The price-to-book of 0.18 and price-to-sales of 0.14 are the only internally consistent anchors available, since reported EPS was negative $1.13 in 2026Q2 and seven of the last ten quarters posted losses, leaving the TTM P/E of -7.45 analytically meaningless. The headline EV/EBITDA of 10.95x warrants caution rather than comfort: EBITDA in this model is inflated by project divestiture timing, and the same figure at 128.8x on a trailing debt-adjusted basis suggests the earnings base, not the valuation, is the distorted input. Against JinkoSolar at 16.03x and First Solar at 7.36x, CSIQ screens in the middle, but the peer group is itself loss-making, so the discount may reflect structural quality differences rather than a temporary dislocation. If the market is pricing an asset-value floor via the 0.18x book multiple, that floor presumes project assets held at cost approximate realizable value, which investors should verify given the 10.95x EV/EBITDA disconnect and the collapse in deferred revenue.

The gross margin path — 19.0%, 17.2%, 16.4%, 14.3%, 11.7%, 29.8%, 17.2%, 10.2%, 25.1%, 13.9% — is far too volatile to represent pricing power in the module business; it more plausibly reflects when Global Energy project sales land in the quarter, consistent with the inter-segment deferral mechanics noted in CSIQ's filings. The more durable signal is the operating line: negative in five of the last six quarters and at -5.9% in 2026Q2, the overhead base is not being absorbed even in quarters with 25% gross margins, which implies SG&A is fixed relative to a shrinking revenue base rather than variable. On this structure, true earning power appears to sit closer to the operating margin than the net margin, since the -1.86% TTM net figure is further depressed by non-operating items and interest cost on a $7.3 billion debt load. A permanent improvement likely requires either a sustained storage and services mix shift or a lower fixed-cost footprint; neither is yet visible in the reported ratios.

ROIC has oscillated between -1.2% and 1.0% across the window with no discernible trend toward the positive side, and ROE of -1.8% confirms the equity base is being impaired rather than compounded — retained earnings fell from $1.6 billion to $1.4 billion while nominal equity stayed near $2.8 billion. Critically, the decay appears driven more by efficiency than by margin alone: asset turnover halved from 0.13 to 0.08 as total assets grew 29.8% to $16.1 billion without a commensurate revenue contribution, indicating that the capacity build has yet to be monetized. Because ROIC here is measured against an asset base that includes project assets carried at cost, the reported figures may actually overstate the capital intensity problem if those assets are worth more than book, or understate it if market IRRs have compressed; that judgment cannot be resolved from the ratios alone. At minimum, the data suggest returns are decaying rather than compounding, which is inconsistent with a capital-intensive business that is still adding fixed assets.

The apparent stability of the CCC is misleading when decomposed: DSO of 90 days is broadly consistent with its historical range, but DIO has deteriorated materially — from 87 days in 2024Q4 to 160 in 2026Q1 and 139 in 2026Q2 — which is consistent with module price declines requiring inventory to be held longer or written down, and the reported write-down risk noted in CSIQ's disclosures supports this reading. The offsetting improvement in DPO, which rose from 56 days in 2024Q2 to 120 days, does not appear to reflect improved supplier bargaining power given the sector-wide oversupply; it may instead indicate delayed payment or extended terms, which functions as short-term financing that is not repeatable indefinitely. Read together with the roughly $2.96 billion cumulative working capital drain disclosed in the cash flow statements, the working capital cycle has been a net source of cash consumption, not a source of flexibility. Investors should monitor whether inventory days normalize as a sign of clearing, or continue rising as a sign of price-driven obsolescence risk.

The leverage profile has moved in one direction: D/E climbed from 1.28 in 2024Q2 to 1.72 in 2026Q2 while equity remained flat at approximately $2.8 billion, meaning the balance sheet grew 29.8% in assets entirely through liabilities — an unusual posture for a business operating at negative ROIC. Interest coverage has flipped repeatedly between roughly 2.8x and negative territory, with -1.12x in 2026Q2 and -1.45x in 2025Q4, which implies earnings available to service debt are not being generated with any consistency; the reported 18.34% gross margin is insufficient to cover SG&A, let alone interest. Liquidity compounds this: the current ratio of 1.02 and quick ratio of 0.78 in 2026Q2 leave almost no buffer, and deferred revenue falling from $362.3 million to zero suggests contracted prepayments are not replenishing the funding base. The disclosed 1.80% Debt/Equity figure referenced in other materials appears inconsistent with the consolidated ratios shown here and warrants verification before any refinancing assumption is made.