GPU Rental Index: Silicon Data Launches Pricing Tracker
- Silicon Data has launched the SDH100RT, a new index designed to bring price transparency to the GPU rental market.
- The SDH100RT tracks the spot rental price of Nvidia H100 GPUs, crunching 3.5 million data points daily from over 30 sources worldwide.
- Li said she envisions compute becoming a critical resource, necessitating sophisticated risk management tools.
Silicon Data’s SDH100RT, a groundbreaking GPU rental price index, is now live, offering unprecedented transparency in the AI development sector. Founded by Carmen li to combat opaque pricing, this index tracks Nvidia H100 GPU spot rental prices globally, pulling data from over 30 sources to deliver a real-time benchmark. This initiative promises to lower AI development costs, improve risk management, adn potentially lower the barriers to entry for smaller companies—a crucial step forward for the industry. Discover how this data-driven approach reveals regional price differences, like the East Coast’s lower costs, and sheds light on the impact of choice chips. Gain valuable insights, courtesy of News Directory 3 and learn how the SDH100RT will transform how AI projects are financed. Discover what’s next as this index expands to cover more GPUs and fuels innovation.
Silicon Data Launches GPU Rental Price Index for AI Transparency
Updated May 30, 2024
Silicon Data has launched the SDH100RT, a new index designed to bring price transparency to the GPU rental market. The company’s founder,Carmen Li,a former chief data wrangler at Bloomberg,believes that the lack of price transparency in AI development has hindered the industry’s growth and made financing more arduous.
The SDH100RT tracks the spot rental price of Nvidia H100 GPUs, crunching 3.5 million data points daily from over 30 sources worldwide. This AI models price index aims to provide a benchmark for GPU rental costs, enabling better risk management and potentially lowering the barrier to entry for smaller AI companies.
Li said she envisions compute becoming a critical resource, necessitating sophisticated risk management tools. She believes the index will lead to cheaper AI chips tools and broader participation in the AI industry.
Before founding Silicon Data, Li observed that manny small companies struggled with unpredictable GPU costs, impacting their profit margins. the SDH100RT addresses this by providing a foundation for financial products, like futures contracts, that can hedge against price swings.
While the index currently focuses on the Nvidia H100, Silicon Data plans to develop indexes for other GPUs, including the Nvidia A100 and chips from AMD and Nvidia’s Blackwell series.
“I really believe compute will be the biggest resource for humanity in the next few years. If my thesis is right, then it will need more sophisticated risk management.”
Carmen Li, Silicon Data founder
Silicon Data’s data reveals insights into GPU rental pricing. Such as,East Coast H100 rental prices have been consistently lower than those on the West Coast. Also, Amazon Web Services’ custom chips have lowered prices for its customers, with Trainium2 units costing less than half the price of an Nvidia H100.
The company also found that the introduction of DeepSeek’s LLMs had a limited impact on spot rental prices. Interestingly,systems with Intel CPUs commanded a premium over those with AMD processors,depending on the interconnect technology used.
What’s next
Li hopes the SDH100RT will enable banks to fund a wider range of AI projects and develop financial products to mitigate risks in the AI industry, fostering further innovation and growth in startups.