Full Report

Daqo New Energy: a low-cost commodity producer in a deep trough

Daqo New Energy makes one product — high-purity polysilicon, the raw material for solar panels — and sells it at whatever the market will pay. That single fact drives everything else. Revenue swung from $1.7 billion in 2021 to $4.6 billion in 2022 and back down to $665 million in 2025; a $1.8 billion profit in 2022 [1] became a $170 million loss in 2025 [2]. The company enters this report near the bottom of its own price cycle, selling below cost, but debt-free and sitting on roughly $2 billion of cash.

What the company does

Daqo produces polysilicon: the ultra-pure silicon that gets melted into ingots, sliced into wafers, and built into the cells and modules of solar panels. It is a commodity — chemically fungible, priced by global supply and demand, sold mostly to Chinese wafer makers under framework contracts that fix volumes but leave price floating at prevailing market rates [3], Operating Results — p.61"). There is no brand, no switching cost, and no pricing power; the only durable edge a producer can hold is cost.

That is where Daqo has staked its position. Its plants sit in Xinjiang and Inner Mongolia, next to cheap coal-fired electricity — the largest input in polysilicon — which the company credits with making it "one of the lowest cost producers around the globe" [4]. Nameplate capacity reached 305,000 metric tons by late 2024, up from 70,000 tons in 2019, after a decade of debt-and-cash-funded expansion [5].

The structure has two features a new reader should hold onto. First, the NYSE-listed entity, Daqo New Energy Corp., is a Cayman Islands holding company; its shares trade as American Depositary Shares, each representing five ordinary shares, and it files with the SEC as a foreign private issuer [6]. Second, the operating business is not wholly owned: the main subsidiary, Xinjiang Daqo, listed separately on Shanghai's STAR Market in July 2021, and the US holding company owns 72.7% of it [7], [8]. A meaningful slice of the profits, losses, and cash on the consolidated statements therefore belongs to minority holders of the A-share subsidiary — a wrinkle that matters for valuation and recurs later in this report.

The cycle is the business

Polysilicon is a boom-bust commodity, and the last five years delivered a textbook cycle. As solar demand outran supply in 2021, the average selling price ran from roughly $9 per kilogram to a peak above $37/kg in late 2022; a wave of new industry capacity then crushed it back below $5/kg through 2024.

Source: FY2024 20-F, quarterly external sales volume and ASP disclosure [9].

The price moved almost eight-fold peak to trough, and Daqo's income statement moved with it. At the top of the cycle in 2022, gross margin was 74.0% and net income reached $1.8 billion on $4.6 billion of revenue [10]. By 2024 the same business ran a negative 20.7% gross margin, and it stayed there in 2025 [11].

Source: revenue and net income attributable to Daqo New Energy Corp. shareholders, FY2025 20-F (2023–2025) [12], Operating Results — p.60") and FY2023 20-F (2021–2022) [13].

Selling below cost

The defining feature of the present moment is that the market price now sits below what it costs Daqo to make the product. Its full unit production cost was $6.78/kg in 2023, $6.44/kg in 2024, and $6.61/kg in 2025 [14]. The annual average selling price fell to $5.66/kg in 2024 and $5.25/kg in 2025 [15]. Even for a low-cost producer, selling at a loss on every kilogram is the state of play.

Source: FY2025 20-F, annual ASP [16] and production cost per kg [17].

Management's response has been to idle capacity rather than chase volume. Daqo produced 123,652 tons in 2025 against 305,000 tons of nameplate capacity — roughly 40% utilization — and sold 126,707 tons, down from a 200,002-ton peak in 2023 [18]. Running the plants harder would lower the per-unit cost through scale, but it would also add supply to a glutted market and burn more cash; the company has chosen restraint.

Source: FY2025 20-F, Operating Results, Revenue Analysis and Production Cost [19], Operating Results — p.60"), [20], [21].

The balance sheet is the floor

What lets Daqo choose restraint is a balance sheet built during the boom. At the end of 2025 it held $980.3 million of cash and restricted cash plus $1,035.6 million of fixed-term deposits [22] — roughly $2.0 billion of liquid assets — with no bank borrowings and current assets exceeding current liabilities by $2.1 billion [23]. Equity attributable to Daqo shareholders was $4.41 billion, alongside $1.51 billion of non-controlling interest in the listed subsidiary [24].

Source: FY2025 20-F, Liquidity and Capital Resources and Consolidated Balance Sheets [25], [26], Consolidated Balance Sheets — p.115").

The cushion is finite, though. Operating cash flow was negative $435 million in 2024, and while 2025 turned modestly positive at $50 million, the swing came from working capital and lower spending rather than a return to profit [27]. Capital returns have been paused: the board authorized $100 million buyback programs in 2024 and again in 2025 but had bought back none of the holding company's shares under them, choosing to conserve cash through the trough [28]. One nuance for later: much of the liquid balance sits inside the 72.7%-owned A-share subsidiary, so the holding company's own look-through claim on it is smaller than the headline.

Two forces outside the income statement

Two facts shape the case as much as price does. The first is a hard geographic constraint: Daqo's polysilicon is made partly in Xinjiang, and under the US Uyghur Forced Labor Prevention Act, goods made wholly or partly there are barred from US import unless proven free of forced labor. Xinjiang Daqo was added to the Act's entity list in June 2022, which effectively shuts Daqo's material out of the US solar supply chain [29]. The company sells to a global-but-mostly-Chinese customer base, so the direct hit is contained, but it caps the addressable market and is a live risk if Europe or others follow.

The second is a potential turn in the cycle. Management attributes the milder 7.3% ASP decline in 2025 — against 50.7% in 2024 — partly to China's "anti-involution" campaign, a policy push since mid-2025 to curb the ruinous overcapacity and below-cost competition across Chinese industry, polysilicon included [30]. Whether that translates into enforced supply discipline, and durable price recovery, is the swing factor for the whole case.

What the market makes of it

The stock has tracked the collapse in earnings. The ADS is down roughly 80% over five years and trades at about $12, near the low end of a 52-week range of roughly $12 to $37, giving a market capitalization near $0.8 billion. That values the equity at about 0.2 times the $4.4 billion of book value attributable to shareholders, and below the consolidated liquid assets on the balance sheet.

Source: market data, early July 2026 (as reported by market data providers); book value from FY2025 20-F [31].

A price this far below book and near the cash line is the market pricing in a long, uncertain trough — and, for some, the possibility that the balance sheet keeps eroding before prices recover. Neither the discount nor the cash is self-explaining; both need the cycle to be read carefully, which is what the rest of this report sets out to do.

The question this report examines

Daqo is a pure play on the polysilicon price, run by a low-cost producer with an unusually strong balance sheet, caught in the deepest down-cycle of its listed life. The central question this report examines is whether Daqo's position near the bottom of the industry cost curve, funded by a debt-free balance sheet holding roughly $2 billion in cash and deposits, can carry it through an oversupply deep enough that it now sells below production cost — and hold out until industry supply rationalizes and prices recover — before the trough erodes the very balance sheet that is the investment's floor. Everything that follows tests one side of that question: how good the cost position really is, how much cash the trough can consume, how the minority-owned structure splits the value, and what would signal the turn.

---

Cost Curve

Daqo's low-cost claim holds up, with two caveats that matter for the trough. Its cash production cost fell to a record $4.46/kg by late 2025 — genuinely in the industry's bottom tier, built on cheap Xinjiang and Inner Mongolia coal power. But the reported full cost of $6.61/kg is inflated by depreciation on idled plants, and at least one rival — GCL's granular silicon — undercuts Daqo on cash cost. The edge is real and shared, not unique.

This chapter tests one side of the report's central question: how good the cost position really is. The prior chapter (The Polysilicon Cycle) established that Daqo now sells below its full production cost. What follows is why that gap exists, where Daqo actually sits on the global cost curve, and what the curve implies about how the glut clears.

