Yinlun (002126): J.P. Morgan initiates Yinlun at Overweight on AI power, liquid-cooling and truck-electrification growth

Overview

J.P. Morgan initiates coverage of China-based thermal-management supplier Yinlun with an Overweight rating. The report expects AI data-center power demand, liquid cooling and commercial-vehicle electrification to shift the company toward faster-growing and higher-margin businesses, supporting earnings growth and a valuation re-rating.

Core views

J.P. Morgan’s central thesis is that Yinlun’s established passenger-vehicle and commercial-vehicle thermal-management operations provide a broad base, while its AI infrastructure exposure is not fully reflected in the market’s view of the company. The institution forecasts roughly 20% revenue CAGR and roughly 30% net-profit CAGR over 2025-28E, with net profit more than doubling by 2028E. Group revenue is projected to rise from Rmb15.7bn in 2025 to Rmb26.6bn in 2028E, while net profit rises from Rmb957mn to Rmb2,173mn. The expected mix shift is important: Digital & Energy is forecast to grow at a 54% CAGR over 2025-28E and expand from 9% of revenue in 2025 to about 20% in 2028E, bringing structurally higher margins than legacy automotive components. The clearest near-term driver is genset cooling supplied to Caterpillar. J.P. Morgan argues that AI data-center construction is increasingly constrained by grid availability, prompting demand for diesel backup engines and gas prime-power engines. It forecasts global data-center power installed base to grow from about 117GW in 2025 to 242GW in 2028, while AI data-center power rises from about 37GW to 148GW. Caterpillar has indicated order visibility extending toward 2030, and J.P. Morgan expects its Power Energy revenue to grow at a 13% CAGR over 2025-30E. Yinlun began volume shipments of generator cooling packages to Caterpillar in 2H24; its multi-year supplier qualification, Mexico production footprint and prices estimated at 20-30% below Modine are presented as advantages. Yinlun’s share of Caterpillar cooling-package requirements is estimated to rise by about 10 percentage points to above 50% by 2028E. The resulting financial contribution is substantial in the report’s forecasts. Genset revenue is projected to increase from Rmb0.5bn in 2026E to Rmb1.5bn in 2027E and Rmb3.2bn in 2028E, a 143% CAGR. This would represent about 3%, 7% and 12% of group revenue in 2026E-28E, respectively, and 6%, 13% and 21% of group gross profit. The report expects genset cooling packages and aftertreatment systems to earn 36-38% gross margins, above the legacy business. Caterpillar’s 2026 cooling-package order is cited at about 3,400 units, up 70% from around 2,000 units at the start of the year, with roughly 5,000 units indicated for 2027. Gas-engine aftertreatment deliveries are expected to begin in 4Q26, reaching 700-800 units in 2027 at about US$130,000 per unit. Liquid cooling is presented as the second AI infrastructure growth leg. Higher AI server power density is increasing demand for coolant distribution units, and plate heat exchangers account for about 10-20% of CDU cost. A CDU typically contains around three PHEs with an ASP of roughly Rmb40,000-50,000 each, implying Rmb120,000-150,000 of PHE content per CDU. J.P. Morgan estimates the global secondary-side CDU PHE market will grow from about Rmb3bn in 2025 to about Rmb13bn in 2028, a 63% CAGR. Yinlun’s long experience in plate forming and vacuum brazing, self-designed furnaces and planned localized capacity are seen as differentiators because furnace design, commissioning and yield ramp-up constrain rapid capacity replication. Its first large-format PHE line is scheduled for 4Q26, with South China and Thailand lines planned for 2027; together, the three lines imply about Rmb1.8bn of annualized output value at full utilization. The base case excludes Google-related orders. Yinlun passed Google’s designated third-party laboratory certification in March 2026 but still needs CDU-integrator system validation, a potential factory audit and purchase-order approval. The report forecasts base-case liquid-cooling revenue of Rmb500mn in 2026E and Rmb840mn in 2028E, a 30% CAGR, with approximately 3% of group revenue and 4-5% of group gross profit over the period. If Google qualification converts into commercial orders, liquid-cooling revenue could reach about Rmb1.4bn by 2028E, contributing roughly 5% of group revenue and 9% of group gross profit; this would raise net-profit forecasts by approximately 4% in 2027E and 6% in 2028E. J.P. Morgan treats this outcome as upside optionality rather than confirmed revenue. Commercial vehicles and non-road machinery provide the resilient earnings base. The segment generated about Rmb3.32bn of revenue in 1H26, up 37% year on year, and about Rmb785mn of gross profit, with a roughly 24% gross margin versus about 18% for the group and about 13% for passenger vehicles. Around 30% of segment revenue comes from overseas markets. The report forecasts segment growth of 28%, 18% and 12% in 2026E, 2027E and 2028E, driven by export demand, domestic heavy-duty-truck replacement and electrification. Thermal-management content is estimated to rise from roughly Rmb14,000-15,000 per ICE truck to Rmb20,000-30,000 per electric truck. J.P. Morgan expects EV heavy-duty-truck penetration to increase from about 25% in 2025 to around 50% by 2030E, above China’s stated 40% 2030 new-energy penetration target. The report expects the mix shift to improve profitability and cash generation. Passenger-vehicle gross margin fell to 13% in 1H26 from 14% in 2025 amid pricing pressure, whereas commercial-vehicle and non-road gross margin held at about 24%. Group gross margin is forecast to recover from 18% in 1H26 to about 21% by 2028E as Digital & Energy scales. ROE is projected to rise to about 21% by 2028E, supported by an 8% net margin and 0.9x asset turnover. Net debt-to-equity is expected to decline from 55% in 2025 to 20% by 2028E, while operating cash flow rises from Rmb1,405mn to Rmb3,405mn and free cash flow reaches Rmb2,425mn. J.P. Morgan values Yinlun using a sum-of-the-parts approach because of the divergent growth and margin profiles of auto parts and Digital & Energy. The base case applies 15x 2028E P/E to passenger- and commercial-vehicle auto parts and 35x to Digital & Energy, producing a Rmb51 Dec-2027 target price, equivalent to 29x 2027E and 20x 2028E blended P/E and 49% upside from the Rmb34.20 reference price. The upside case, incorporating Google-related PHE orders while retaining the valuation framework, produces a Rmb55 value and about 61% upside. The report argues that a single blended multiple would understate the value of the AI power and liquid-cooling optionality.

