A tech supplier is auctioning an enormous AI supercomputer setup via eBay, asking nearly $60 million for a cluster equipped with 9,600 Nvidia A100 GPUs and 4,200 AMD EPYC processors. This large-scale infrastructure could serve AI development and scientific workloads, presenting a rare procurement option for large compute needs.
- 300-rack AI supercomputer listed for $59.7 million
- Features 9,600 Nvidia A100 GPUs and 4,200 AMD EPYC Rome CPUs
- Includes free ground shipping and optional expedited delivery
What happened
Core 4 Solutions has put up for sale a massive AI supercomputer on eBay, encompassing 300 server racks built with cutting-edge hardware from Nvidia and AMD's prior generation components. The installation includes 9,600 Nvidia A100 GPUs and 4,200 AMD EPYC Rome processors, coupled with substantial memory and networking equipment, making it one of the largest such setups publicly listed for direct purchase.
While the Nvidia A100 GPUs utilize the Ampere architecture which is now superseded by newer Nvidia architectures like Blackwell, they still offer powerful acceleration for demanding AI and machine learning workloads. Similarly, the AMD EPYC Rome CPUs represent an earlier generation but deliver significant core counts and performance. The seller includes extensive infrastructure such as power, cooling, and networking hardware essential for operation.
Why it matters
Listing a supercomputer of this scale on an open marketplace like eBay is unusual and highlights potential new pathways for organizations to acquire high-performance compute environments outside traditional procurement channels. It presents a rare opportunity to secure substantial AI compute capacity at a defined price, sidestepping long vendor lead times and customization waits.
The $59.7 million asking price, while high, reflects more than just hardware costs by including critical infrastructure and support equipment. Despite the equipment being from an earlier generation, the setup remains valuable for many AI, ML, and scientific modeling applications that rely on large-scale parallel processing and memory capacities.
What to watch next
Interest levels indicated by watchers and a cart addition suggest some potential buyers are evaluating the listing, but serious negotiations would require thorough evaluation of the system’s condition, power needs, and software compatibility. The logistics of transporting 300 server racks, even with free shipping options, will also challenge potential purchasers given the size and setup requirements.
Future developments could include more listings for surplus or used large-scale AI infrastructure as data centers refresh hardware or firms refocus investments. How the market adapts to such direct sales could influence procurement trends and the secondary market for high-performance computing equipment.