Two costs, and the gap between them

Daqo reports its economics on two bases, and the distinction carries most of the analytical weight in a trough. Cash cost strips out depreciation and share-based compensation; total production cost — $6.61/kg for 2025 — includes them [1]. In a normal year the gap between them is modest. In this down-cycle it is wide, and it is wide for a revealing reason: with the plants running at roughly 40% of nameplate, the fixed depreciation charge is spread across far fewer kilograms. Management attributes the gap to "total idle facility-related costs, primarily non-cash depreciation expenses," which lift the per-unit full cost even as the underlying cash cost falls [2].

The two lines moved in opposite directions through 2025. Cash cost dropped every quarter to a company record of $4.46/kg in the fourth quarter, down from $5.12/kg in the second [3][4]. Total cost fell too — but largely because rising output shrank the idle-facility charge, from $7.26/kg in the second quarter to $5.83/kg in the fourth [5][6].

Source: Q2–Q4 2025 and Q1 2026 earnings calls; the gap between the two bars is chiefly idle-facility depreciation [7][8][9]. Q1 2026 bars are derived from the reported sequential change of +3% cash and +2% total [10].

The practical consequence: the price at which Daqo bleeds cash is far below the price at which it reports a loss. At a market price around $5/kg, Daqo loses money on a full, GAAP basis — the state the prior chapter described — yet each kilogram still clears its cash cost with room to spare. The full-cost figure governs the reported loss; the cash cost governs how long the company can keep running. For a business whose survival case rests on outlasting the trough, the $4.46/kg cash cost is the number that matters, and low utilization has not raised it.

Where the cash cost comes from

Polysilicon made by the modified Siemens process is, before anything else, an electricity product: the process deposits silicon from gas at high temperature and consumes large amounts of power. Daqo's plants sit next to abundant, cheap coal-fired electricity in Xinjiang and Inner Mongolia, which the company credits — together with scale and process work — for making it "one of the lowest cost producers around the globe" [11]. It states plainly that it believes its production cost is "one of the lowest and our product quality is one of the best in China" [12].

Two features make this a cost position rather than a true moat. First, it is location- and process-based, not proprietary: any producer siting a Siemens plant next to the same cheap western-China power can approach the same electricity bill, and the largest Chinese producers have done exactly that. Second, the Siemens route's dependence on power is itself a competitive vulnerability, because an alternative process — the fluidized bed reactor, which grows granular silicon from silane gas — uses markedly less electricity. Daqo names REC, GCL and one joint venture as fluidized bed reactor operators and concedes the method "may have significantly lower production costs" [13]. That concession is not hypothetical.

The rest of the curve

GCL Technology, the second-largest producer, reported first-half 2025 granular-silicon cash cost of RMB26.22/kg — roughly $3.6/kg at prevailing exchange rates, and below Daqo's cash cost in the same period [14]. GCL's advantage has long carried an asterisk: granular silicon historically sold at a discount to Daqo's dense-chunk product on quality, so GCL's lower cost was partly offset by a lower price. That asterisk is fading — by mid-2025 GCL's granular product was trading above traditional N-type dense material for the first time — which makes the cost gap more threatening over time, not less [15]. Its 480,000 MT of granular capacity is larger than Daqo's 305,000 MT [16].

The rest of the industry sorts cleanly by cost, and the high-cost end is already leaving. REC Silicon permanently shut its Moses Lake granular plant and its Butte polysilicon line in 2024, exiting solar-grade polysilicon entirely after persistent quality problems [17]. Wacker, the only Western producer left at scale, saw its solar-grade polysilicon "declined markedly," hurt by "Germany's persistently high energy costs," and is redirecting the division toward higher-margin semiconductor-grade material — targeting a polysilicon EBITDA margin above 30% by 2030 rather than competing on solar tonnage [18][19]. Among Chinese peers, Xinte Energy cut first-half 2025 output by about 77% year over year [20] and ran with prices that "remained below its production costs," a scale of curtailment that mirrors Daqo's own [21].

Sources: GCL H1 2025 results [22]; Daqo Q4 2025 call [23]; Xinte H1 2025 [24]; Wacker FY2024 [25]; REC Silicon FY2025 [26]. Tongwei's polysilicon-only cash cost is not separately disclosed in the corpus.

The pattern that emerges is not the one a cost-leader story usually implies. The producers being forced off the curve — REC, and Wacker in solar — are the Western, high-energy operators, and their exit is already happening. But the tonnage that has to be disciplined to clear the glut sits in the low-cost Chinese tier — Daqo, GCL, Tongwei, Xinte — whose cash costs are all low enough to endure a long trough. Daqo does not out-survive the marginal producer on cost, because its closest rivals are as durable as it is. That reframes how the market rebalances: not through the low-cost leaders bankrupting each other, but through coordinated supply restraint.

What the cost floor actually buys

The first quarter of 2026 showed what the cost floor does and does not do. Prices slid back below production cost, and Daqo — following China's industry self-regulation guidelines — declined to sell below cost. Its sales volume collapsed to 4,482 metric tons against 43,402 tons produced, as it chose to build inventory rather than ship at a loss; the realized price was $5.96/kg [27]. The result was a gross loss of $139.4 million and a gross margin of negative 521%, reversing the small gross profit of $15.4 million booked in the fourth quarter of 2025 [28][29].

That episode is the cost position working as designed. A low cash cost does not deliver profit at a $5/kg market price — it delivers the option to wait: to idle capacity and withhold volume without the cash burn that would force a higher-cost producer to keep selling into the glut. The evidence for the low-cost claim is that Daqo can afford this restraint at all; the strongest fact against reading it as a moat is that its main rivals can afford the same restraint, and one of them makes the product for less. The read would change if granular silicon's quality gap closed enough for GCL's cost edge to translate into durable share gains at Daqo's expense, or — in the other direction — if utilization normalized and pulled Daqo's full cost back toward its cash cost, restoring reported profitability without prices needing to reach today's $6.61/kg full-cost mark.

The line to watch is not the headline loss but the spread between Daqo's quarterly cash cost and its realized price. As long as that spread stays positive, the balance sheet the next chapters examine is being defended, not spent — the condition the whole investment case depends on.

---

Where the Cash Sits

The roughly $2.0 billion of liquidity that anchors the case is real and, through the first quarter of 2026, barely eroding. But for a holder of the New York-listed ADS it is structurally distant: only $311 million sits at the Cayman parent, about $1.7 billion is inside the Chinese operating subsidiaries, $3.23 billion of subsidiary net assets are legally barred from being paid up, and dividends flow only when the subsidiary earns a profit — which it did not in 2024 or 2025 — losing about a quarter to minority holders on the way.

The floor is holding

At December 31, 2025 Daqo held $980.3 million of cash, cash equivalents and restricted cash plus $1,035.6 million of fixed-term deposits, with no bank borrowings and current assets exceeding current liabilities by $2,110.3 million [1]. Three months later the pile was intact: management reported about $2 billion of cash and highly liquid assets at March 31, 2026 and zero debt [2].

Sources: cash and deposits and restricted net assets, FY2025 Annual Report (Form 20-F) [3] [4]; parent cash and minority interest, Schedule I and Consolidated Balance Sheets [5] [6].

The floor is not eroding quickly. Full-year 2025 operating cash flow was a positive $49.7 million despite a $216.1 million net loss, because non-cash depreciation of $236.7 million and share-based compensation of $55.8 million more than covered the loss [7]. The first quarter of 2026 is a warning of how fast a bad quarter can bite: operating cash flow swung to negative $147.5 million as the company stockpiled unsold inventory rather than sell below cost [8]. Even so, headline liquidity held near $2 billion [9]. Inventory converts back to cash when it eventually sells, so the underlying operating burn at a cash cost near $4.5/kg is far smaller than a single quarter's working-capital swing implies. At any plausible pace, group solvency is a question of years, not quarters — which is the real strength of the balance sheet, and the counter to everything that follows.

