027. The margin guide is “gross margin will be around like 15%-20% and the net profit margin will be, like, high single digits.” — Xu Xiaoyu, Management Representative (speaking on behalf of CEO) · 2026-10-27 The near-term product is unglamorous and real: transformers, switchgear and bus duct — integrated solutions, leveraging Daqo Group's four decades of electrical equipment manufacturing.
Two cautions. First, Daqo Group is a related party under common ownership, and when pressed on how that arrangement is reviewed and priced, the answer stayed at the level of "committed CapEx around $20 million" and where the cash sits (~$300 million offshore; roughly RMB 10 billion at Xinjiang Daqo). Second, the market is currently de-rating precisely this theme: the solid state transformer basket (ETN, GEV, POWI, NVTS) is down roughly 7% over 90 days, and AI data center exposure is a 30-day decliner across a wide field of tickers. Daqo is entering a fashionable vertical just as the tape stops paying for it — plausible as a 2027 earnings kicker, unhelpful as a 2026 narrative rescue.
What to watch
Three markers. One, whether Q4 shipments (guided above a 35-45kt production range, with $140 million of customer advances to be delivered against) actually draw down inventory without another ASP give-back. Two, whether the January 1 energy-consumption standard produces real capacity exits rather than another round of conference calls — the difference between a RMB 43 floor and a RMB 50 market. Three, whether a company with $2.06 billion of liquid assets and a $744 million market value can finally convert "a question of timing" into a program. Nine-month operating cash burn of $215 million is the clock ticking in the background; the robust balance sheet buys patience, but patience is not a strategy, and this is the third quarter in which it has been offered as one.