Compute Exchange Launches Secondary GPU Hardware Marketplace as Demand for Refurbished AI Infrastructure Accelerates
apnews.com ∙ Friday, July 17, 2026
Top line
Compute Exchange has launched a new secondary marketplace for refurbished NVIDIA GPUs to help organizations scale AI infrastructure cost-effectively.
Summary
Compute Exchange has introduced a secondary marketplace specifically for used and refurbished NVIDIA GPUs, addressing the growing corporate demand for cost-effective AI infrastructure. By moving beyond its original focus on reserved compute capacity, the platform now facilitates the trading of physical hardware like H100s and A100s. To ensure market transparency and hardware reliability, the service manages the end-to-end procurement process and utilizes SiliconMark for independent component verification. This development caters to a diverse range of buyers—from AI startups to enterprise operators—who are increasingly opting for proven, production-grade hardware to scale their AI workloads efficiently.
Highlights
Compute Exchange has launched a new marketplace dedicated to buying and selling used and refurbished NVIDIA GPUs.
The initiative addresses growing demand for cost-effective AI infrastructure as organizations seek to scale capacity without relying solely on the latest hardware generations.
Market activity shows significant interest in proven production hardware, specifically NVIDIA H100 and A100 models.
The platform facilitates transaction support including sourcing, logistics, and execution while maintaining a market-neutral stance.
SiliconMark is utilized by the platform to provide independent verification of hardware condition and quality for buyers.
Demand for secondary hardware ranges from hundreds to tens of thousands of GPUs per request.
Compute Exchange is expanding its original business model, which focused on reserved capacity, to a comprehensive platform for physical AI hardware procurement.
The service caters to diverse participants, including cloud providers, AI startups, and large-scale enterprise organizations.
Related Items
Compute Exchange Launches Secondary GPU Hardware Marketplace as Demand for Refurbished AI Infrastructure Accelerates
apnews.com ∙ Friday, July 17, 2026
Top line
Compute Exchange has launched a new secondary marketplace for refurbished NVIDIA GPUs to help organizations scale AI infrastructure cost-effectively.
Summary
Compute Exchange has introduced a secondary marketplace specifically for used and refurbished NVIDIA GPUs, addressing the growing corporate demand for cost-effective AI infrastructure. By moving beyond its original focus on reserved compute capacity, the platform now facilitates the trading of physical hardware like H100s and A100s. To ensure market transparency and hardware reliability, the service manages the end-to-end procurement process and utilizes SiliconMark for independent component verification. This development caters to a diverse range of buyers—from AI startups to enterprise operators—who are increasingly opting for proven, production-grade hardware to scale their AI workloads efficiently.
Highlights
Compute Exchange has launched a new marketplace dedicated to buying and selling used and refurbished NVIDIA GPUs.
The initiative addresses growing demand for cost-effective AI infrastructure as organizations seek to scale capacity without relying solely on the latest hardware generations.
Market activity shows significant interest in proven production hardware, specifically NVIDIA H100 and A100 models.
The platform facilitates transaction support including sourcing, logistics, and execution while maintaining a market-neutral stance.
SiliconMark is utilized by the platform to provide independent verification of hardware condition and quality for buyers.
Demand for secondary hardware ranges from hundreds to tens of thousands of GPUs per request.
Compute Exchange is expanding its original business model, which focused on reserved capacity, to a comprehensive platform for physical AI hardware procurement.
The service caters to diverse participants, including cloud providers, AI startups, and large-scale enterprise organizations.