Most of it is one tier down

The complication is where the money is. Daqo New Energy Corp. is a Cayman Islands holding company; its only material asset is a 72.8% stake in Shanghai-listed Xinjiang Daqo, carried on the parent's own books at $4,097.9 million under the equity method. The parent itself held just $311.2 million of cash at year-end 2025 [10] [11]. The other ~$1.7 billion sits inside the PRC subsidiaries — and $2,066.6 million of the group's liquid assets are denominated in RMB behind China's capital controls [12].

Source: parent cash from Schedule I; subsidiary balance is group liquid assets of $2,015.9M less the $311.2M parent balance [13] [14].

That the auditors required a Schedule I parent-only statement at all is itself the tell: it is mandated only when the restricted net assets of consolidated subsidiaries exceed 25% of consolidated net assets [15]. Here they do — comfortably.

Restricted inside China

The subsidiary cash is not fully fungible with the parent's. Under PRC law the group's Chinese subsidiaries are prohibited from distributing their registered capital, statutory reserves and the proceeds of Xinjiang Daqo's STAR-market IPO and follow-on offering — a restricted portion that amounted to $3,232.8 million at December 31, 2025 [16]. A further $172.3 million is locked in statutory common reserves that PRC entities must build from profits [17]. These figures are net-asset restrictions, not a cash lock-box, and the operating cash inside Xinjiang Daqo still funds the plants day to day; what they constrain is the amount that could ever be moved up to the Cayman parent as a dividend, loan or advance.

The upstreaming tap runs on profits

For cash to reach an ADS holder, Xinjiang Daqo must pay a dividend, and that requires distributable profit it is not currently generating. Its post-IPO policy commits it to distribute at least 30% of average distributable profit over the trailing three years — but only when profitable — and the group recorded no withholding tax on subsidiary dividends in 2024 or 2025 because the listed group made losses [18]. Two frictions apply even when the tap is open: dividends from a PRC subsidiary to a foreign parent carry a 10% withholding tax [19], and because Xinjiang Daqo is only 72.8% owned, roughly 27% of any dividend it declares leaves the group entirely, to its STAR-market minority holders.

The cash flow up to the parent shows the effect. Money reaching the holdco — captured by the parent's own operating cash flow, essentially dividends received plus interest — has collapsed as the subsidiary swung from profit to loss.

Sources: cash up to parent from Schedule I Condensed Statement of Cash Flows [20]; subsidiary dividend to minorities and parent buyback from Consolidated Statements of Cash Flows [21].

The pattern is coherent. In 2023, when the subsidiary was still profitable, roughly $700 million flowed up to the parent and funded a $485.9 million ADS buyback — while $303.7 million of the same dividend cycle left the group in cash to minority holders [22] [23]. By 2025 only $11.7 million reached the parent and the two authorized $100 million ADS buybacks stayed untouched [24]. The paused buyback that earlier chapters flagged is not only caution about the trough; it also reflects that the holdco has $311 million to work with and cannot cheaply pull more from China. Notably, the one repurchase still running is inside China — Xinjiang Daqo bought back $7.8 million of its own STAR shares from minorities in the first quarter of 2026, cash deployed at the subsidiary's much higher multiple rather than against the ADS trading near a fifth of book [25].

What it means for the ADS

The $2 billion protects Daqo the enterprise for years. It is a weaker per-share floor. Of the $5,916.3 million of consolidated equity, $1,509.6 million belongs to Xinjiang Daqo's minority holders, not to ADS holders [26]. Netting the minority's share out of the subsidiary cash leaves ADS holders a look-through claim on roughly $1.55 billion of the liquid assets — the parent's own $311 million plus about 72.8% of the ~$1.7 billion below it — or near $23 per ADS.

Sources: ADS price of $12.25 at July 2, 2026 (market data, as reported); book value and look-through cash derived from equity attributable to Daqo shareholders of $4,406.7M and 67.7M ADS outstanding (338.3M ordinary shares ÷ 5) [27].

The arithmetic cuts both ways, which is why it is worth stating plainly rather than as a verdict. Even after the minority haircut, the look-through cash of about $23 per ADS sits well above the $12.25 price, and the ~$0.8 billion market capitalization is below the group's liquid assets on any measure — so trapped cash and minority leakage do not, by themselves, explain a stock at 0.19 times book. The market is discounting something further: continued below-cost losses that could turn the slow burn faster, the friction and delay in ever getting China's cash to a New York shareholder, and the entity-list and convertibility risks that sit on top. The evidence points to a balance sheet that is a genuine solvency backstop but a soft per-share floor; the main risk to that read is a prolonged trough that forces the parent to inject its $311 million downward instead of returning it. What would change it, in either direction, is Xinjiang Daqo's return to profit — the single event that reopens the dividend tap, the buyback, and the bridge between the cash and the shareholder.

---

The Listing Gap

Daqo's main asset is a 72.8% stake in Xinjiang Daqo, which is itself listed on Shanghai's STAR Market. That market prices the stake at roughly $3.9 billion. The entire NYSE ADS trades for about $0.83 billion. The same operating company is marked near book value by Chinese investors and at 0.19x book by American ones. This chapter measures that gap, sets it against the discounts to cash and book, and asks what it can and cannot mean.

One company, two prices

Daqo New Energy is a Cayman holding company whose only material operation is Xinjiang Daqo, the polysilicon producer. Xinjiang Daqo listed on the Shanghai Stock Exchange's STAR Market in July 2021, and successive share placements there cut Daqo's stake from full ownership to 72.8% as of March 31, 2026 [1]. The result is two live quotes for essentially the same assets: the STAR line 688303, priced in renminbi by mainland investors, and the NYSE ADS line DQ, priced in dollars.

The two markets disagree by roughly five times. On July 1, 2026, Xinjiang Daqo closed at RMB17.98 per share, down about 33% year-to-date, on roughly 2.1 billion shares — a market value near RMB37–38 billion. At the filing's December 31, 2025 translation rate of RMB7.0169 to the dollar [2], that is about $5.3–5.5 billion for the whole subsidiary. Daqo's 72.8% of it is worth roughly $3.9 billion. Meanwhile the ADS holds 338,330,684 ordinary shares — 67.7 million ADS [3] — that at $12.25 on July 2, 2026 capitalize the entire holding company at about $0.83 billion.

Sources: STAR value derived from Xinjiang Daqo (688303) close of RMB17.98 on 2026-07-01 and ~2.1B shares, converted at RMB7.0169 [4]; the 338,330,684 ordinary shares outstanding and parent cash from the FY2025 20-F [5]; ADS market cap from NYSE close of $12.25 on 2026-07-02, as reported.

Set beside the $0.83 billion market cap, the arithmetic is stark: the value of just Daqo's listed stake, at the price a real market is paying for it every day, is about 4.7 times the price of the whole company that owns it — and that is before adding the $311.2 million of cash sitting at the Cayman parent [6]. The conclusion is not sensitive to the loose ends. Move the exchange rate to a current ~7.2, or use the lower end of the reported market cap, and the stake is still worth $3.5–4.0 billion — several times the ADS.

The gap in per-ADS terms

The same divergence, expressed per ADS, shows the NYSE price sitting below every measure of what stands behind it. Book value attributable to Daqo shareholders is $4,406.7 million, or about $65 per ADS [7]. The look-through claim on the group's cash — after the minority and trapped-cash haircuts established in Where the Cash Sits — is about $23 per ADS. The look-through value of the STAR stake plus holdco cash is about $63 per ADS. The ADS trades at $12.25.

