Exclusive | 25-Year-Old Hong Letong Leading Team, Axiom Just Raised $140 Million

The investment community has learned of the latest funding round—

Hong Letong’s AI startup Axiom announced the completion of a $200 million Series A funding round, led by Menlo Ventures, with continued investment from Greycroft, Madrona Venture, B Capital, Toyota Ventures, and other existing shareholders. The company’s valuation now reaches $1.6 billion (approximately 11 billion RMB).

Hong Letong is 25 years old, born in Guangzhou, and attended South China Normal University Affiliated High School, winning multiple awards in math competitions. She later studied at MIT and Oxford University, and during her PhD at Stanford, she chose to dive into entrepreneurship.

The investment community has previously had private conversations with Hong Letong, when Axiom was still in its early stages. I recall her writing in her social circle during Axiom’s founding: “Wishing myself to be both a flower and a tree. Colorful and passionate, standing alone and proud.”

Image Source: MIT

Beyond individual choices, a more prominent era is emerging—post-2000 founders are collectively stepping onto the AI stage.

Just raised 1.4 billion

Founded less than a year ago, valued at 11 billion

The story began with a conversation in a café.

It was a late autumn weekend in 2024. 23-year-old Hong Letong met Shubho Sengupta, then Director of Meta AI Research, at a café near Stanford.

During several hours of discussion, they explored intersections of their research fields and how AI might solve some of the world’s most challenging mathematical problems. Soon after, Hong Letong made a calm yet resolute decision: to drop out of Stanford and start Axiom.

Stories of prodigies always come with a halo. Quickly, Hong Letong’s name spread quietly and rapidly in the venture capital circle. At that moment, she rarely talked about trends or disruption; instead, she focused on her curiosity about the problems themselves and her clear judgment of technological possibilities.

Today, Axiom announced it had completed a $200 million Series A round, reaching unicorn status less than a year after its founding.

Just five months earlier, Axiom had completed its seed round. At that time, the startup conveyed a firm belief: mathematics is the correct foundation for building AI reasoning capabilities.

In December last year, Axiom’s core system scored a perfect 12 out of 12 on the Putnam Competition, often called the “Olympics of Undergraduate Mathematics.” Only five people had achieved this in nearly a century. Subsequently, the system autonomously proved several open conjectures in number theory.

Solving mathematical problems is just the tip of the iceberg. Axiom is transferring this “absolutely correct” mathematical reasoning ability into the domain of code verification through transfer learning, aiming to address various issues in generative AI.

Currently, the AI industry faces a sharp contradiction: while the capabilities of large models are skyrocketing, their reliability remains a black box. Nonsensical outputs in daily applications may be harmless; but in finance, national defense, or critical infrastructure, any probabilistic errors could lead to disaster.

“We are entering a world where AI can almost write all software, but behind it hides an almost unmentioned problem: all code is unverified, which is a huge risk,” said Matt Kraning, partner at Menlo Ventures. “AI will write all code, but mathematics will verify whether it’s correct.”

Axiom’s answer is Verified AI.

Simply put, its core advantage lies in formal verification: using the Lean programming language to convert mathematical proofs into executable programs, fundamentally ensuring correctness. Here, large models no longer guess answers based on probabilities but transform code into rigorous mathematical logic to prove results. Every step of reasoning must pass the test of a deterministic verifier.

“Scientific breakthroughs often require two steps: proposing hypotheses and then verifying them with proofs. We founded Axiom to infinitely compress the time from curiosity to truth,” said Hong Letong. She believes AI’s recursive self-evolution is imminent. Regarding Axiom’s ultimate mission, she left a powerful note:

“Verified AI is not just about fixing AI’s flaws. The real goal is to push AI’s limits and serve as a stepping stone toward superintelligence.”

A Post-2000 Female Scholar Leading the Team

From Guangzhou

The story of Axiom cannot be separated from the unstoppable rise of Hong Letong.

