Author argues AI data vendors lack sustainable demand

The author argues that building a sustainable business in the AI training data market is difficult because data acts as a depleting one-off endowment rather than an annuity. While there are hundreds of data vendors, the buyer pool is extremely narrow, dominated by a handful of major AI labs that spend vastly more on compute than on data. Consequently, this dynamic creates market volatility and leaves vendors with little pricing power or long-term lock-in. Open article →