Sources: book value of $4,406.7 million and share count from the FY2025 20-F [8]; look-through cash derived in Where the Cash Sits; look-through STAR value from the 688303 quote as above; ADS price NYSE close 2026-07-02, as reported.

That the STAR-based figure (~$63) lands almost exactly on book value (~$65) is not a coincidence — it is the tell. Xinjiang Daqo's own equity is close to $5.55 billion (the $1,509.6 million minority interest is 27.2% of the subsidiary's net assets [9]), so its ~$5.4 billion STAR market value is roughly one times its own book. The mainland market prices the operating company at about book; the NYSE prices the identical company at 0.19x book.

Source: STAR price-to-book derived from the 688303 market value and the subsidiary's implied book (minority interest of $1,509.6M at 27.2% [10]); ADS price-to-book from ADS market cap over attributable equity, as reported.

Why the discount is real but not a free lunch

A 5x gap on identical assets looks like an arbitrage. It is not one, and the reasons it is not are the same forces that explain the discount.

There is no bridge between the two share pools. An ADS holder cannot convert into or redeem Xinjiang Daqo A-shares; China's capital account walls the STAR line off from foreign holders. The filing states the point plainly — holders of the ordinary shares and ADSs "may have limited opportunities to purchase Xinjiang Daqo's shares" even after the STAR listing [11]. You cannot buy the cheap instrument and sell the dear one, so the prices are free to diverge and stay diverged.

The value you do own is hard to move up to the ADS. As Where the Cash Sits established, cash reaches the Cayman parent only as a dividend that requires subsidiary profit, carries a 10% withholding tax, and leaks about 27% to the STAR minorities — and that upstreaming has collapsed to $11.7 million. Restricted net assets of $3,232.8 million are barred from distribution altogether [12]. Daqo cannot sell its 72.8% into the market either: it is a strategic control block, subject to approvals and lock-ups, not a liquid holding it can cash at the screen price. So the $3.9 billion mark is genuine as a valuation but not monetizable at will.

And the two investor bases price different risks. STAR is a retail-heavy market that has long assigned mainland A-shares a premium over their offshore equivalents, so part of the gap is the A-share mark being generous — Xinjiang Daqo is loss-making too, having reported a first-quarter 2026 net loss of about RMB800 million, yet still trades near book. The ADS, by contrast, carries the discounts a global investor demands for this specific structure: the entity-list and UFLPA overhang on Xinjiang polysilicon, US delisting risk, a subsidiary run by a separate board owing duties to its own minorities [13], and the collapsed dividend tap.

The measured read: the ADS discount is not the market saying these assets are worth little. The very same assets are marked near book, in size, by a real market next door. The discount is the price of the wall between an ADS holder and those assets — capital controls, minority leakage, a dormant dividend, and US-specific legal risk. It is a valuation fact, not a closable trade. The strongest argument on the other side is that a wall this durable can justify a permanent discount, and that a look-through value you can neither sell nor upstream is worth materially less than its mark.

What would move it

The gap narrows on anything that lowers the wall. A return to subsidiary profitability would reopen the dividend and, with it, the ADS buyback that has stayed paused through two $100 million authorizations — repurchasing stock at 0.19x book is heavily accretive per share, which is why the pause is the sharper signal (see Where the Cash Sits). A resolution of the delisting and UFLPA overhang would remove a discrete discount the STAR line does not carry. Working the other way: a deeper or longer trough that erodes both books, tighter capital controls, or a forced US delisting would keep the two markets apart, and could pull the STAR mark down toward the ADS rather than the reverse.

What to watch is specific and checkable: whether Xinjiang Daqo returns to a reported profit and resumes upstreaming dividends; whether Daqo restarts the ADS buyback; and the 688303 versus DQ price ratio itself, which prices the wall in real time. The mainland mark says the assets are worth roughly their book. Whether an ADS holder ever captures that is the question this structure leaves open.

---

Waiting on Beijing

Daqo's case rests on polysilicon prices recovering above cost before the trough spends its balance sheet. The past year shows that recovery is a policy outcome, not a market one: prices cleared cost in late 2025 only after Beijing intervened, then relapsed below cost within two quarters when enforcement stalled [1]. Demand offers no independent lift. The live test is whether China's price law is enforced above cost around mid-2026.

The rebound, and the relapse

For most of 2024 and the first half of 2025 the polysilicon market did what a competitive commodity in deep oversupply does — it fell until it was below everyone's cash cost. China spot polysilicon slid from roughly RMB39–45/kg in April 2025 to RMB32–35/kg by the end of June, and Daqo ran its plants at about 34% of nameplate, scaling back sales rather than adding supply [2].

What turned it was not demand clearing the glut but the Chinese state. From late June 2025 the government made "anti-involution" — curbing ruinous below-cost competition — an explicit priority: a People's Daily article on June 29, a July 1 statement by President Xi on regulating disorderly low-price competition, and a July 2 Ministry of Industry symposium with fourteen solar manufacturers [3]. Prices responded: they surged more than 50% from the mid-2025 lows to RMB50–56/kg by year-end [4], and Daqo's annual ASP decline slowed to 7.3% in 2025 after falling 66.7% and 50.7% in the two prior years [5]. Daqo booked a positive gross margin in Q4 2025 for the first time in the trough.

The rebound did not hold. By the end of Q1 2026 spot had fallen back to roughly RMB35–37/kg — below production cost again — and Daqo took a $98.4 million inventory impairment, driving gross margin to negative 521% [6]. The late-2025 recovery, in other words, was the direct product of a policy push, and it reversed when the follow-through did not arrive on schedule.

Spot is the midpoint of reported RMB market ranges converted at ~RMB7.0/USD; "Daqo cost" is FY2025 full production cost, "price floor" is the mooted RMB53–54/kg price-law minimum. Sources: Q2 2025 presentation [7]; FY2025 Annual Report [8]; Q4 2025 call [9]; Q1 2026 call [10].

What enforcement would take

Management has been explicit about the mechanism it is waiting on. Under the amended Pricing Law, sales "should not be below the industry level cost," and Daqo's read is that the floor would sit at "at least RMB53–54 per kilogram" — roughly $7.6/kg, and about 15% above its own $6.61/kg full cost [11]. If that floor were set and enforced, Daqo would return to reported profit and, as one of the lowest-cost producers, expect to regain share. That is the bull mechanism in a sentence.

The obstacle is the distance between rhetoric and enforcement. The policy scaffolding is real and building: Beijing has formally designated anti-involution a national priority in the 15th Five-Year Plan, revised the Anti-Unfair Competition Law and drafted a Pricing Law amendment, and proposed a mandatory national standard imposing strict per-unit energy-consumption limits on polysilicon production — an instrument that would fall hardest on high-cost, high-energy producers rather than on Daqo [12]. But as of the Q1 2026 call the price had not yet been set: a fresh government "round of price determination" was expected around mid-2026, with the enforcement method still under discussion and Daqo, in its own words, "in observation mode" [13]. The tell is in the futures curve: at the time of the Q4 2025 call the polysilicon future traded at RMB46.3/kg — near Daqo's full cost, and well below the RMB53–54 the company hoped policy would deliver [14]. The market, in short, was not pricing enforcement.

The demand ceiling

Even a policed floor would only stop prices falling; it does not create the demand that clears a 600,000-tonne inventory mountain. By the end of Q1 2026, industry polysilicon inventory had climbed to nearly 600,000 metric tons — with Tier-1 producers alone holding at least three months of stock — even as industry monthly supply fell to about 93,000 tonnes at just 39% utilization [15][16]. Supply has been cut hard; the overhang persists because demand is not growing fast enough to absorb it.

The demand picture is one of decelerating growth, not decline. Global solar installations reached roughly 580 GW in 2025 — a record — but growth slowed to about 9% after rising in the mid-30s-percent range in the prior two years [17]. And the demand that exists is increasingly policy-timed rather than organic. China added a record 93 GW in May 2025 alone — then just 14 GW in June — as developers rushed to beat a May 31 cutoff for market-based pricing, pulling demand forward and leaving an air-pocket behind it [18].

