Axiom Math, founded by Hong in March, has quickly become a rising star in the AI industry, raising $64 million in seed funding in September. Hong, 24, has built a team of 17 employees, many of whom come from Meta’s Fundamental Artificial Intelligence Research lab, Meta’s GenAI team, and Google Brain, which merged into DeepMind in 2023.
Axiom is tackling advanced math, which AI researchers consider crucial for achieving superintelligence. “Math is the perfect sandbox for building superintelligence,” Hong told Forbes.
Early passion for math
Growing up in Guangzhou, China, Hong has been passionate about mathematics since childhood. She taught herself English to read advanced math textbooks and attended South China Normal University Affiliated High School. In high school, she was one of only four girls on her provincial math Olympiad team and achieved strong results in competitions such as the Hua Luogeng Cup and the National High School Mathematics League, as reported by Chinese news outlet 36Kr.
While training in math Olympiad programs, she developed a fascination with research math.
“I was always very interested in mathematical discoveries,” she said in a recent interview with The Wall Street Journal. “Olympiad math is a constant dopamine hit, but research math is banging your head against the wall. It’s pain and suffering. I like that part.”
Hong’s talent in math earned her admission to the Massachusetts Institute of Technology (MIT), where she majored in mathematics and physics. There, the first-generation college student excelled, writing nine research papers and taking 20 advanced mathematics courses. In 2023, she won the prestigious Frank and Brennie Morgan Prize, awarded for outstanding research in number theory and probability.
As a Rhodes Scholar at Oxford University, Hong earned a master’s in computational neuroscience before heading to Stanford for a law degree and math Ph.D.
 |
|
Carina Hong, founder of AI startup Axiom Math. Photo courtesy of MIT |
The turning point
On weekends, Hong studied at a coffee shop near Stanford’s campus, where she regularly drank matcha lattes. It was there that she met Shubho Sengupta, an AI researcher at Meta, by chance, who would later join Axiom as its first member. They often discussed the potential of developing AI that could solve the world’s toughest math problems and discover new ones.
During one of her morning runs, as she considered leaving school to start a company, Hong recalled advice from Lisa Su, CEO of chip giant AMD: run toward the hardest problems. She dropped out as soon as Axiom’s seed-funding round closed last summer.
Hong’s company is named after the mathematical term for a basic truth that can be the starting point of an entire theory. She said Axiom’s aim is to build an “AI mathematician” capable of solving complex math problems, generating proofs, and checking its own work.
The idea is to convert mathematical knowledge from textbooks, research papers, and journals into a software program that can create and verify new problems. Ultimately, Hong hopes the AI will propose conjectures, theories that are likely true but not yet proven.
“We’re not building another chatbot that mimics solutions,” Hong said. “We’re teaching AI to prove theorems. That’s a fundamentally different challenge and one worth pursuing.”
The hunt for talent
In less than a year, Hong has recruited top talent from Meta’s FAIR lab, including Sengupta. Other key hires include Francois Charton, who solved a 100-year-old math problem, and Aram Markosyan, an AI scientist who led safety and fairness research at Meta.
Hong’s recruiting success has drawn attention from Silicon Valley and beyond. Ken Ono, a world-renowned mathematician and Hong’s former mentor, joined Axiom to help push the limits of its AI systems. The 57-year-old University of Virginia professor said his decision was driven by intellectual curiosity rather than financial gain. “I’m not doing this for the money,” Ono said, explaining that he turned down more lucrative offers from larger AI companies.
Ono’s task at Axiom is to develop complex problems that challenge the AI and create benchmarks to assess its performance. “Think of it like a map for a sailor,” he said. “Before you set out to discover a new land, you need to know where you are and what’s already been explored.”
Hong’s ambition has been key to attracting top tech talent to Axiom since its earliest days, when the office consisted of a plastic folding table and a friend’s spare couch, according to Business Insider. Hong said the company’s commercial applications extend beyond mathematics, another draw for recruits, including “any domain where you need provably correct reasoning,” such as hardware and software verification, quantitative finance, and cryptography.
“One thing I heard from some of the top researchers and mathematicians I’ve recruited to Axiom is that solving for mathematical superintelligence will be their legacy,” Hong said. “When the problem is hard enough, talent density gets very high, and that makes you a magnet for other great thinkers.”