MetaX (688802) Report Interpretation

Overview

This earnings review assesses MetaX's 2Q26 results, the gradual C600 and C700 product ramp, and the launch of its XiJing S Series AI SuperNode solution. Goldman Sachs maintains Buy, expecting customer penetration and chip shipments to support longer-term growth despite near-term revenue-estimate reductions.

Core views

MetaX began mass production of its C600 chip in May 2026 and continues development of the C700 platform for 2H26. Goldman Sachs expects the new platforms to take time to ramp, but anticipates rising revenue contribution in 2H26 and 2027E as the company penetrates new cloud-service-provider and enterprise customers. Its China base case forecasts industry AI-chip demand at 5.9 million units in 2026E, 21.2 million in 2028E and 39.2 million in 2030E. The report identifies AI-chip profitability and local supply development—especially capacity and yield-rate ramp-up—as central requirements for sustainable growth. For 2Q26, revenue increased 28% YoY and 36% QoQ to Rmb762m, below Goldman Sachs estimates. Gross margin was 55.1%, down from 60.1% in 1Q26 but in line with the institution's forecast. Gross profit reached Rmb420m, up 25% YoY, while operating profit remained negative at Rmb7m. Net income nevertheless turned positive to Rmb711m from a Rmb99m loss in 1Q26 and Rmb47m a year earlier, beating Goldman Sachs' Rmb12m net-loss estimate because fair-value changes in company investments produced higher-than-expected non-operating income. The report therefore distinguishes the earnings beat from the underlying operating ramp. The company also launched its XiJing S Series SuperNode solution at WAIC 2026. The system supports high-density configurations of 64 GPUs per rack, low data latency, and both scale-up and scale-out expansion. Goldman Sachs views the platform as suitable for large-model training and inference because it supports expert parallelism and tensor parallelism. It argues that a full-stack, directly integrated design could enable faster deployment and lower maintenance costs for customers, supporting adoption. Reflecting the quarterly result, Goldman Sachs raises its 2026E net-income forecast to Rmb514m from Rmb42m, primarily because of the non-operating fair-value gain. It lowers 2026E and 2027E revenue forecasts by 9% and 2%, respectively, to Rmb4,001m and Rmb10,204m as C600 and C700 require time to ramp. Revenue forecasts for 2028E through 2030E rise by 1–2% on a more favorable view of customer penetration, while gross-margin and operating-expense-ratio forecasts for 2027E–2030E remain largely unchanged. The model projects revenue of Rmb21,452m in 2028E, Rmb30,953m in 2029E and Rmb41,489m in 2030E, with operating margin improving from -1.7% in 2026E to 18.2% in 2030E. Goldman Sachs raises its 12-month target price to Rmb932 from Rmb912 and maintains Buy. The valuation uses a 68x target EV/EBITDA multiple on 2030E EBITDA, discounted back to 2027E at a 12.7% cost of equity. The cost-of-equity inputs are beta of 1.5, a 3.0% risk-free rate and a 6.5% market-risk premium. The target multiple reflects peer trading EV/EBITDA relative to forward EBITDA growth and margins, alongside MetaX's projected 45% 2031E EBITDA growth and 22% 2031E EBITDA margin.

Analysis framework

Goldman Sachs reviews the quarterly income statement against its estimates, separates operating performance from the non-operating fair-value gain, then revises revenue and earnings forecasts according to the expected C600/C700 ramp. It connects product and customer-adoption developments to longer-term chip volumes and margins, and values MetaX using discounted forward EV/EBITDA.

Methodology notes

• Valuation methodsEV/EBITDA valuation

  Discounted 2030E EV/EBITDA valuation

  Goldman Sachs applies a 68x EV/EBITDA multiple to 2030E EBITDA and discounts the resulting value back to 2027E using a 12.7% cost of equity to derive its 12-month target price.

• Industry AnalysisSupply-demand framework

  China AI-chip unit-demand outlook and local supply ramp

  The report links MetaX's growth outlook to projected China AI-chip demand and to local manufacturing capacity and yield-rate improvement.

Asset mapping & comparison

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

• MetaX (688802.SS)

  Primary covered AI-chip company; expected to benefit from C600 ramp-up, new client penetration and SuperNode adoption.

  Strengths

  C600 entered mass production in May 2026; the XiJing S Series offers 64 GPUs per rack with low latency and integrated deployment.

  Weaknesses

  2Q26 revenue was below Goldman Sachs estimates, and C600/C700 ramp-up leads to lower 2026E and 2027E revenue forecasts.

  Comparison

  The 68x target EV/EBITDA multiple is informed by peer trading multiples relative to forward EBITDA growth and margin.

  Risks

  Slower China cloud-capex growth, fiercer AI-chip competition and advanced-node supply-chain constraints.

Key data

• 2Q26 revenueRmb762mUp 28% YoY and 36% QoQ; 35% below Goldman Sachs estimates.
• 2Q26 gross margin55.1%In line with Goldman Sachs estimates, versus 60.1% in 1Q26.
• 2Q26 net incomeRmb711mTurned positive and beat Goldman Sachs' Rmb12m net-loss estimate due to fair-value changes in investments.
• 2026E revenue forecastRmb4,001mReduced 9% from Rmb4,415m as C600/C700 ramp-up takes time.
• 2026E net-income forecastRmb514mRaised from Rmb42m, mainly on higher-than-expected 2Q26 non-operating income.
• 12-month target priceRmb932Raised from Rmb912; based on a 68x 2030E EV/EBITDA multiple.
• China AI-chip units5.9m / 21.2m / 39.2mGoldman Sachs base-case forecasts for 2026E / 2028E / 2030E.

Impact & implications

The report sees near-term revenue expectations constrained by the gradual C600/C700 ramp, but believes the product transition, wider CSP and enterprise penetration, and the SuperNode offering can underpin longer-term scale and profitability. Its target-price increase reflects the higher 2026E non-operating income outcome while the long-term valuation framework remains unchanged.

Risks

• China cloud-capex growth could be slower than expected.
• Competition among AI-chip providers could intensify more than expected.
• Advanced-node supply-chain constraints could restrict execution.

What to watch

• The pace of C600 shipment ramp-up and C700 platform progress in 2H26.
• Revenue contribution from new CSP and enterprise customers.
• Adoption of the XiJing S Series SuperNode solution.
• Local AI-chip capacity and yield-rate ramp-up in China.