Annual growth decelerated from roughly +36% (2024) to +9% (2025). Source: FY2025 Annual Report, Item 3 Risk Factors [19].

Two structural headwinds sit on top of the deceleration. First, China is removing the export subsidy that props up its module makers: the value-added-tax rebate on exported solar products falls from 9% to 6% on April 1, 2026 and is abolished entirely from January 1, 2027 [20]. Polysilicon itself is sold under a 13% VAT with no refund, so the cut does not hit Daqo directly — it bites indirectly, by raising the delivered cost of Chinese modules abroad and dampening the export demand that pulls polysilicon through the chain [21]. Management flagged exactly this dynamic on the Q1 2026 call: a January–February export rush to beat the rebate deadline, then downstream buyers holding off on polysilicon purchases amid uncertainty about demand after April 1 [22]. Second, the entity-list and UFLPA barriers established earlier (The Polysilicon Cycle) keep Daqo's Xinjiang-made polysilicon out of the largest premium-priced end market, so the US installation base does little for its order book.

How Daqo is playing it, and what to watch

Daqo has turned its low cost into a waiting strategy rather than a share-grab. Through the trough it has repeatedly declined to sell below cost: in Q1 2026 it sold only 4,482 tonnes against 43,402 produced, recognizing a $5.96/kg ASP on that thin volume while holding the rest as inventory in anticipation of a policy-led recovery [23]. It has tied its production plan directly to enforcement: guidance is to hold utilization at 50–55%, but management has said plainly that if the price law is not enforced and rivals keep selling below cost, "then we will lower our utilization" and sell closer to market [24]. The company's own volumes have become a live signal of whether it believes the recovery is real.

That framing yields a short list of checkable markers, each with a filing line and a threshold:

Source: synthesized from Q1 2026 and Q4 2025 earnings calls and the FY2025 Annual Report, as cited above [25].

The measured read: the recovery the whole case waits on is real but conditional, and the condition is policy rather than the cost curve doing its work unaided. The strongest fact for the bulls is that the mechanism has already fired once — prices jumped more than 50% in H2 2025 when Beijing pushed [26], and Daqo briefly turned gross-margin positive [27] — so this is a demonstrated lever, not a hope. The strongest fact against them is that it reversed within two quarters, that a 600,000-tonne inventory overhang and decelerating demand give the free market no help, and that enforcement kept slipping into "observation mode." What would settle it is the mid-2026 price determination: a credible above-cost floor with teeth turns Daqo's balance sheet from a defensive buffer into a return; another lapse leaves the company idling plants and drawing down cash while it waits for the next push.

---

Capital Allocation

Daqo is a family-controlled company that has run its balance sheet through the cycle with a growth reflex. It spent roughly $2.7 billion building new capacity into the price collapse — one plant that started up in 2024 has still not been given a ramp-up schedule "due to adverse market conditions" — and it has left two authorized $100 million buyback programs entirely unused while the ADS trades at 0.19x book, choosing to reserve capital for industry consolidation and a fresh RMB6 billion diversification into data-center power.

Who decides

Control sits with three generations of one family. Xiang Xu has been both chairman and chief executive since August 2023, taking the combined role from his father, founder Guangfu Xu; his daughter Xiaoyu Xu, a former J.P. Morgan banker, became deputy chief executive in October 2024 [1]. The directors affiliated with Daqo Group — the two Xus, Xiaoyu Xu, Dafeng Shi and Fei Ge — together beneficially own 35.6% of the ordinary shares, a block the filing says gives them "substantial influence" over mergers, asset sales, board elections and other significant corporate actions, and whose interests "may not always be aligned" with those of other shareholders [2].

Source: FY2025 Annual Report (Form 20-F), Item 6.E Share Ownership [3].

The board is a majority of independents — six of eleven directors — and an audit committee reviews related-party dealings [4]. Two features temper how much comfort a minority holder should take from that. First, disclosure is thin: as a foreign private issuer Daqo reports executive pay only in aggregate — about $2.4 million for all directors and officers combined — with no individual figures and no CEO pay ratio. Second, the founding family's ownership at the listed company is a minority of the votes, so alignment runs to the enterprise, not automatically to the per-ADS value a New York holder cares about — the distinction that the earlier chapters on cash location (Where the Cash Sits) and the cross-listing (The Listing Gap) turned on.

The record: building into the fall

The clearest read on management is what it did with capital during the boom. Between 2021 and 2024 Daqo more than tripled nameplate polysilicon capacity, from 105,000 tonnes to 305,000 tonnes, through the Phase 5A and 5B plants in Baotou, Inner Mongolia [5]. That build cost about $2.7 billion of property, plant and equipment, concentrated in 2022 and 2023 — the very years the polysilicon price was rolling over from its Q4 2022 peak of $37.41/kg toward the $4.62/kg trough of Q4 2024.

Sources: FY2025 Annual Report (Form 20-F), Consolidated Statements of Cash Flows, for 2023–2025 [6]; FY2023 Annual Report (Form 20-F), Consolidated Statements of Cash Flows, for 2021–2022 [7].

The Phase 5B plant, a 100,000-tonne line, began production in May 2024 straight into the glut [8]. By the time of the FY2025 report it still had no ramp-up schedule: the company states it "currently do[es] not have an expected ramp-up schedule for our Phase 5B facilities or our polysilicon project for the semiconductor industry, primarily due to adverse market conditions" [9]. Part of that expansion was funded by a July 2022 sale of Xinjiang Daqo shares on the Shanghai STAR Market that raised about RMB11 billion but diluted the Cayman parent's stake in its own operating subsidiary to 72.8% [10] — the minority leakage that the cash and listing chapters already priced.

The fairer reading is not that management is reckless. Those Phase 5 approvals were made when the price still looked durable, and low-cost capacity is precisely the asset that lets Daqo wait out the trough it helped create. The discipline showed up once the cycle turned: capex fell to $359 million in 2024 and $173 million in 2025, and Phase 5B was left idle rather than force-ramped. But the pattern is a growth reflex — add tonnes when cash is plentiful — and the industry-wide version of that reflex is what produced the oversupply the whole investment case now waits on.

The buyback that hasn't happened

The forward choice is sharper. Daqo's board authorized a $100 million repurchase program in July 2024 and a second $100 million program in August 2025. Neither has been used at all [11]. The company has never paid a dividend on its ordinary shares or ADSs [12].

That inaction is recent, not habitual. In 2023, when the operating subsidiary was still profitable and paying cash up to the Cayman parent, Daqo repurchased $485.9 million of its own ADSs. As the trough deepened, ADS repurchases collapsed to $5.0 million in 2024 and to zero in 2025 [13].

Source: two $100M programs authorized July 2024 and August 2025, none used [14]; price/book per prior chapters.

Source: FY2025 Annual Report (Form 20-F), Item 5 financing-activities discussion; $200M authorized across the 2024 and 2025 programs, $0 used [15] [16].

The timing is what makes the pause notable. The ADS now trades below its own look-through cash and at roughly a fifth of book — the cheapest the stock has been in its listed life, and by the arithmetic of a 0.19x-book price the most accretive moment to retire shares. Management does not dispute the value. On the Q4 2025 call it said it sees "tremendous value and intrinsic value in our shares," but judged it "essential to wait for more clarity on the policy implementation and also the outcomes before proceeding" — a "wait-and-see stance" to "optimize the timing" of the program [17]. A quarter earlier the explanation was more concrete: the repurchase was "paused pending clarity on the capital required for industry consolidation" [18], and management added that after it announced the program the share price had risen to $31 — about 35% above late-August levels — so it would begin buying "after we have a more clear picture of what the consolidation looks like" [19].

