AfterQuery reportedly reaches $3.2B valuation in record YC ascent

Introduction

The market for AI training data is moving beyond the question of whether a model can produce the right answer. A growing group of startups is targeting a more demanding goal: teaching models and agents how professionals actually approach and complete work. According to reporting from TechCrunch, citing Forbes, AfterQuery has reportedly completed a new financing round at a valuation of $3.2 billion. The company had not confirmed the transaction or provided details when the report was published, so the financing should still be treated as unverified.

If the valuation is accurate, AfterQuery would be the fastest company in Y Combinator’s history to move from launch to unicorn status, according to YC partner Gustaf Alströmer. Its founders are now 22 and 23 years old and joined Y Combinator’s Winter 2025 batch roughly 18 months ago. In April, the San Francisco startup announced a $30 million Series A at a $300 million valuation. The reported new valuation would represent an increase of more than ten times in only five months.

Key points

• The financing remains a reported event: AfterQuery had not immediately responded to requests for comment. Investors, round size, and deal terms have not been disclosed in the available material.
• Early commercial traction is notable: In April, the company said it had reached a $100 million annualized revenue run rate and was working with several leading AI labs.
• Its named customers span the AI ecosystem: The company has identified Nvidia, legal technology company Legora, and South Korean AI lab Motif Technologies as customers.
• Its training proposition is different: Rather than only using experts to judge whether answers are correct, AfterQuery aims to encode the patterns, decisions, and reasoning that professionals use to complete tasks.

Why it matters

AfterQuery belongs to a broader category of expert-in-the-loop data companies associated with startups such as Mercor and Scale. As foundation models improve at general knowledge and routine question answering, the next bottleneck may be the ability to execute complex, domain-specific workflows. A model may need to understand the objective, select appropriate steps, and produce work in a way that resembles a doctor, lawyer, or other specialist—not merely return a plausible response.

That shift could expand the role of training-data providers. Their value may depend on access to qualified professionals, careful task design, evaluation systems, and the ability to turn expert behavior into repeatable training signals. It also creates a potential bridge between model development and agent deployment, where success is measured by completed work rather than isolated answers.

The reported valuation nevertheless raises important questions. The available information does not establish how durable AfterQuery’s revenue growth is, how large or concentrated its customer relationships may be, or whether expert-generated training processes can become a lasting competitive moat. The company’s next challenge will be converting rapid fundraising momentum into reliable delivery, repeatable workflows, and sustained demand.

Regardless of the final transaction details, the episode points to a broader change in AI infrastructure. Training data is increasingly being treated not as static labeling, but as a product layer that connects model capabilities with real professional processes.

Source: TechCrunch AI