After surging to a high of around $6 in July, the B200 price retreated somewhat, but has recently rebounded slowly to $5.7, still significantly higher than the levels of May and June. This indicates that demand for high-end, latest-generation GPUs remains strong, with the market willing to pay a premium for the B200. Meanwhile, the prices of H100 and A100 have stagnated or declined slowly, with older-generation GPUs not rising in tandem with the B200. This suggests that current demand is shifting from older generations to the latest, and AI computing power pricing is shifting from a "comprehensive shortage" to a "structural shortage of next-generation high-end GPUs."
-------- 1. A100, H100, and B200 are NVIDIA's three generations of GPUs, from oldest to newest. Newer versions of GPUs offer better performance and are better suited for large-scale, complex models, resulting in higher rental prices.
2. Due to the high price and rapid updates of GPUs, many companies choose to lease chips from cloud service providers and data centers to save on investment. Therefore, tracking GPU rental prices can indirectly observe market demand and gauge the health of the AI computing power market.
3. The GPU rental price index primarily tracks daily rental quotes for NVIDIA H100, B200, and A100 GPUs from cloud service providers, data centers, and the private rental market. After removing regional differences and outliers, the price is uniformly converted to "how many US dollars per GPU per hour."
4. An increase in the index typically indicates a tighter GPU supply and stronger demand; a decrease suggests increased supply or cooling demand. By observing price trends and comparing the price difference between new and old GPUs, investors can understand the current state of AI computing power.