Two readings compete here. The charitable one is that reserving dry powder for a consolidating industry — where a low-cost, debt-free balance sheet could buy distressed capacity cheaply — may create more value than shrinking the share count. The skeptical one is that the group did keep buying stock through the trough, just not the cheap one: Xinjiang Daqo continued to repurchase its own STAR shares, at a mark near book value, spending $7.7 million in 2024 and about $0.9 million in 2025 [20] [21]. Capital returned inside China defends the A-share mark; the far cheaper ADS goes unbought.

The new growth engine

While the buyback waits, a new use for capital has appeared. In June 2026 Daqo signed an agreement to build a manufacturing base in Kunshan for energy solutions aimed at AI data centers — storage systems, solid-state transformers and batteries — with a first phase of about RMB2.1 billion and a total of roughly RMB6.0 billion, per company announcements. Management has begun framing this on its calls, describing data-center power and even space-based solar as "a significant new growth engine for the sector" and casting the balance sheet as the tool to "capitalize on the market recovery and these long-term growth" opportunities [22].

For a company whose entire investment case rests on a cash cushion of about $2 billion outlasting a polysilicon trough, a multi-billion-renminbi push into an unrelated business is the capital-allocation event to watch. It is early and small relative to the balance sheet, and diversification away from a single commoditized product has a logic. But it points the same direction as the Phase 5 build: capital flowing to new capacity and new markets rather than to the discounted equity, at a moment when the equity is the cheapest asset management could buy.

The read this chapter lands on: management has been a capable operator and a genuine defender of the enterprise balance sheet — it stopped spending when the cycle turned and refuses to sell polysilicon below cash cost — but its allocation instincts run to tonnage and new ventures over per-ADS value. The strongest fact against that read is that Daqo did execute a large ADS buyback in 2023 when cash was available, so the machinery exists and can restart. What would change the read is simple and falsifiable: the company using even a fraction of its $200 million of authorized buybacks at a fifth of book, or channeling a recovery in subsidiary earnings up to the ADS rather than into the next growth project.

---

Carrying Value

Property, plant and equipment is the single largest thing Daqo owns — about $3.4 billion at the end of 2025, more than half of both its assets and its book value. That carrying value is not a fixed floor. Daqo wrote $175.6 million off its plants in 2024 as polysilicon prices fell, avoided a further charge in 2025 only because prices rebounded, and books the rest on a recoverability test that a sustained relapse below cost would re-arm. The cash floor is harder than the book floor.

The book value is mostly plant

The balance sheet that earlier chapters treated as a floor is not all cash. Of $5,916.3 million of consolidated equity at end-2025, gross property, plant and equipment nets to $3,399.1 million after $971.0 million of accumulated depreciation and $177.3 million of cumulative impairment [1]. That plant is 52.7% of total assets and 57.5% of total equity [2]. The roughly $2.0 billion of cash and deposits examined in Where the Cash Sits is the other half — and the two halves behave very differently in a trough.

Source: FY2025 Annual Report (Form 20-F), property, plant and equipment note [3] and Consolidated Balance Sheets [4]; ratios derived from those figures.

This matters because the ADS is valued on book. The listing gap in The Listing Gap is a discount to a book value of $4,406.7 million attributable to DQ shareholders [5], and the largest input to that book value is the carrying amount of plants running near 40% of nameplate. Whether that carrying amount holds is an accounting-quality question, not a market one.

The 2024 write-down, and the one that did not happen in 2025

Daqo has already marked its plants down once in this cycle. It recognized a $175.6 million long-lived-asset impairment in 2024 — nothing in 2023, nothing in 2025 [6]. The charge was "mainly related to our older polysilicon facilities," and management attributes it to "the continuous downward trend of the polysilicon selling prices" impairing the recoverability of those assets [7].

The 2025 result is the more revealing number. Impairment fell "from $175.6 million in 2024 to nil in 2025, mainly because the polysilicon selling prices rebounded substantially in 2025, which enhanced the recoverability of the carrying amount of our long-lived assets" [8]. In other words, the plant carrying value survived 2025 not because operations improved but because the H2 2025 policy-driven price rebound described in Waiting on Beijing lifted the forecast enough to clear the test. The same lever that turned the reported loss less negative also kept the book value intact.

Inventory tells the parallel story on the current-asset side. The write-down against inventory ran $4.5 million in 2023, jumped to $81.4 million in 2024, and eased to $21.5 million in 2025 as prices recovered [9]. Then, in the first quarter of 2026, the relapse below cost forced a fresh $98.4 million inventory impairment, and gross margin was "negative 521% compared to 7% in the fourth quarter of 2025" [10]. Inventory, carried at the lower of cost or net realizable value [11], re-marks every quarter; the plant does not, but it is tested on the same prices.

Source: FY2025 Annual Report (Form 20-F), Item 5 impairment discussion [12] and Consolidated Statements of Cash Flows [13]. The Q1 2026 inventory charge of $98.4M is not shown.

Why a price relapse re-arms the test

The reason the carrying value moves with prices is the mechanics of the impairment test. Daqo evaluates its plants for impairment "whenever events or changes in circumstances (such as a significant adverse change to market conditions)" indicate the carrying amount may not be recoverable, comparing that carrying amount to "future undiscounted cash flows" and, only if those fall short, writing the asset down to fair value based on discounted cash flows [14]. The fair-value estimate "incorporates significant assumptions including revenue growth rate, gross margin, and discount rate," which "might be affected by expectations about future market and economic conditions" [15]. Those are the same variables that Waiting on Beijing showed hinge on whether Beijing enforces an above-cost price floor by the mid-2026 cost re-determination.

The undiscounted-cash-flow first step is a genuinely high bar to fail — an asset is impaired only when the sum of all future undiscounted cash flows over its life falls below carrying value, so low-cost, long-lived plant clears it far more easily than a mark-to-spot would. That is why Daqo could avoid a 2025 charge on a modest price recovery. But the bar is not immovable: the older facilities failed it in 2024, and Daqo has done this before — in the prior down-cycle it "incurred fixed asset impairment charges related to the Phase 1 polysilicon facilities" in 2013 [16]. A relapse to sustained below-cost pricing pulls down the revenue and gross-margin assumptions that the recoverability of the older plants depends on, and it re-tests the newest capacity — the $2.7 billion Phase 5 build examined in Capital Allocation, whose Phase 5B line is idle with no ramp schedule — as it enters the depreciable base.

The subjective levers: useful lives and the tax asset

Two accounting judgments sit under the plant carrying value, and both lean toward keeping it high.

The first is depreciable life. Depreciation of plant rose from $148.9 million in 2023 to $206.4 million in 2024 to $236.7 million in 2025 [17], and management projects a further increase in 2026 "driven by an increase in our projected depreciable asset base" — Phase 5B moving from construction into service [18]. The sensitivity is large: shortening the estimated useful lives of all depreciable assets by five years would raise annual depreciation by roughly $248.8 million, while lengthening them five years would cut it by about $67.0 million [19]. Against $665.4 million of revenue, a $248.8 million swing in a single non-cash line is material — and this connects to the full-versus-cash-cost gap in Cost Curve: the reported full cost is inflated by exactly this depreciation, spread over idle capacity.

The second is the deferred tax asset. Daqo carries $115.7 million of net deferred tax assets at end-2025, including a net-operating-loss carryforward that grew from $38.0 million to $78.2 million and an impairment-related asset of $31.1 million, against a valuation allowance of just $5.0 million [20]. Recognizing those assets almost in full — after two consecutive loss years — embeds a management judgment that Daqo will earn enough future taxable profit to use them. That is the same bet on recovery the impairment test makes; a rising valuation allowance in a future filing would be management's own signal that it doubts it.

Source: FY2025 Annual Report (Form 20-F), Critical Accounting Estimates [21] and Income Taxes note [22].

