Odaily reported that Alex Svanevik, chief executive officer of on-chain analytics platform Nansen, said the artificial intelligence industry’s bubble may burst when companies begin to use Chinese large models effectively. In his view, the U.S. regulatory environment may limit that process, but the broader trend remains that Chinese models continue to become more efficient and are able to run on hardware that is not at the very leading edge.
Svanevik also noted that global GPU supply is increasing, including chips that are not made by Nvidia. At the same time, rental prices for H100 and H200 GPUs have recently declined, which he described as a reflection of changes in the supply-demand structure for computing power. He raised the question of “how to explain the decline in GPU rental prices” as a market signal.
His comments framed the issue around two developments happening at the same time: improving model efficiency and expanding compute supply. Under that combination, the AI infrastructure market may be entering a repricing phase, with the previous pricing structure around high-end compute facing adjustment as enterprises gain access to Chinese models and more GPU capacity becomes available.