Ethan Mollick recently shared some intriguing data about AI spending patterns. The numbers, sourced from a Ramp report, seem to paint a rosy picture for OpenAI. But it’s worth digging a bit deeper.

According to the report, OpenAI is experiencing impressive retention and growth numbers. A whopping 82% of companies that spent on OpenAI a year ago are still customers today. Even more striking, these customers reportedly increase their spending by an average of 25% every month in the first year of service.

But as I’ve commented on LinkedIn, the methodologies used by Ramp are somewhat questionable. They note that “Once usage exceeds a certain scale, companies tend to shift payments from cards to AP”. But is this universally true? Mistral, for instance, has always required upfront purchase of credits, just like OpenAI does now. And what about spending on Anthropic, which might be hidden within AWS bills?

The data also raises questions about other AI vendors. As Joel M. pointed out in the LinkedIn thread, the graph shows only OpenAI growing significantly, while others remain relatively flat. Is this a true reflection of value, or merely the result of OpenAI’s marketing buzz as the perceived market creator?

I proposed an alternative interpretation: could this be the distinction between “foundation models” (like OpenAI and Midjourney) versus “narrow AI use-cases”? In other words, selling shovels (foundation models) keeps booming, but revenues from actual digging (specific AI applications) might be tanking.

It’s also worth noting the potential biases in data collection. Pawel Cieslicki raised a valid point about Midjourney’s usage data, which might be skewed due to their Discord-based interface. Without access to internal analytics, it’s challenging to get an accurate picture.

In conclusion, while the data seems to suggest that companies are finding value in OpenAI deployments, we should approach these numbers with caution. The AI market is complex and rapidly evolving, and simplistic interpretations of spending data may not tell the whole story. It’s crucial to look beyond the surface and consider the broader context.