The AI compensation premium is small but visible in the corpus.

The AI premium in remote compensation is real, but in our snapshot it is smaller than the headlines suggest — and it lives more in the tails than at the median.

Across the 1,897 active remote listings in the RemNavi corpus that publish a numeric salary band, Data/ML roles post a median of $242,000 per year. Software Engineering, the next role-group up by disclosed-listing volume, posts $237,500. The gap at the median is $4,500 — roughly 1.9%. That is not zero, but it is also not the 30-50% premium often quoted in industry commentary about AI-product compensation. The interpretation has to be more careful.

Where the premium actually shows up is in two places: the upper percentiles, and the named-employer leaderboard.

At the top quartile (P75), Data/ML posts $307,500 against Software Engineering's $286,000. That is a $21,500 spread — five times the median gap, in the same direction. Within Data/ML's 91 disclosed listings, the dispersion is wider: P25 sits at $183,150 (below SWE's P25 of $199,500), and P75 sits at $307,500 (above SWE's $286,000). Data/ML is a wider band centered slightly higher than SWE, not a uniformly higher band. The middle is comparable; the top is genuinely paid more.

The named-employer leaderboard tells the second half of the story. Two AI-product employers — OpenAI (363 listings, $291,000 median) and Perplexity (16 listings, $260,000 median) — sit in the top four positions by disclosed median, despite neither posting the highest individual salary on the board. Postman ($370,000 median, 19 listings) and Cloudflare ($277,500 median, 32 listings) bookend them. The shape: AI-product employers have moved their hiring posture toward consistently higher disclosed bands across more roles, rather than offering one outlier role at a striking number. That kind of move is more contagious — it sets a benchmark candidates can credibly cite in negotiations elsewhere.

The watch-item for H2: whether the Data/ML disclosed share (currently 91 of 1,897 disclosing listings, ~4.8%) grows as ML-engineering teams expand at employers who already have transparency practice. If it does, the median may tick up another $5-10K through composition effect alone. If it doesn't, the AI premium remains a tail effect at the high end of the band — visible to anyone reading the leaderboard, less visible to anyone reading the median.

The next Market Note will surface the first post-EU-directive disclosure snapshot in late June.