AI and data: the distance between a title and a tool

128,813 data professionals across the three markets. Databricks is declared by 3.6% of them and Snowflake by 3.2%.

The market has data people. It does not, on this measure, have data platform people in anything like the same quantity. That is the single clearest expression in the corpus of the problem this business exists to solve: a title is abundant and a capability is not, and only one of the two is visible in a search.

Across a pool of 22,853 AI and machine learning professionals: Amazon Bedrock 178, Vertex AI 206, Azure AI Foundry 189. Each is under one per cent of the pool.

Those three counts sum to 573 declarations, and that is the number to be careful with. A professional can declare more than one of these platforms, so the count of distinct individuals is lower than 573 and the extract cannot say how much lower. Even taken as an upper bound it is a small number for three national markets.

This is what replaces the adjective. Claims about AI talent nobody else can find are ordinarily unfalsifiable; this is a measurement, with a date on it and a stated method, and it can be checked.

The practical consequence is a change of kind rather than degree. A funnel assumes enough candidates at the top that filtering produces a shortlist. At these quantities there is no funnel to filter. The work becomes identifying named individuals, understanding what they have actually built, and approaching them properly, which is a different activity with different economics.