A Precursor Ventures panel (YouTube video) debating whether data-as-a-service is venture-backable. Speakers argue many data entrepreneurs overestimate market demand for raw datasets; data tends to be valuable primarily when transformed into inputs for other business processes (e.g., ML training data or inputs for self-driving stacks), rather than as a standalone product.
But to Oren's point, my sense has been a lot of data entrepreneurs end up being overly optimistic about the universe of people who will see value in their data. And some of it feels very like an ex post rationalization. It's like, well, because I've gone through all this effort to collect this data, there must be people who will find it valuable. And if I think about what is what appears to be valuable today, it's the ability to transform data into some other useful input. Whether that's hey, we have a bunch of training data that you can use to train your LLM. We have a bunch of data we can use to train your self driving car. It appears that data that enables some other business process is valuable as an input, but as an input to another business, not as a business in and of itself. And.
