SILICON Data, a market-intelligence firm that focuses on the cost of graphics processing units, created what it says is the first daily index to track the specialized chips vital to powering artificial intelligence.
The company’s Silicon Data H100 Rental Index tracks the hourly cost of renting a GPU, according to founder and Chief Executive Officer Carmen Li.
“There is zero transparency now for GPU costs,” Li said in an interview. “This is the first step, to help market participants gain transparency into the convoluted world of AI costs.”
Graphics processing units are the workhorse for training AI models. The chips are designed to handle thousands of tasks simultaneously, a concept known as parallelism. A data center may use hundreds or even thousands of these processors, which can cost thousands of dollars.
Founded in 2024 and backed by proprietary-trading firms DRW and Jump Trading, New York-based Silicon Data is building a platform that aims to bring price transparency and structure to the GPU market. The goal is to turn GPUs into a benchmarked asset class that can be traded like any other financial instrument, Li said.
The firm sources its pricing data across markets and technology companies, including so-called hyperscalers and neoclouds, according to Li. The index is normalized based on a variety of factors including GPU types, platforms and hardware configuration. That’s because GPU prices can be affected by a wide range of factors, including their brand, performance metrics, and hardware and software specifications.
Silicon Data’s index is also based on historical data. It re-calibrates based on the importance of each factor at the time. Bloomberg LP, which publishes the index, is the parent of Bloomberg News.
The primary users of the index are banks that license and distribute it for internal risk management, hedge funds that license the underlying data, and asset managers or private equity firms that are financing or building out data centers where GPUs are used.
Though the index itself can’t be traded yet, it provides a benchmark for traders and others to follow as they would a stock index. It also gives insight into the cost of goods for those building AI products or in need of GPU computing power.
It’s about “creating a transparent market and helping every participant understand what their infrastructure is actually worth,” Li said. –BLOOMBERG
The post Silicon Data creates first-of-its-kind index for AI chips appeared first on The Malaysian Reserve.
