Someone posted this table saying it lists the top 30 OpenAI customers with consumption over 1 trillion tokens.
Salesforce is at the 4th position. (Nevermind the position: that table was sorted by last name (John E. @ Salesforce was 4th in that order) ! Nonetheless, the point is what the +1T tokens imply.)
IF TRUE... it would indicate that enterprise-scale AI adoption is being done predominantly via Salesforce.
It doesn't look like we are at mere pilot integrations anymore but full workloads.
Processing over a trillion tokens implies continuous use - powering customer-facing or internal intelligence layers with summarization, recommendations, forecasting, sentiment extraction, case classification, etc.
At GPT4 API rates, 1 T tokens would cost around $7 Billion (50/50 blend of input/output).
But according to the latest Salesforce's earnings call, Data Cloud and AI ARR reached $1.2 B in Q2. They must have a significant enterprise discount.
Salesforce is both a consumer and competitor to OpenAI (see Salesforce's own XGen models).
Will Salesforce push to invest further in its own LLMs or tighten Azure/OpenAI partnerships?
Will OpenAI strategically favor Salesforce integrations in its roadmap?