Analysis framework

The report builds its case from three earnings drivers—Caterpillar genset cooling, AIDC liquid cooling and commercial-vehicle electrification—then tests them against market demand, product content, customer qualification barriers, capacity plans and segment margins. It translates these operational assumptions into segment revenue, profit, cash-flow and balance-sheet forecasts before applying a segment-specific SOTP valuation.

Methodology notes

• Valuation methodsSOTP (Sum-of-the-Parts) Valuation

  Sum-of-the-parts valuation using different P/E multiples for mature auto-parts operations and Digital & Energy.

  J.P. Morgan applies 15x 2028E P/E to passenger- and commercial-vehicle parts and 35x to Digital & Energy, reflecting the latter’s higher expected growth and margins.

• Corporate Fundamentals and FinanceDuPont analysis

  ROE decomposition through net margin, asset turnover and equity multiplier.

  The report attributes projected ROE improvement mainly to higher margins and faster asset turnover as higher-margin Digital & Energy businesses scale.

• Industry AnalysisUpstream-Midstream-Downstream Transmission

  AI data-center demand is linked to power generation, CDU demand and Yinlun’s cooling-component content.

  The report traces AI compute growth through data-center power constraints, genset demand, liquid-cooling adoption and component orders.

Asset mapping & comparison

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

• Yinlun (002126.SZ)

  Primary covered company; expected to benefit from AI data-center power demand, liquid cooling and commercial-vehicle electrification.

  Strengths

  Multi-year Caterpillar qualification, Mexico production, cost competitiveness, vacuum-brazing expertise and self-designed furnace capability.

  Weaknesses

  Passenger-vehicle margin pressure and dependence on unconfirmed liquid-cooling order conversion.

  Comparison

  The report sees Modine’s strategic focus on data-center cooling as creating an opportunity for Yinlun to gain genset cooling share at Caterpillar.

  Risks

  High Caterpillar concentration, delayed PHE qualification or order conversion, capacity underutilization and geopolitical localization pressures.

• Caterpillar (CAT US)

  Key customer and demand channel for Yinlun’s genset cooling packages and aftertreatment systems.

  Strengths

  AI data-center-related power demand and long order visibility support its engine-production outlook.

  Comparison

  Yinlun is positioned against incumbent supplier Modine in Caterpillar’s cooling-package supply chain.

  Risks

  Changes to Caterpillar procurement, sourcing, production plans or pricing could materially affect Yinlun.

• Modine (MOD US)

  Competitor and prior leading Caterpillar genset cooling-package supplier.

  Weaknesses

  The report says management attention and capital allocation are increasingly focused on North American data-center cooling.

  Comparison

  Yinlun’s estimated 20-30% price advantage and Mexico footprint may help it take share.

Key data

• Target priceRmb51.00 by Dec-202749% upside from the Rmb34.20 closing price on 29 Sep 2026
• Group revenue forecastRmb15.7bn in 2025 to Rmb26.6bn in 2028EApproximately 19% CAGR
• Net profit forecastRmb957mn in 2025 to Rmb2,173mn in 2028E31% CAGR over 2025-28E
• Genset revenueRmb0.5bn in 2026E to Rmb3.2bn in 2028E143% CAGR; projected to contribute 21% of group gross profit by 2028E
• Liquid-cooling upsideRmb1.4bn potential 2028E revenueAssumes Google-related order conversion; base case is Rmb840mn and excludes those orders
• EV HDT penetrationAbout 25% in 2025 to about 50% in 2030EJ.P. Morgan forecast versus China’s 40% 2030 target

Impact & implications

The report expects AI infrastructure businesses to increase Yinlun’s growth, gross-margin and earnings-mix quality relative to its legacy automotive operations. Its investment conclusion depends chiefly on Caterpillar order execution, conversion of liquid-cooling qualifications into volume orders and continued strength in commercial-vehicle exports and electrification.

Risks

• Caterpillar represented about 16% of Yinlun’s commercial-vehicle and non-road revenue, or about 6% of group revenue, in 2025; order, sourcing, inventory or pricing changes could have an outsized earnings impact.
• Google-related PHE applications remain in validation, with no confirmed volume orders; delays or unsuccessful validation could defer revenue conversion.
• PHE capacity is being added before confirmed demand, creating utilization, depreciation, staffing and margin risks if orders ramp slowly.
• Broader US localization or de-risking policies could lengthen qualification cycles, increase compliance requirements or limit overseas opportunities.

What to watch

• Caterpillar cooling-package volumes, supply-share gains and the rollout of higher-value aftertreatment products.
• Progress through CDU-integrator validation, potential Google factory audit and purchase-order approval.
• Ramp-up and utilization of Yinlun’s 4Q26 PHE line and the planned South China and Thailand lines in 2027.
• Commercial-vehicle export momentum, domestic heavy-duty-truck replacement demand and EV HDT penetration.
• Passenger-vehicle pricing pressure and the pace of group gross-margin recovery.