What the auditor flagged

The independent auditor's one critical audit matter is not impairment. It is revenue cut-off for domestic polysilicon sales — the risk of "inconsistency between the date of customer acceptance and the date of revenue recognition" — which the auditor singled out as an "especially challenging, subjective, or complex" judgment requiring extensive procedures [23]. That the plant carrying value was not elevated to a critical audit matter — after a $175.6 million charge the prior year — is a mark in Daqo's favor: an independent auditor did not judge the 2025 recoverability conclusion to be the hardest call in the file. Revenue cut-off is worth noting on its own — timing recognition around period-end is where a commodity producer under volume pressure has the most discretion — but the filing gives no evidence it was mishandled, only that it is where the audit spent its effort.

The read

The evidence points to a book value that is sound today but softer than the headline equity figure implies, and soft in a specific, checkable way. Roughly $2.0 billion of the equity is cash and deposits that no impairment can touch — impairments are non-cash and leave the solvency floor from Where the Cash Sits intact. The other ~$3.4 billion is plant whose carrying value is a direct function of the polysilicon price path: it was cut $175.6 million in 2024, held flat in 2025 only because prices rebounded, and would be re-tested if they relapse. The strongest fact against reading this as fragile is the undiscounted-cash-flow test itself, which low-cost long-lived assets pass comfortably, plus the auditor's decision not to flag it — the 2025 book value is a defensible conclusion, not an aggressive one.

What would change the read: a return to sustained below-cost pricing through the mid-2026 cost re-determination, which re-arms the test on the older facilities and on Phase 5B as it enters service; a fresh long-lived-asset impairment or a rising valuation allowance against the deferred tax asset in a 2026 filing, either of which is management conceding the recovery assumptions have weakened; or, on the other side, an enforced above-cost floor, which would lift the very forecasts that keep the plant — and the tax asset — whole. The book-value floor and the price recovery are not two separate questions; they are the same one, read off the balance sheet.

---

The Value Range

At $12.25, the DQ ADS trades below the cash the parent looks through to and at roughly a fifth of book. The downside is bounded by a hard, impairment-proof pool of about $2.0 billion in cash and deposits; the upside — a re-rating toward book or the Shanghai listing's mark — and the soft $3.4 billion plant value both depend on Beijing enforcing an above-cost polysilicon price by mid-2026. This chapter reconciles the report's pillars into that range and names what to watch. It is scenarios, not a verdict.

What the price implies

The arithmetic is stark. Daqo has 338,330,684 ordinary shares outstanding — 67.7 million ADS at five shares each [1]. At $12.25 that is an $0.83 billion equity value against $4.41 billion of equity attributable to the ADS and $5.92 billion of total equity including the minority [2]. Book value is about $65 per ADS; the market pays 0.19 times it.

Source: ADS price as reported, July 2, 2026; shares, equity and book value from the FY2025 Annual Report, Consolidated Balance Sheets [3]; price-to-book derived.

Four reference points frame the range. The price sits below the cash the ADS looks through to, well below both book and the value the Shanghai STAR market puts on the same operating company, and near the very bottom of what a liquidation-style read would support. Each anchor was built earlier in this report; the point here is that they bracket a wide gap.

Sources: price as reported; look-through cash per Where the Cash Sits; look-through STAR mark per The Listing Gap; book value from the FY2025 Annual Report [4].

Two floors, hard and soft

The floor under the ADS is not one number but two, and they behave differently in a prolonged trough — the distinction Carrying Value drew. The hard floor is the liquidity: $980.3 million of cash and restricted cash plus roughly $1.0 billion of fixed-term deposits, no bank borrowings, and a working-capital surplus of $2.19 billion at year-end 2025 [5]. That pool cannot be impaired by a lower polysilicon price. The soft floor is property, plant and equipment carried at $3.40 billion — 52.7% of the $6.45 billion balance sheet [6]. Its carrying value is price-contingent: the FY2025 impairment was nil only because prices rebounded, after a $175.6 million charge the year before.

Source: liquidity and PP&E from the FY2025 Annual Report, Consolidated Balance Sheets [7]; ADS equity value as reported.

Both floors sit above the ADS price, but neither is fully available to an ADS holder. The hard cash is largely trapped: only $311.2 million of it is U.S.-dollar cash at the Cayman parent, $3.23 billion of consolidated net assets are legally barred from distribution, and the upstreaming tap has collapsed with subsidiary losses [8] [9]. That is why a discount to book and to look-through cash can persist without being an arbitrage, as Where the Cash Sits and The Listing Gap established. The floors bound the downside; they do not by themselves close the gap.

The hinge: an above-cost price by mid-2026

What turns the balance-sheet option into a return is a polysilicon price above Daqo's cost, and in this cycle that is a policy outcome rather than a market one. In FY2025 the average selling price was $5.25/kg against a full production cost of $6.61/kg — a full year sold below cost [10] [11]. The first quarter of 2026 was worse: spot fell back below cost, Daqo took a $98.4 million inventory write-down, gross margin was negative 521%, and the net loss reached $88.4 million, or $1.31 per ADS [12]. Rather than sell into that, management withheld product — sales collapsed to 4,482 metric tons against 43,402 produced — to comply with the authorities' self-regulation guidelines [13].

Sources: ASP and cost from the FY2025 Annual Report [14] [15]; Q1 2026 ASP from the Q1 2026 call [16]; mooted floor per Waiting on Beijing.

The mechanism is now written into national policy. Anti-involution is a stated priority of China's 15th Five-Year Plan, the competition and pricing framework has been amended, and a proposed mandatory standard would cap energy consumption per unit of polysilicon — a rule that favors a low-energy producer [17]. On the Q1 call, management described a government consensus to enforce the price law, a fresh round of cost determination expected around midyear, and renewed minimum-price guidance to follow — but enforcement itself had not begun, leaving Daqo in what it called "observation mode" [18]. The mooted floor sits above Daqo's cost, as Waiting on Beijing detailed; if it is set there and enforced, Daqo returns to profit, and if it is not, the below-cost wait continues. The value range is most sensitive to which of those happens.

Three scenarios

The pillars combine into a spread that is asymmetric by construction: the downside is anchored to cash that cannot be impaired, while the upside is anchored to book and to the Shanghai mark. What separates them is enforcement and time.

Sources: cost floor [19]; utilization guidance [20]; enforcement path [21]; value anchors per Where the Cash Sits, The Listing Gap and Waiting on Beijing.

Two features of the downside are worth holding onto because they cut against the bear. First, even in a below-cost year the enterprise did not bleed: FY2025 operating cash flow was a positive $49.7 million, cushioned by $236.7 million of depreciation and non-cash write-downs, so the hard cash floor erodes slowly rather than abruptly [22]. Second, management has held utilization at 50-55% and can drop it further to cut the burn, as it has before [23]. The strongest fact against the upside is that the trigger is not in the company's hands — enforcement has been discussed, not delivered — and that the roughly 600,000 metric tons of industry inventory would cap any price rebound even if enforcement arrives [24].

There is a second condition on the upside that has nothing to do with polysilicon. A return to subsidiary profit only reaches the ADS if management chooses to send it there. Two $100 million ADS buyback programs sit unused at 0.19 times book while capital is reserved for consolidation and a new data-center venture [25]. As Capital Allocation argued, recovered cash could as easily fund tonnage or diversification as the discounted ADS. The upside case therefore needs two things to break the same way: policy, then priorities.

What to watch

Each of these is falsifiable, appears in a specific disclosure, and would move the read. They are the checkpoints between now and a resolved thesis.

Sources: enforcement and utilization from the Q1 2026 call [26] [27]; impairment, deferred-tax and buyback items per Carrying Value and Capital Allocation.