Born in 2001 in Guangzhou to Chaoshan parents, she showed extraordinary mathematical talent from a young age. She attended South China Normal University Affiliated High School. In high school, she was the only female among four members of the Guangdong provincial team in math Olympiads, achieving top results in competitions like the Hua Luogeng Cup and the National High School Mathematics League.

In 2019, she was admitted to MIT, studying mathematics and physics double majors. After three years, she received the Rhodes Scholarship to Oxford University, becoming one of only four Chinese recipients. The Rhodes Scholarship is one of the world’s oldest and most prestigious international awards, often called the “Nobel Prize for undergraduates.”

Later, she earned a master’s degree at Oxford. During that time, she also researched AI and machine learning at University College London, expressing her interest: “What will be the future interaction between AI and scientists? That’s the research I want to pursue.”

By August 2024, Hong Letong was pursuing a PhD in Mathematics and Law at Stanford. “I’ve always been a researcher,” she said in an interview, emphasizing her desire to solve truly difficult technical problems. Soon, she entered the AI world while still a doctoral student.

During the global buzz around DeepSeek, she shared her thoughts: “A small, focused, independent team. A group of idealists working together. They are highly capable and hands-on. The most precious thing is the intertwined belief of ideals and mission. That’s the story of DeepSeek, and it’s also the story I want to write myself.”

Today, Axiom has assembled a core team that can be called a “dream team”:

• CTO Shubho Sengupta, former Meta AI Research Director, involved in Google’s distributed training systems, and one of the earliest developers of CUDA technology;

• Lead Scientist François Charton, a pioneer in introducing Transformer models into mathematics, who used large models to overturn a 30-year-old open conjecture;

• The most impressive is the addition of founding mathematician Ken Ono, a renowned expert with Guggenheim and Sloan awards, former vice president of the American Mathematical Society, who resigned from his tenured position at the University of Virginia to join Axiom full-time.

A well-known story is that during Ono’s 40-year teaching career, he mentored ten Morgan Prize winners, among whom Hong Letong is one.

“She has a solid mathematical foundation, incredible operational efficiency, and the ability to attract top talent worldwide,” said an investor. “This is the most impressive founder I’ve seen in Silicon Valley over the past twenty years, without exception.”

Currently, Axiom has over 30 employees, with hiring accelerating. Her story has only just begun.

The AI Era

Post-2000s Collective Debut

The AI wave has become the stage for the post-2000 generation.

Just last week, Lingchuang Intelligence, founded just over a year ago, announced it had completed a total of 2 billion RMB in angel and Pre-A funding, with a star-studded investor lineup. Co-founder Chen Yuanpei, born in 2001, studied at Peking University and Stanford, mentored by Fei-Fei Li.

Yong Fengyu, founder of Youliqi, is also a post-2000. She holds a PhD from Yale University and a bachelor’s from the University of Michigan in computer science, having worked at DeepMind. Last week, her company also completed nearly 300 million RMB in a new funding round.

The list goes on. AI programming companies founded by four post-2000s—Anysphere, with a valuation of about $29 billion; 19-year-old Chinese-American Serena Ge’s data company Datacurve, which raised over $100 million; and Mercor, with three 22-year-old co-founders, becoming some of the youngest self-made billionaires globally.

A more direct impact is seen in the billionaire rankings: AI has become the most youthful track. The 2026 Hurun U40 Global Self-Made Billionaires List shows 27 AI entrepreneurs with over $1 billion, accounting for a quarter of the list, with an average age of just 32, totaling nearly 700 billion RMB in wealth.

Witnessing these waves of change, many investors are impressed: in just two or three years, these young AI entrepreneurs have demonstrated vision, resilience, and maturity beyond their age—arguably the most outstanding group they’ve seen in their careers. Some even admit that they wouldn’t invest if there wasn’t a core post-2000 partner in the team.

Why must it be young people? Fu Jixun, managing partner at GGV Capital, explained: “Every era’s innovation is often led by the young because they have no baggage.” Each era has its own innovations and inertia, sometimes even a burden. To some extent, today’s AI is an opportunity for the post-95 and post-00 generations.

On the eve of technological restructuring of the world, having no baggage is often the best luggage.