Silicon Data Reports High Demand for Older AI Chips
The firm highlights a robust market for previous-generation GPUs as supply chain constraints persist.
Updated on Oct. 21, 2026 in Semiconductors
Live Poll
Is now a good time for businesses to invest in older technology instead of the newest?
Silicon Data CEO Carmen Li announced strong market demand for older AI hardware, driven by high costs associated with newer GPU models. The company also confirmed it secured $30.5 million in Series A funding to expand its tracking services.
Why it matters
Persistent chip shortages have kept prices for top-tier hardware elevated, forcing buyers to seek previous-generation alternatives. As compute costs rise, businesses are increasingly relying on rental markets to access high-performance AI processing power.
On-demand H100 rental rates start at $2 per hour, while newer B200 GPUs command approximately $5 per hour. Silicon Data launched its H100 Rental Index in May 2025 to track these fluctuations.
The players
Carmen Li
She serves as the CEO of Silicon Data, a market intelligence firm focused on the AI compute hardware industry.
Silicon Data
This startup provides data and indices on GPU rental rates and hardware valuations for the global AI compute market.
CME Group
This global markets company provides derivatives and clearing services across a wide range of asset classes.
The details
Silicon Data tracks GPU rental rates and residual values to provide market intelligence for AI compute users. The startup is now preparing to launch cash-settled GPU futures and options in collaboration with CME Group to manage volatility in the sector.
Timeline
April 2024: Silicon Data was founded.
May 27, 2025: The company launched the H100 Rental Index.
August 2026: The firm secured $30.5 million in Series A funding.
Mid-October 2026: H100 residual values were recorded at 61% of purchase price.
Fall 2026: Silicon Data plans to launch GPU derivatives with CME Group.
The Tech Race
The emergence of financial derivatives for GPU compute marks a shift toward treating AI hardware as a tradable commodity similar to energy or precious metals. This follows the industry-wide focus on the Nvidia H100 GPU, which has become the de facto benchmark for AI infrastructure scaling.
Companies looking to deploy AI models may face significant cost fluctuations when renting cloud compute power. Understanding these rental indices can help businesses predict their operational expenses and plan hardware procurement more effectively.
The takeaway
The maturation of the GPU rental market reflects the growing importance of AI compute as a foundational business cost. Organizations should prepare for continued price volatility in hardware by utilizing market intelligence data to guide their infrastructure investments.
What happens next
Silicon Data is scheduled to launch cash-settled GPU futures and options in partnership with CME Group during the fall of 2026.
Further reading
For more information on the evolving landscape of chip supply, visit the Semiconductors section.
Live Poll
Is now a good time for businesses to invest in older technology instead of the newest?
