DataTribe says confidential computing accelerates GPU compute commoditization into tradeable market
SMH•SMH•SMH•SMH•DataTribe said GPU compute could become a tradeable commodity as AI inference standardizes, citing a $600 billion physical buildout that could support a $6 trillion-$30 trillion derivatives market. It also cited H100 rental rates falling 60%-70% from peak levels.
DataTribe said GPU compute is shifting toward a tradeable commodity as AI inference standardizes around stateless APIs. It cited an estimated $600 billion physical buildout that could support a $6 trillion-$30 trillion derivatives market, while H100 rental rates have fallen 60%-70% from peak levels as supply rises.
DataTribe identified security, uptime, jurisdiction and data sovereignty as constraints on a liquid spot market for GPU-hours. It cited confidential computing as a potential catalyst and Gartner's projection that more than 75% of processing in untrusted infrastructure will be secured “in use” by 2029.
