Gf Securities released a research report stating that, based on a 5% residual value rate for traditional electronic products, the AI electronic waste industry is projected to reach a scale of 500 billion yuan by the end of the 15th Five-Year Plan period. Considering the high-value nature of servers and typical case residual value rates, it is expected that recycling of AI servers alone will generate an annual market worth 60 to 76 billion yuan during the 2027-2028 period. The institution believes that the formalization of AI waste recycling is an inevitable trend, and industry leaders possessing formal waste electrical and electronic equipment (WEEE) dismantling qualifications are expected to benefit substantially.

The main points from Gf Securities are as follows:

**Trillion-Yuan AI Industry Booms, 500 Billion in AI Waste Value Awaits Tapping** The generative AI wave is sweeping the globe. During a press conference at the Fourth Session of the 14th National People's Congress in March, Zheng Shanjie, head of the National Development and Reform Commission, stated that the scale of the AI-related industry is expected to exceed 10 trillion yuan by the end of the 15th Five-Year Plan period. This rapid development, coupled with the increasing complexity of related electronic products, is leading to exponential growth in AI electronic waste. A report in Nature warns that without a circular economy strategy, e-waste from generative AI could increase nearly 1,000-fold between 2020 and 2030. Based on a 5% residual value rate for traditional electronics, the AI e-waste industry is forecast to reach 500 billion yuan by the end of the 15th Five-Year Plan.

**Policy Shift Accelerates Formalized Recycling, First Wave of AI Server Recycling Expected in 2027-28** On March 1, 2026, the updated "Technical Specification for Pollution Control in the Treatment of Waste Electrical and Electronic Products" came into effect. This revision significantly expands the scope of national mandatory control to include, for the first time, AI servers and intelligent consumer devices, while also raising requirements for the entire process from collection and transportation to treatment and supervision, marking the arrival of a formalized era for AI waste recycling.

As computing power infrastructure leads AI development, the AI server recycling market is poised to take off first. According to disclosures from the China Academy of Information and Communications Technology (CAICT), China's AI server shipments reached 630,000 units in 2024, a year-on-year increase of 93%, with cumulative shipments over six years exceeding 1.7 million units, representing a market size of $22.1 billion. Based on a typical lifespan of 3-5 years, AI servers procured in 2024 will enter their replacement cycle around 2027-2028. Given the high value of servers and typical residual value rates, the recycling market for AI servers alone is projected to generate 60 to 76 billion yuan annually during the 2027-2028 period, and this is just the beginning.

**High Value and High Barriers Coexist, Leading WEEE Dismantlers Set to Benefit First** AI server recycling differs from traditional e-waste due to several characteristics: 1. **High Residual Value:** For example, the residual value rate for an H100 server in 2025 is as high as 80-90%, with a single card's recycling price reaching $28,000-$35,000. 2. **High Concentration:** TrendForce data shows that in 2022, just eight major global and domestic companies, including Microsoft, Google, Alibaba, and Tencent, accounted for 77% of AI server procurement; future AI waste will largely originate from these sources. 3. **High Complexity:** High-value AI servers require different handling methods—such as near-new reuse, whole-unit refurbishment, component resale, or material extraction—depending on their condition, demanding high comprehensive capabilities from recyclers. 4. **Stringent Requirements:** Coolant from liquid cooling systems cannot be directly discharged, and data security destruction requirements create hard entry barriers. Coupled with increasing environmental focus (as some waste is hazardous) and the carbon traceability and ESG demands of leading AI companies, the formalization of AI waste recycling is inevitable. Industry leaders with formal WEEE dismantling qualifications are well-positioned to benefit significantly. For instance, Zhejiang-based dismantling leader Dadihuanbao has recently begun laying the groundwork for server recycling business through a newly established subsidiary.

AI waste recycling represents an overlooked niche within the AI era. Companies such as Dadihuanbao, Huaxin Environment, and Zhongzai Zihuan are suggested for attention.

**Risk Warning** Risks include potential rapid depreciation and iteration of AI servers, fluctuations in secondary market demand and price uncertainty, and changes in the competitive landscape along with the risk of new entrants.