The report's spine has been whether a low-cost, debt-free balance sheet can carry Daqo through a trough deep enough to sell below cost until prices recover. The evidence points to a genuine floor and a genuine option, priced as though neither will pay off: at $12.25 the ADS is below its own look-through cash, and the assets behind it are marked near book by the same investors who own the Shanghai listing. What that option is worth turns on a policy decision expected by mid-2026 and on whether management then routes the proceeds to the discounted ADS. Until the mid-2026 cost re-determination lands and the first post-enforcement quarter is reported, the range stays wide — bounded below by cash, above by book, and unresolved in between.

---

Cash Quality

The investment case rests on a roughly $2 billion balance-sheet floor, so the first question a skeptic asks is whether that cash — and the revenue behind it — is real. On the evidence it is: Daqo sells a single product for payment in advance, carries no trade receivables, and drew a clean internal-control opinion with only a revenue-timing item flagged by its auditor. But the floor's provenance and trajectory matter more than its size. Close to $1 billion of it was minted by monetizing boom-era receivables in 2023, a reservoir now drained, and 2025's positive operating cash flow was almost entirely depreciation of idle plant, not economic profit.

How the revenue is booked

Daqo's revenue is unusually hard to inflate, which is not something one can say of most companies selling into a glut. The company requires advance payment before it ships, so it carries essentially no trade accounts receivable — the balance was nil at the end of 2021 and 2022, and the line does not appear on the 2024 or 2025 balance sheet at all [1]. What is not prepaid is generally settled in bank acceptance drafts — bank-guaranteed instruments, not open customer credit [2]. The usual channel-stuffing playbook — booking sales on generous credit to customers who may never pay — has little room to operate here.

Almost all of that revenue is domestic and recognized at a single point in time. Domestic sales were $660.0 million of $665.4 million total in 2025 — 99.2% — with exports a rounding error at $5.4 million [3]. Product is a fungible commodity, returns are immaterial, and there are no warranties, rebates, or incentive arrangements to estimate [3].

That leaves timing as the one judgment the auditor singled out. Deloitte — Daqo's auditor since 2008 — reported a single critical audit matter: revenue cut-off for domestic polysilicon sales, "because there exists the risk of inconsistency between the date of customer acceptance and the date of revenue recognition" [4]. The procedures were the standard cut-off battery — tracing samples of shipments on both sides of year-end to receipt notes signed by customers — and the auditor issued an unqualified opinion on the financial statements and on internal control [4]. The matter is about which period a genuine, prepaid-for shipment lands in, not whether the sale or the customer exists. For a company whose revenue is essentially one product recognized on delivery, a cut-off CAM is close to the default; its presence is a reason to watch quarter-boundary timing, not evidence of a problem.

Where the balance-sheet floor came from

The cash pile that anchors the whole case is real, but it was largely built by draining working capital accumulated in the boom rather than by steady operating generation. The clearest window is notes receivable — the bank acceptance drafts Daqo takes in partial settlement. That balance swelled to $1,131.6 million at the end of 2022 as the boom peaked, then was monetized down to $116.4 million a year later [5].

Source: Consolidated balance sheets, FY2022 and FY2025 Annual Reports (Form 20-F) [5][2].

That drawdown is why 2023 operating cash flow of $1,616.0 million towered over net income of $652.9 million: a single working-capital line, the release of notes receivable, contributed $987.5 million — 61% of the year's operating cash and 1.5 times net income [6]. It was cash the company had already earned in 2022 and was now collecting, not new operating profit. With the reservoir down to $135.5 million, that lever cannot be pulled again at anything like the same scale.

Source: Consolidated statements of cash flows, FY2025 Annual Report (Form 20-F); net income is consolidated, including non-controlling interests [6].

Over the full 2021–2025 cycle, the picture is reassuring in aggregate: cumulative operating cash flow of about $4.33 billion against cumulative net income of roughly $3.33 billion, a conversion ratio of 1.3 [6]. Reported earnings did turn into cash. The catch is that the conversion happened at the top of the cycle, and the mechanisms that drove it — the note release and, on the funding side, the follow-on offerings covered elsewhere — are behind the company, not ahead of it.

What "positive operating cash flow" was in 2025

Management highlighted that Daqo "generated positive operating cash flow in 2025" — $49.7 million on cash received from product sales of $753.9 million [7]. The figure is accurate, but its composition is the point: it exists because a loss-making business adds back the depreciation of plant it is barely running.

Source: Consolidated statements of cash flows, FY2025 Annual Report (Form 20-F); non-cash add-backs and working-capital change grouped from reported reconciling lines; free cash flow = operating cash flow less purchases of property, plant and equipment [6].

Depreciation of property, plant and equipment alone was $236.7 million in 2025 — nearly five times the reported operating cash flow — and it is rising as idle capacity, including the Phase 5B lines, enters the depreciable base [6]. Strip out non-cash charges and the $49.7 million is a thin residue on top of a large depreciation add-back, not evidence the business is throwing off cash. After $173.0 million of capital spending, free cash flow was about negative $123 million [6]. The floor held in 2025 because roughly $2 billion sits in deposits and the burn is slow — not because operations funded themselves.

One working-capital line did most of the work of keeping the headline positive. Notes payable to third parties — supplier financing Daqo arranges through bank drafts, backed by restricted cash at the same banks — rose from $6.8 million to $103.0 million [8], a $98.8 million operating inflow [6]. That single swing is larger than the year's entire reported operating cash flow; without it, 2025 operations would have consumed cash. It is disclosed and collateralized, so this is a note on the quality of the headline, not an allegation — but it means "positive operating cash flow" leaned on a step-up in supplier financing that cannot repeat indefinitely.

The demand signal in the advances, and the 2026 reversal

The same accounts that make revenue clean also carry a demand signal. Advances from customers — prepayments made to lock up supply — are a barometer of how badly buyers want the product. They stood at $293.6 million at the end of 2021, when polysilicon was scarce, and had fallen to $62.0 million by the end of 2025 [9][10]. In a glut, no one prepays for a commodity they can buy on demand; the collapse in advances is the mirror image of the pricing chapter, read from the balance sheet.

Source: Revenue recognition notes, FY2022 and FY2025 Annual Reports (Form 20-F); total of short- and long-term contract-liability balances [9][10].

By the first quarter of 2026 the cash quality had turned decisively. Operating activities used $147.5 million of cash, against $38.9 million in the prior-year quarter, and gross margin was negative 521% after a $98.4 million inventory write-down as period-end prices sat below production cost [11][12]. The thin positive of 2025 was a function of a particular quarter's prices and a supplier-financing inflow; when prices relapsed, the burn resumed. This is consistent with the floor eroding slowly rather than the business having found a self-funding footing.

The read

The reported numbers hold up on the tests that usually catch trouble. Cash comes in before product goes out, so there are no trade receivables to inflate; a single commodity recognized on delivery leaves little estimation latitude; the auditor found internal control effective and flagged only cut-off timing; and across the full cycle, earnings converted to cash at 1.3 times. The ~$2 billion floor is genuine cash, not an accrual illusion — which is the single most important thing it needs to be, because it is the downside of the whole case.

Where the caution lies is provenance and direction, not honesty. That cash was largely harvested from the boom — nearly $1 billion of it from monetizing 2022's notes receivable in 2023 — and those reservoirs are now near-empty. The 2025 "positive operating cash flow" was depreciation of idle plant plus a step-up in supplier financing, and by early 2026 operations were consuming cash again. The floor is a tank being drawn down, not a spring being refilled. What would sharpen the read in either direction is specific and checkable: a trade-receivable balance appearing on the balance sheet or a rising credit-loss allowance on notes receivable would signal Daqo is extending credit to move volume; continued reliance on growth in notes payable to keep operating cash positive would show the headline is being propped; and a return of customer advances would be the earliest balance-sheet sign that demand — and pricing power — is coming back.