The Zhitong Finance App learned that Guojin Securities released a research report saying that AI computing power construction continues to drive the upgrading of chip cooling requirements. The power consumption of a single GPU has increased from 400W to 1400W, and TIM has become a key material to guarantee the performance and reliability of high-power chips. The carbon-based route is expected to open a domestic breakthrough window. The industrialization opportunities for graphene thermal pads are clear. It is recommended to focus on companies that have actively laid out graphene thermal pads and have a first-mover advantage.
Guojin Securities's main views are as follows:
AI computing power upgrades drive increased chip cooling requirements, and traditional TIM performance bottlenecks gradually became apparent
There are slight undulations on the surface of the chip, package cover, and radiator. TIM transfers heat from the chip to the radiator or liquid cooling plate by filling gaps in the interface. Its performance directly affects the chip temperature and long-term operation reliability. According to the package structure, a package with a cover usually connects the chip to the package cover plate through TIM1, and the cover plate to the radiator through TIM2; the uncovered package uses TIM1.5 to directly connect the bare chip to the radiator or cold plate to shorten the heat transfer path, but places higher requirements on material flexibility, interface fit, and assembly stress. The power consumption of a single Nvidia GPU was raised from 400W of the A100 to a maximum of 1400W on the GB300 platform; at the same time, 2.5D/3D packaging integrates chips, HBM and other devices in a larger package area, and differences in material thermal expansion and package warpage make it more difficult to fit the interface. Therefore, TIM must not only quickly export heat, but also adapt to interface deformation and maintain long-term stability. Traditional polymer TIMs such as silicone grease and gels are limited by low thermal conductivity substrates, and it is easy to increase material stiffness; although materials such as liquid metal have high thermal conductivity, they still face challenges such as conductivity, leakage, corrosion, and assembly adaptation, driving the continuous iteration of high-performance TIM.
Graphene TIM takes into account efficient thermal conductivity, soft fit and long-term stability, and the industrialization process continues to advance
The advantages of graphene TIM materials are reflected in: (1) Strong thermal conductivity. Graphene has a high intrinsic in-plane thermal conductivity, and a heat transfer path in the thickness direction is formed through structural design. The thermal conductivity of the disclosed high-performance graphene thermal conductive gasket is about 90-200W/ (m·K); (2) The interface has good adhesion. After graphene is compounded with a flexible substrate, the gasket has compression resilience and can adapt to warping and assembly gaps in large packages; (3) It has good long-term stability. The flexible solid state structure helps reduce material migration and pumping out in the thermal cycle and maintains interfacial contact. The technical route gradually evolved from early dispersing graphene as a filler in polymers to constructing a three-dimensional continuous thermal conductivity network, controlling the vertical orientation of graphene, and transforming the material's in-surface thermal conductivity into the longitudinal thermal conductivity required for the chip. On the market side, AI infrastructure construction is driving the deployment of high-power chips, and applications such as GPUs, ASICs, and server CPUs are increasing in demand for high-performance TIM, and expanding to scenarios such as optical modules and power semiconductors. According to QYResearch data, the global TIM market is expected to grow from about US$2,012 billion in 2024 to US$4.148 billion in 2031, with a compound growth rate of about 10.7% from 2025 to 2031; market expansion and increase in penetration rate of high-end materials are expected to jointly open up room for growth of graphene TIM.
The carbon-based route opens a domestic breakthrough window, and the leading companies with graphene thermal conductive gaskets have significant card position advantages
The traditional TIM market is dominated by international giants DuPont, Dow, and Henkel, and the competitive pattern is relatively solidified; while graphene thermal conductive gaskets are an emerging segment, the market pattern has not yet been defined, and domestic manufacturers are expected to take the lead in breaking through. Graphene thermal conductive gaskets have the triple barriers of material design, mass production process, and customer certification. Currently, there are still few manufacturers that have achieved large-scale batch supply. Among them, Hong Fu Cheng is one of the few companies in the industry that can mass-produce graphene thermal conductive gaskets and achieve large-scale applications. The company mass-produces 130W/ (m·k) vertically oriented graphene thermal conductive gaskets with thermal resistance as low as 0.04-0.06°C·cm2/W. It has supplied leading global AI chip customers in batches, and achieved a breakthrough in the small-batch supply of TIM1 thermal interface materials. Benefiting from product release, the company's revenue from carbon-based thermal conductive gaskets increased from 35.61 million yuan to 252.266 million yuan in 2023-2025, with a year-on-year growth rate of over 138% for two consecutive years. In 2025, the gross profit margin was 65.6% and after deducting non-net interest rate of 37.4%, significantly leading the industry. At the same time, domestic manufacturers such as Sinopec Technology, Siquan New Materials, and Feirongda are also speeding up the deployment of vertically oriented graphene TIMs, and the industrialization process is expected to accelerate. As demand for AI computing power cooling continues to increase, graphene thermal pads are expected to become a key entry point for domestic manufacturers to accelerate their breakthrough into the high-end.
Risk Alerts
Demand for AI chips falls short of expected risk; customer verification and batch introduction fall short of expected risk; large-scale production and cost control fall short of expected risk; competition and penetration rate of alternative materials fall short of expected risk.