After a morning spent reviewing a dataset, reading research papers and playing with cutting-edge AI models in his Manhattan Apartment, Edwin Chen takes a short walk to the swank three-story Starbucks Reserve Roastery on Ninth Avenue.

Dressed in a Vuori navy T-shirt with a tiger-adorned canvas tote slung over one shoulder, Chen heads downstairs and settles in at a dark corner table. Sipping a small green tea “because ordering coffee here takes too long,” the founder and CEO of Surge AI, a data labeling and AI training firm, then launches into a nonstop two-hour discussion about everything from Silicon Valley culture (he hates it) to his rivals (“they’re all body shops”) to how humans might interface with aliens if they came to Earth. “They don’t speak English. So how would you communicate with them? How would you decipher their language? Hopefully there’ll be some mathematical way to do it.”

This dilemma is also explored in his favorite short story, a 1998 piece by science fiction author Ted Chiang. “Story of Your Life” became the basis for the movie Arrival, in which a linguist tries to talk to aliens by identifying patterns in their speech and writing. It was also part of Chen’s inspiration for starting Surge in 2020, he says, adding that he wants his data labeling company to encode the “richness of humanity.” For him, that means getting the smartest humans (inclu­ding professors from Stanford, Princeton and Harvard) to train AI, translating their specialized knowledge to the 1s and 0s underpinning large language models. In addition to the Ivy League brainiacs, Chen employs an army of a million-plus gig workers from more than 50 countries around the world who help come up with questions that might stump AI, evaluating the models’ responses and writing criteria that help AI generate a perfect response. “I really do think that what we’re doing is so critical to all the AI models that without us, AGI [artificial general intelligence, tech lingo for when AI will match or surpass human capabilities] just won’t happen,” Chen says. “And I want it to happen.”

Long-winded, brilliant and eccentric, Chen is perhaps the most successful tech entrepreneur you’ve never heard of. That’s because until very recently he wanted it that way, despite being well known in the AI community. The data scientist who did stints at Twitter, Google and Facebook eschewed traditional venture capital and left the Bay Area fishbowl seven years ago, elec­ting to fund Surge himself, starting with “a couple million” of savings from his decade in Big Tech. “One of the reasons why we bootstrapped is that I’ve always hated the Silicon Valley status game,” says Chen, who describes the typical VC-backed Valley startup as a “get-rich-quick scheme.” He also hates the idea of raising so much money and then needing to spend it. In his opinion, that leads to massive overhiring. He points out that Surge has just 250 employees, including full-time, part-time and consultants. By contrast, Scale AI, its big rival, has four times as many staffers with less revenue.

Surge, which helps tech companies get the high-quality data they need to improve their AI models, brought in $1.2 billion in revenue in 2024, less than five years after its founding, from customers including Google, Meta, Microsoft and AI labs Anthropic and Mistral. (It helped train Google’s Gemini and Anthropic’s Claude.) It’s been profitable from nearly day one, accor­ding to Chen. Based on those numbers, the company is worth an estimated $24 billion. Surge is in talks to raise $1 billion at a $30 billion valuation, though the round hasn’t closed yet.

Chen’s decision to fund Surge himself has paid off handsomely: His approximately 75% stake in it is worth an estimated $18 billion, enough to make him the wealthiest newcomer on this year’s Forbes 400 list of the richest Americans. At age 37, he is also the youngest member.

Chen is just as likely to ask someone in an interview to discuss David Foster Wallace or linguistics as he is to ask them to code or problem-solve on a whiteboard. “We value creativity,” he says.

Surge claims its approach isn’t like older forms of data labeling, in which people—often from less developed countries in the Global South—are paid pennies per hour to sit in front of compu­ters and identify the difference between a cat and a dog. Instead, Chen’s data annotators, which include professionals and professors, follow........

© Forbes