Surge AI’s Quiet Leap to Success

Multi-billion-dollar valuations in the AI sector are not unheard of. But multi-billion-dollar valuations supported by billion-dollar revenues are not so common. Bootstrapped company Surge AI is one such rare example of an AI-focused organization that has soared in valuation but is also delivering on its financials.

Surge AI’s Offerings

San Francisco-based Surge AI was set up by Google, Facebook, and Twitter alum Edwin Chen. As AI models become more complex, the fact remains that behind some of the most impressive AI systems are people labeling and clarifying data to train it.

After graduating from MIT, Chen worked in various positions at Twitter, Google, and Facebook, where he was involved in content moderation and recommendation algorithms. At each of these places, Chen found that it was difficult to get high-quality human-labeled data at scale.

He left Twitter in 2020 to address this situation and set up Surge AI. He wanted to build a solution which shifted the labeling landscape from low-quality, low-skill mind-set to a richer offering that could capture the range of human skills, creativity and values that AI systems needed to possess. Chen wanted to create a data labeling company that could encode the “richness of humanity.” He did not want data annotation to come only from people who either didn’t get paid enough or did not have the right context to make an informed judgment. He wanted to hire people who understood context and had a deep understan­ding of language. He wanted smart humans to train AI to be able to simplify their personal specialized knowledge into a code that could train LLMs.

With the growing importance of reinforcement learning from human feedback (RLHF) in training advanced AI systems, Surge AI has seen an increase in demand for nuanced datasets. Surge AI has capitalized on this by employing highly skilled contractors instead of large pools of low-wage labor.

Surge AI’s data annotators follow instructions to interact with online chatbots. It uses a “human-in-the-loop” approach, where AI generates its own data and labels it but humans critique its performance. Some of the instructions for Surge’s annotators thus include asking them to coax the chatbot to give out an incorrect or a toxic response. But once the chatbot does that, the annotators then offer a better response. They are even asked to compare different AI responses to the same question and explain why one is better than the other to train the AI into generating a perfect response.

Surge AI’s Financials

Chen believed in bootstrapping and wanted to stay away from third party financing. Surge AI has 250 employees and recorded $1.2 billion in revenues in 2024. Its customer list includes Google, Meta, Microsoft, and Anthropic to name a few. Surge AI claims that it has been profitable from day one.

Surge is not the only player in the market. Its competitors include Scale AI, Labelbox, and Super Annotate. While financials for these companies are not widely published as well, Scale AI is estimated to have recorded $870 million in revenues in 2024.

Scale AI was in the news recently when Meta announced plans to invest $14.3 billion in the company to acquire a 49% stake. The investment valued Scale AI at $29 billion from $13.8 billion that it was valued at in 2024. Meta is counting on the investment, and the hiring of Scale AI’s CEO Alexandr Wang to help accelerate its push into the “superintelligence” research lab. But the move also hurt Scale AI as companies like Google decided to end their relationship with the company. Scale AI’s loss of market ended up being Surge AI’s gain.

For the longest time, Surge stayed away from venture funding. It believes that third party funding invariably leads to unnecessary hiring and a bloated organization. But Surge is now looking to raise $1 billion in funding in a round that will value it close to $30 billion.

Photo Credit: Gerd Altmann from Pixabay