


Sitting in a conference room at Mercor’s office in San Francisco’s South of Market district, CEO Brendan Foody recalls the day in June that changed everything. Meta had announced it was buying almost half of data labeling giant Scale AI for $14 billion and poaching its star CEO Alexandr Wang. Mercor, a smaller rival that recruits PhDs and other experts to train models for AI labs, saw an immediate opening.
“I was initially surprised,” Foody tells Forbes. “Then it slowly transitioned from surprise to excitement and enthusiasm around the future.” His cofounder and CTO Adarsh Hiremath chimes in: “It just doesn't happen too often in startups where your biggest competitor gets torpedoed overnight.” That’s because Scale’s tie-up with Meta means many of the big AI labs, worried about a loss of neutrality, no longer want to work with them, the founders argue. During the conversation, Hiremath struggles to contain the excitement of his new pet, a 6-week-old Bernese Mountain Dog named Zeus.
The puppy’s exuberant energy is an apt metaphor for Mercor itself, a company that has become one of the AI era’s poster children for Silicon Valley’s twenty-something founders. At age 22, Mercor’s leaders Foody, Hiremath and Surya Midha are all Thiel Fellows, members of conservative billionaire investor Peter Thiel’s program to dole out $100,000 grants every year to young people in exchange for foregoing college. The three longtime friends, who met on their Bay Area suburb’s high school debate team, started Mercor in 2023 before their fellowships.
The buzzy startup, backed by heavy hitters including blueblood venture firm Benchmark, Twitter cofounder Jack Dorsey and former U.S. Treasury Secretary Larry Summers, is a company in transition. Last month, it hired former Uber product chief Sundeep Jain as the startup’s first president, the company told Forbes, injecting tech-giant experience into the neophyte operation. Mercor will soon move to Instagram’s old offices in downtown San Francisco, a bigger space for its fast growing team. And this year the company made its debut at No. 89 on Forbes’ Cloud 100 list, our ranking of the world’s top private cloud computing companies.
Even Mercor’s primary business has changed — sort of. The company wasn’t always focused on data labeling, and it’s still not the long term goal. Its original conception was modernizing recruiting with AI and high-tech matching algorithms. Mercor built a platform where applicants were interviewed by an AI avatar and placed with a company looking to hire talent, like a next-gen Deloitte or Accenture. Then OpenAI’s ChatGPT, released only a few weeks before Mercor’s founding, touched off a race among the tech giants to train the most advanced AI models. Mercor, which means “market” in Latin, found its sweet spot staffing people to train those AI models, and tilted the entire company in that direction.
Don’t call it a pivot, Foody insists, saying Mercor still sees itself as a recruiter. Data labeling still fits with Meror’s core purpose of matching workers with companies. Whatever you call it, it’s clearly paid off. In March, Mercor said it had $100 million in annualized revenue run rate, and now tells Forbes it made $6 million in profit in the first half of the year. The company said it’s grown nearly 60% every month for the past six months. Foody declined to reveal how much of the business data labeling now represents, but said it’s the “core growth driver” of the company.
The data labeling space is already crowded, even with Scale’s diminished stature. Surge, an older company founded in 2016, is reportedly valued at $25 billion. Turing AI, valued at $2.2 billion, raised $110 million in July. And Invisible, a smaller firm last valued at $500 million in 2023, has become a go-to partner for OpenAI and Microsoft.
Reached for comment, Scale AI pushed back against Mercor’s assertions. “It’s not surprising that Mercor is spreading lies about its biggest competitor,” Scale AI spokesperson Joe Osborne said in a statement. “We remain independent and neutral, and have said so publicly many times.”
Foody says Mercor stands out from the pack because of its focus on providing clients with highly-skilled experts, like PhDs or lawyers, who typically make between $90 and $150 an hour. That type of expertise is necessary for training the most advanced reasoning models, which often need domain experts to teach them how to “think” through multi-step requests. (The founders’ favorite staffing placements for AI model trainers include a chess grandmaster and private detective.) Foody also touts Mercor’s matching algorithms, which allows the company to pick the best person for each specific project, he said.
Mercor says it’s made inroads with the top AI labs seeking model trainers, including OpenAI. Applied Compute, a new startup focused on AI reinforcement learning that was founded by ex-OpenAI staffers, uses Mercor to provide datasets on specific topics, like finance. “They attract the caliber of talent that a lot of these other platforms aren’t able to,” said Yash Patil, CEO of Applied Compute, who met the Mercor founders when they were all Thiel Fellows together.
