Chinese factories are dismantling thousands of RTX 5090s to power AI servers

A newly surfaced video reveals the extent to which Chinese factories are dismantling high-end RTX 5090 graphics cards to repurpose their components for AI server use. The footage shows workers removing GPU dies and memory modules from brand-new cards, which are then transplanted into blower-style designs. This practice is reportedly being carried out on an industrial scale, with thousands of units being processed daily. The video highlights the increasing demand for AI-specific hardware, which is driving this large-scale repurposing of consumer-grade GPUs.

The RTX 5090, one of Nvidia's most powerful consumer graphics cards, is being targeted in this process. The video shows the GPU die and memory being desoldered and moved to custom card designs. This suggests that Chinese manufacturers are sourcing these components to meet the needs of AI developers and data centers. The practice raises questions about the sustainability of such operations and the potential impact on the availability of consumer-grade GPUs in the market.

The scale of the operation is significant, with thousands of RTX 5090s being dismantled each day. The video shows a continuous flow of GPUs being processed in industrial workshops, indicating a well-organized system. This level of activity suggests that the demand for AI hardware is outpacing the supply of dedicated AI chips. The repurposing of consumer-grade GPUs is becoming a viable alternative for companies looking to build AI servers without waiting for specialized hardware.

This practice could lead to increased costs for consumers, as the supply of high-end GPUs like the RTX 5090 may become more limited. It may also contribute to vendor lock-in, as companies that rely on repurposed components may find themselves dependent on specific suppliers. Additionally, the environmental impact of dismantling and repurposing GPUs on such a large scale could raise governance concerns. Market reactions may vary, with some viewing this as a creative workaround and others seeing it as a sign of growing supply chain challenges.

The situation is still developing, with no immediate resolution in sight. The video has sparked discussions about the long-term implications of this practice for both the consumer and AI markets. As demand for AI hardware continues to rise, the pressure on manufacturers to find alternative solutions may only increase. This could lead to further innovations in hardware repurposing or a shift toward more dedicated AI chip production.