Cambricon’s third-quarter revenue missed Goldman Sachs’ forecast. Why is its long-term growth outlook unchanged?
Third-quarter revenue doubled year over year and gross margin continued to improve, but tight supply and the timing of purchases by major customers left revenue well below Goldman Sachs’ forecast. Goldman cut its 2026 earnings forecast, largely retained its medium-term growth outlook, and lowered its target price following a decline in industry valuations.
Original source：Goldman SachsFirst published on Unhidea：Original source date：
Key takeaway
Goldman Sachs attributed Cambricon’s third-quarter revenue shortfall mainly to supply constraints and purchase timing and maintained its outlook for a medium-term ramp-up. However, its model requires a sharp jump in fourth-quarter revenue, with growth accompanied by lower margins.
Revenue doubled. Why cut the earnings forecast?
Cambricon reported third-quarter 2026 revenue of 3.531 billion yuan, up 104% year over year and 13% quarter over quarter, but 57% below Goldman Sachs’ original forecast of 8.178 billion yuan. Strong growth and a miss against expectations coexist because Goldman had set a high revenue benchmark.
Gross margin rose from 56.1% in the second quarter to 58.7%, also exceeding Goldman’s forecast of 50%; it was still insufficient to offset the revenue shortfall. Third-quarter EBITDA was 1.533 billion yuan, 19% below the bank’s forecast; net profit was 1.404 billion yuan, 21% below expectations.
Accordingly, Goldman cut its 2026 revenue forecast by 23% to 18.438 billion yuan and its net profit forecast by 11% to 5.515 billion yuan. It raised its full-year gross margin forecast from 51.3% to 53.4%, cushioning some of the impact of the revenue revision.
How will supply and purchase timing affect the subsequent ramp-up?
Goldman attributed the revenue shortfall mainly to tight raw material supplies and the timing of purchases by leading cloud customers. Cambricon has a concentrated customer base, so when major customers make purchases significantly affects revenue recognition in any single quarter.
Inventory reached 11.7 billion yuan at the end of the third quarter, up 41% quarter over quarter. Goldman believes this can support subsequent AI chip shipments. However, wafer supply constraints remain the bank’s principal downside risk, so the inventory build and supply constraints need to be considered together.
The company aims for fourth-quarter revenue growth of at least the third quarter’s 13% quarter-over-quarter rate. Goldman’s model, meanwhile, forecasts fourth-quarter revenue of 8.911 billion yuan, up approximately 152% from the third quarter. The two thresholds differ substantially, and Goldman’s full-year forecast depends on a more concentrated, larger-scale realization of revenue in the fourth quarter.
What will drive medium-term growth, and what is the cost to margins?
Goldman expects capital spending by China’s leading cloud platforms to grow by 105%, 21% and 14%, respectively, over 2026–2028. Taking this together with domestic substitution and upgrades to chip specifications, it remains positive on Cambricon. Expanding customer budgets create room for demand, while product upgrades support an improved product mix and higher selling prices.
The bank’s model forecasts cloud chip sales rising from 303 thousand units in 2026 to 1 million units in 2027, with the average selling price increasing from 8,459 dollars to 12,673 dollars. Higher volumes and prices support revenue growth of 392% in 2027; the model projects 93% revenue growth in 2028, while the company’s target is to maintain triple-digit revenue growth through 2028.
Goldman largely retained its revenue and earnings forecasts for 2027–2028, but the ramp-up does not coincide with a rise in margins: in the model, gross margin falls to 51% in 2027, the expense ratio rises from 21.7% in 2026 to 33.9%, and net margin falls from 29.9% to 15.2%.
Buy rating maintained. Why lower the target price?
Goldman lowered its 12-month target price from 1,841 yuan to 1,680 yuan while maintaining its Buy rating. The key change is the valuation multiple: industry valuations fell amid concerns about the memory cycle and the sustainability of AI spending, and the bank reduced its target EV/EBITDA multiple from 27 times to 23 times.
The valuation remains based on 2030 earnings: Goldman raised its EBITDA forecast for that year by 8% to 65.821 billion yuan, citing an improved product mix, scale efficiencies and returns on earlier R&D investment. It then applies a multiple of 23 times and discounts the result to 2027 using a cost of equity of 12.7%. The higher long-term earnings forecast still did not offset the lower valuation multiple.
