Daqo (DQ) Q2 2026 Earnings Call Transcript

Daqo New Energy reported Q2 2026 revenue of $62.7M (up from $26.7M in Q1) but still generated a net loss of $81.2M (vs. $88.4M in Q1), with gross loss narrowing to $82.7M. Gross margin remained deeply negative (−132%) as polysilicon ASP fell to $4.04/kg, though inventory impairment provisions declined to $55.7M from $98.9M. Management cited easing pricing pressure as China’s anti-involution/energy consumption standards roll out (6.3 kgce/kg effective Jan. 1, 2027) and reiterated a market-oriented sales strategy plus a new AIDC power-infrastructure expansion (RMB 2B Phase 1). Liquidity stayed strong at ~$1.92B and the company guided Q3 2026 polysilicon production of 40,000–45,000 metric tons.

Analysis

The market is likely to overweight the policy narrative and underweight the inventory math. A durable recovery in DQ depends less on headlines about “self-discipline” and more on whether downstream buyers actually clear stocks fast enough to absorb a still-excess supply base; until that happens, any price floor just turns into volume rationing and intermittent output cuts. That means the next 1-3 months are more about signaling than earnings power, while the real structural inflection, if it comes, is 6-18 months away and depends on enforcement of energy-intensity rules forcing closures rather than voluntary compliance.

For competitors, the first-order loser is the high-cost tier of Chinese polysilicon producers with weak liquidity and older plants; second-order losers are wafer and module makers that lose input-cost relief while end demand stays soft, so their working capital may stay trapped longer. DQ’s balance sheet buys time, but time is not the same as value creation: cash can fund survival and a speculative pivot, yet the AIDC initiative is still a pre-revenue option with execution risk, customer-concentration risk, and a long validation cycle. NVDA is a narrative beneficiary from the 800V DC framing, but the linkage is currently more branding than economic exposure.

Contrarian view: the consensus is too willing to extrapolate “policy support” into a clean price rebound. A forced industry cleanup could indeed improve margins eventually, but the path is likely messy, with more shutdowns, delayed utilization recovery, and false starts in spot pricing before any true floor emerges. If polysilicon transactions fail to hold above cash cost for multiple weeks, or if Q3 guidance comes in only through lower utilization rather than better realized pricing, the bullish thesis should be treated as a value trap rather than a cyclical turn.

Key Decisions for Investors

• Stay underweight/avoid DQ for now; the near-term setup is a dead-cat-bounce risk, not a confirmed cycle turn. Reassess only if spot pricing sustains above cash cost for 4-6 weeks and Q3 mix shifts from volume rescue to margin expansion.
• If trading DQ tactically, prefer a short-dated call spread sale or put spread purchase on any policy-driven rally; thesis is that the market will fade once it realizes inventory is still the binding constraint. Falsifier: evidence of real, high-volume transactions clearing materially above cash cost.
• Long NVDA / short DQ as a narrative-vs-fundamentals pair: NVDA captures monetizable AI infrastructure demand today, while DQ’s AIDC story is a 2027-2030 option with uncertain conversion. Best entry is on any DQ spike tied to policy headlines.
• Watch downstream solar names and module/wafer proxies for a lagged margin squeeze; if polysilicon prices are propped up without a demand pickup, the pressure shifts to converters within 1-2 quarters. This is a cleaner way to express the “higher input cost, weak end demand” mechanism than owning DQ.
• Set an alert for January 2027 enforcement and plant shutdown announcements. That is the real catalyst window for a structural re-rating; absent hard shutdowns, the current policy support should be treated as rhetorical rather than investable.