Cambricon faces short-term revenue pressure, but margin improvement supports the Overweight view

Overview

This report updates the risk-reward scenarios and earnings forecasts for Cambricon Technology. Morgan Stanley believes that weaker second-quarter revenue growth creates short-term pressure on 2026 performance, but the company's deep collaboration with major cloud service customers, expansion of domestic AI computing infrastructure, improving supply-chain visibility, and new product iterations still support long-term growth. It therefore maintains its Overweight rating and Rmb1,408 target price.

Core views

Core highlights include demand for cloud training and inference chips driven by domestic generative AI infrastructure spending, Cambricon's strategic position in the domestic high-performance AI chip market, the product-market fit of MLU590, and the performance upgrade brought by MLU690. Absolute valuation multiples are high, but the research believes they can be supported by strong growth, improving supply, and the company's strategic positioning in domestic AI chips.

Analysis framework

The report combines earnings forecast revisions, 2026 expected price-to-sales valuation, bull-base-bear scenario analysis, and long-term valuation parameters to assess the impact of changes in revenue growth, gross margin, market share, and supply-chain execution on the target price.

Methodology notes

• Scenario analysisBull-base-bear risk-reward framework

  Three valuation scenarios are formed through assumptions on revenue growth, market share, customer orders, and gross margin.

  The bull-case scenario corresponds to Rmb2,660 and 85x 2026E price-to-sales, the base-case scenario corresponds to Rmb1,408 and 45x, and the bear-case scenario corresponds to Rmb709 and 23x.

• Relative valuationExpected price-to-sales valuation

  Measures the value of a high-growth AI chip company based on 2026 expected revenue and different price-to-sales multiples.

  The report believes the current absolute levels of P/E and P/S ratios are high, but can be partly explained by high growth, improving supply visibility, and the strategic position of domestic AI chips.

• Long-term valuation parametersCost of equity and long-term growth assumptions

  Uses cost of equity, payout ratio, and medium- to long-term growth parameters to support the valuation model.

  Key assumptions include an 8.4% cost of equity, a beta of 1.06, a 2.0% risk-free rate, a 6.0% equity risk premium, a 40% long-term payout ratio, a 16% medium-term growth rate, and a 6.0% terminal growth rate.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• Cambricon Technology (688256.SS)

  Core covered company in the report and a Chinese domestic high-performance AI chip supplier

  Strengths

  Deep cooperation with major cloud service customers, and MLU590 has product-market fit; benefits from domestic AI infrastructure expansion, domestic substitution, pricing power, and operating leverage.

  Weaknesses

  Valuation multiples are at a high level, domestic wafer manufacturing and HBM costs are relatively high, and yields remain challenging during the supply-chain transition.

  Comparison

  It has a key strategic position in China's domestic high-performance AI chip market, and the reasonableness of its valuation mainly depends on above-industry growth and sustained share gains.

  Risks

  Slowdown in AI capital expenditure, weaker-than-expected commercialization of large models, lack of consecutive large orders from leading customers, market share decline, price competition, rising wafer costs, and insufficient capacity utilization.

Key data

• Stock ratingOverweightThe industry view is Attractive, and the rating typically corresponds to a relative risk-adjusted return assessment over the next 12 to 18 months.
• Target priceRmb1,408.00Maintained unchanged in this update.
• Reference closing priceRmb1,199.93As of August 7, 2026.
• Potential upside to target priceApproximately 17.3%Calculated based on the target price and the reference closing price.
• 2026 expected EPSRmb11.3Lowered by approximately 3% from the previous Rmb11.6.
• 2027 expected EPSRmb19.0Raised by approximately 3% from the previous Rmb18.4.
• 2028 expected EPSRmb24.6Raised by approximately 1% from the previous Rmb24.5.
• Base-case scenarioRmb1,408Assumes a 2025-2028 revenue CAGR of 99%, with gross margins of 55% and 52% in 2026 and 2027, respectively.
• Bull-case scenarioRmb2,660Assumes a 2025-2028 revenue CAGR of over 130% and continued expansion of domestic high-performance AI chip market share.
• Bear-case scenarioRmb709Assumes a 2025-2028 revenue CAGR below 60%, and gross margins in both 2026 and 2027 falling below 40%.

Impact & implications

The earnings forecast adjustment shows some slowdown in short-term revenue momentum, but improved margin and EPS expectations for 2027-2028 suggest the investment thesis is shifting from pure revenue expansion toward the realization of scale effects, pricing power, and operating leverage. If domestic AI capital expenditure, customer procurement, and supply yields improve as expected, the target price still offers upside; if orders, share, or gross margin fall short of expectations, the high valuation will amplify downside risk.

Risks

• Domestic AI capital expenditure and large-model commercialization progress more slowly than expected.
• Failure to continuously secure large-scale orders from leading customers, or loss of share in the domestic AI chip market.
• Yield improvement during the production shift to SMIC is slower than expected, affecting supply and costs.
• Relatively high domestic wafer manufacturing and HBM costs may pressure gross margins.
• Intensifying price competition, rising wafer manufacturing costs, or insufficient capacity utilization could drive gross margins below expectations.
• Absolute valuation multiples are high, and weaker-than-expected growth or margins could trigger a significant valuation correction.
• Morgan Stanley disclosed that it holds 1% or more of a certain class of ordinary equity securities of Cambricon Technology, which may create a potential conflict of interest affecting research objectivity.

What to watch

• SMIC-related production yields and supply-chain transition progress in the second half of 2026.
• MLU590 order visibility and the procurement pace of major cloud service customers.
• Whether MLU690 can launch as planned in 4Q 2026 and achieve approximately 2.2x performance improvement.
• Domestic generative AI infrastructure capital expenditure and large-model commercialization progress.
• Whether gross margins in 2026 and 2027 can reach the base-case assumptions of 55% and 52%, respectively.
• Whether the company can continue to increase its share of the domestic high-performance AI chip market.
• Subsequent quarterly revenue growth and whether 2026 EPS forecasts are adjusted further.