Mathematics-focused artificial intelligence startup Axiom Quant Inc. stepped out from the shadows today to announce it has raised $64 million in a funding round that values it at $300 million.

The lead investor in the round was B-Capital, with Greycroft, Madrona Ventures and Menlo Ventures also participating.

The startup, which does business as Axiom Math, is looking to build what can only be described as an “AI mathematician,” or essentially a large language model that’s designed to focus on the hardest math problems and solve them with unprecedented accuracy. To do this, it will also need to validate its work, which is why it’s designed to generate detailed proofs of each different step it takes as it arrives at its answers.

Ultimately, what Axiom Math wants to do is take all of the world’s math textbooks, journals and scientific papers and bundle them into an advanced software program that will then be able to create newer math problems, together with verifiable solutions that can be tested by humans.

The startup was founded by Stanford University Ph.D. student and math wizard Carina Hong and former Meta Platforms Inc. AI researcher Shubho Sengupta, who helped develop that company’s Llama LLMs and write software tests for them.

Hong has an impressive background. She obtained a bachelor’s degree in mathematics and physics from the Massachusetts Institute of Technology, where she authored more than nine research papers, and later earned a master’s degree from Oxford University. In 2023, she was named as the winner of that year’s Frank and Brennie Morgan Prize, which is an annual award given to undergraduate students for superior mathematics research.

Hong told Forbes in an interview that she met Sengupta for the first time at a coffee shop near Stanford University, where they spent hours discussing the intersection of AI and math and the idea of creating a model that could specialize in the latter. Naturally, they decided to build it, and from there, Axiom Math was born.

The co-founders hope that their model will be able to create “entirely new knowledge” by proposing new theories that appear to make sense, but have not been proven mathematically. “Math is the perfect sandbox for building superintelligence,” Hong said.

Axiom Math faces some tough competition in the field of AI math, notably from AI giants such as OpenAI and Google LLC’s DeepMind, which have both achieved standout results at the International Math Olympiad. That competition, which is designed to test the knowledge of human mathematicians, involves solving six extremely complex math problems, and both OpenAI and Google managed to solve five out of six.

But Axiom Math has bigger ambitions than just solving complex math equations. While it’s currently focused on creating and solving new problems, it hopes that its work can later be applied to all kinds of industries, such as crunching numbers in financial services, designing airplanes and computer chips, and building quantitative trading algorithms.

B-Capital Partner Ida Girma said in a blog post announcing the round that he sees math as the next frontier for AI, noting that while LLMs can often produce “astonishing results,” they’re still extremely unpredictable because they make so many errors in their reasoning processes.

“True precision in quantitative intelligence may well require reimagining reasoning within the sandbox of mathematics,” he said. “Axiom is leading this bold approach. By combining AI, programming languages and mathematics, it is creating the foundation for verified quantitative reasoning that’s provably correct.”

Image: SiliconANGLE/Meta AI