AfterQuery serves frontier labs and tech giants

This customer list shows AfterQuery is selling mission critical post training work, not generic labeling. At roughly $100M in annualized revenue with every US based frontier lab and major tech buyers in the mix, the math points to a business built on a few deep programs where a vendor is trusted with sensitive model failure analysis, expert workflows, and custom environments that can move real capability metrics.

• The product is naturally sold account by account. AfterQuery gets paid to fix a specific failure mode, like a model using the wrong customer support workflow or failing inside a terminal based coding task. That pushes pricing toward multi million dollar programs tied to hard domains, expert labor, and bespoke RL environments, not per seat subscriptions.
  1 sacra
• The buyer set mirrors the rest of the frontier data market, where revenue is concentrated in a tiny number of labs and hyperscalers. Surge generated about $1.2B from roughly 12 frontier labs in 2024, and Turing and Mercor also built large businesses by serving OpenAI, Anthropic, Meta, Google, and adjacent enterprise accounts with expert data and eval work.
  3 sacra 4 sacra 5 sacra 6 sacra
• What separates AfterQuery is where it sits in the stack. Scale and Surge bundle broad labeling and orchestration. Mercor and Turing lean on large expert marketplaces. AfterQuery goes deeper on research style benchmark creation, domain specific evals, and custom sandboxes, which makes it look less like a labor marketplace and more like a specialized capability improvement partner.
  1 sacra 3 sacra 4 sacra 5 sacra 6 sacra

The market is heading toward fewer vendors owning more of each customer's post training workflow. To keep these accounts, AfterQuery will likely turn one off custom projects into reusable finance, coding, legal, and enterprise agent packages, so its foothold inside the top labs becomes a broader data, eval, and environment platform before larger rivals bundle the category.