Chapter 2

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.