The adage about comparing apples and oranges is well illustrated by the many news reports suggesting the “AI trade” is alive and well after quarterly reports from Alphabet and Azure, which show robust cloud computing revenue growth. That is contrasted with the revenue issues OpenAI seems to be having.
While all three are important contestants in the AI ecosystem, their business models, revenue drivers, and cost structures are fundamentally different.
The core of the disconnect lies in the distinction between selling the picks and shovels (infrastructure) and selling the gold (the end-user application).
In other words, the value chain roles are different. Google Cloud and Azure sell infrastructure services (picks and shovels). Their revenue is driven by renting massive amounts of compute operations.
OpenAI’s revenue is based on model operations. Success depends on software sales to end-users.
Also, Google and Microsoft are somewhat vertically integrated: infrastructure operations plus apps.
The "AI Trade" for cloud providers is currently about scale. For OpenAI, the trade is about efficiency (profit margins).
The former is about industrial demand for compute services. The latter is about customer demand for a specific AI model.
To be sure, if end user demand for model services breaks down, so will demand for AI compute services. But OpenAI’s issues seem company specific, essentially revolving around margin issues and growth rates, compared to the supporting investment in compute facilities.
