According to Forbes, citing informed sources, AI, a startup in training data that was established about 18 months ago, is undergoing a new round of financing, with a valuation of $3.2 billion. Compared to its valuation of $300 million 5 months ago, the company has expanded rapidly in a short period of time.
Valuation has increased by more than ten times in five months
Reports show that AfterQuery completed a $30 million Series A financing round 5 months ago, with a valuation of $300 million at that time. The latest round of financing has not yet been finalized, and the company has not announced the lead investor.
Forbes reports that a source stated that AfterQuery has already achieved profitability and has identified lead investors for this round of financing. The company has not commented on this transaction.
- Latest valuation: $3.2 billion
- Valuation five months ago: $300 million
- Previous Series A financing amount: $30 million
Pay to access professional reasoning data.
AfterQuery was founded by 23-year-old Spencer Mateega and 22-year-old Carlos Georgescu. Initially, the two wanted to create an AI agent for financial scenarios, but during testing, they found that mainstream models still fell short in making complex professional judgments.
Thereafter, the company adjusted its direction to provide "reasoning data" for the AI laboratory. This type of data is not ordinary text; rather, it consists of the thought processes through which professionals such as doctors, lawyers, engineers, and financial analysts write down their approaches to problem-solving. It is used to train models to learn how to make judgments, rather than merely memorizing facts.
Mateega Previously stated on the X platform that the company's annual recurring revenue has risen to the "hundreds of millions of dollars" level, which is higher than the 100 million dollars in April of this year.
Institutions such as NVIDIA have become customers.
The report mentions that NVIDIA has used data from AfterQuery to train its open-source Nemotron model. Customers of AfterQuery also include Thinking Machines Lab, founded by former OpenAI Chief Technology Officer Mira Murati, as well as the legal AI company Legora.
This demand is not an isolated phenomenon. As the high-quality training texts that can be directly scraped from the open internet gradually diminish, the demand from AI for data generated by real experts is on the rise. Compared to general web page content, this type of data is more suitable for training models to handle long-chain, specialized, and complex tasks.
Training data track competition intensifies
The shortage of training data is driving a rapid increase in the valuation of related companies. Reports mention that Scale AI, co-founder of a data annotation company, and Alexandr Wang became representatives in this field as early as 2021. Another competitor, Mercor, is currently in talks with NVIDIA for a new round of financing, with a valuation said to potentially reach $20 billion, doubling from $10 billion in October last year.
AfterQuery believes that the difference between them and their competitors lies in the fact that they first use their own software to filter the difficulty of tasks, and then employ their own models to verify the effectiveness of the data, proving to clients that these materials can indeed enhance the performance of the models, rather than simply selling the annotation results.