The company has its detractors. One rival data labeling CEO praised Surge for its execution and high volume, while saying Mercor wasn’t on their radar as much when it came to big deals with customers. “I see them a lot less,” the CEO said. However, they agreed that Mercor’s bread and butter is in staffing highly-skilled experts.
The focus on expert-quality data has gotten the attention of investors. Mercor raised $100 million in February from backers including Benchmark and General Catalyst, vaulting its valuation to $2 billion, eightfold from $250 million just a few months before. Now, Peter Fenton, a Benchmark partner who was an early investor in Twitter and Yelp, is joining the board, after Victor Lazarte, the previous board member representing Benchmark, announced in July that he’s leaving to start his own fund. “I think Mercor has proven itself as the highest quality data provider,” Lazarte told Forbes. “The people building models understand that the quality of your data is more important than the amount of data that you need.”
There are pitfalls to letting AI weigh in on hiring decisions. In general, language models can generate biased responses based on training data. Foody argues Mercor protects against bias even more than traditional recruiting, because its AI isn’t allowed to see certain identifiers about candidates, including name, gender and race. Foody said the company also isn’t using interview data to train its AI recruiter, a concern of some detractors. “One of the main reasons not to train on interviews is that, on a standalone basis, they're not super useful,” said Foody, adding the company would need an amount of data closer to the size of the whole web for it to be valuable.
When Scale’s sale to Meta sent shockwaves through the industry, it also cast doubt on the future viability of data labeling as a business. Observers wondered why Wang, the CEO of the sector’s most dominant player, would suddenly jump ship if he didn’t see something foreboding on the horizon. Foody takes the opposite view: The Meta deal validated the industry, he argues, since the social giant found it valuable enough to pony up billions for a 49% stake. Human AI trainers will be necessary as long as there are complicated concepts to teach models, and he’s betting it will be a long time before that business dries up.
Now Mercor is trying to get to the next level. When the company hired Jain as president last month, he became the proverbial “adult in room” — like when Google brought in Eric Schmidt to coach young founders Larry Page and Sergey Brin. Jain said his main goal is to scale all of Mercor’s systems. That includes onboarding and management processes, as well as creating better systems for tracking and reporting data for customers. As he settles in, he’ll also have to get used to the youthful environment. “I’m significantly increasing the average age,” said Jain, 54, laughing.
That’s certainly evident in Mercor’s headquarters. On a recent Friday afternoon in July, the office is bustling. Framed inspirational quotes hang on the wall, from both tech icons and Mercor engineers alike. “No significant innovation has ever come out of a large company,” says a poster quoting Vinod Khosla. It hangs a few feet away from another quoting a rank-and-file Mercor engineer: “Welcome aboard the rocket ship.” Copies of Peter Thiel’s Zero to One are scattered throughout the office. On an office bar cart beside bottles of Dom Perignon and Don Julio 1942 are rows and rows of Monster energy drinks and a lone canister of Loaded Hot Dog Pringles.
Foody, who grew up in Menlo Park, always had Silicon Valley in his blood: his mom worked for Meta’s real estate team and his dad founded a graphics interface company in the 90s before turning to startup advising. One of his first ventures, started as a high schooler at age 16, was a company to get his friends promotions on Amazon Web Services, the ecommerce giant’s cloud platform, charging them $500 each. Foody reaped the benefits of his parents’ entrepreneurial wisdom during dinner table conversations. His dad’s advice: create jobs and sell to “rich customers in pain,” instead of hawking products to your friends, Foody recalled. “We joke that Mercor epitomizes both of those. We create jobs and we sell to the wealthiest customers.”
The trio originally started the company to match engineers in India with U.S. companies in need of freelance coders. Intrigued investors began wining and dining them. “They’re not just prodigies,” Benchmark’s Fenton told Forbes. “They’re forces beyond categorization.” To court the founders, he whisked them away on a helicopter tour of San Francisco. Felicis Ventures, another backer, flew the founders to Las Vegas in a private jet to race Ferraris around the F1 track. Foody finished the race first, said Felicis partner Sundeep Peechu, who recalled him taking turns aggressively. “The appropriate risk-taker at the company should be the CEO.”
Thiel, who invested in the company’s series A, has also opened up his network. Foody recalled attending his politics-themed Christmas party, where guests could choose to either take a red pill or a blue pill — a play on The Matrix that has come to symbolize the chasm between left and right wing politics. (Foody said he took neither pill.)
Mercor’s longer term goal is to be able to match every person with an appropriate job for them. In the future, the company wants to be able to place lawyers at law firms and doctors at hospitals. For now, though, the money is in training AI, and it looks like it will be that way for the foreseeable future as Scale becomes a presumably less looming threat. “They've left a very meaningful void in the space,” said Foody.