Image generated by Google’s Nano Banana
Harmonic, the math-focused AI startup founded by Robinhood’s CEO, has reached a $1.45B valuation after a major funding round to scale its provably correct AI models.
What if AI could think like a mathematician instead of a mimic? Harmonic’s new funding suggests that moment is closer than ever.
The AI startup, founded by Robinhood CEO Vlad Tenev, has raised $120 million to accelerate its development of Aristotle, its flagship mathematical reasoning model. The Series C round, led by Ribbit Capital with participation from Sequoia Capital, Index Ventures, Kleiner Perkins, and new investor Emerson Collective, values Harmonic at $1.45 billion.
The investment will fund continued development of Aristotle, Harmonic’s first MSI-class model, which, earlier this year, produced formally verified solutions to five of six International Mathematical Olympiad problems — a first for any AI system and a benchmark that positions the platform at a gold-medal level of mathematical performance.
Aristotle, an AI architecture that embodies Harmonic’s mission to achieve “Mathematical Superintelligence” (MSI), is designed to solve complex mathematical problems with provable correctness and without the hallucinations that undermine traditional generative models.
“Aristotle is now in the hands of mathematicians and researchers, who are using it to accelerate progress across many fields, like cryptography, topology, and scientific computing,” the company said in a blog post. “They are already using it to prove open conjectures and find errors in LLM-generated proofs, ushering in a new era of scientific collaboration and discovery.”
Beyond competitive math, Aristotle can generate provably correct reasoning that could mitigate the limitations of current LLMs, which produce probabilistic outcomes. Its formal verification pipeline addresses that gap by producing results that can be checked end-to-end via Lean4 code that external proof-checkers can formally verify.
This is a significant development that proves large-scale AI can move beyond pattern-matching into domains where precision is non-negotiable, opening the door for AI systems to work directly with scientists, engineers, and mathematicians on problems that once demanded deep specialized expertise.
If Harmonic can scale MSI into broader tools and APIs, organizations could use models like Aristotle for code verification, algorithm design, risk modeling, and mission-critical simulations without relying on opaque reasoning.
Harmonic’s new cash infusion demonstrates that investors are betting that Aristotle’s development of this reliability layer will become foundational as AI workloads move into higher-stakes environments.
As the rush to build global AI infrastructures continues at a growing pace, Harmonic’s new funding also highlights a growing investor appetite for post-LLM architectures that emphasize provability over probability, a shift researchers have been crying out for as new and iterative models flood nearly every aspect of business and scientific development.
Harmonic’s bet on correctness could become a competitive edge in the years ahead, as the capital behind its MSI model, now at $295 million, suggests a growing recognition that raw scale might not be enough.
As enterprises demand systems that can verify rather than trust implicitly, and as researchers push for tools that can withstand academic rigor, MSI is a plausible direction for the next wave of AI innovation.
A recent deep dive into Microsoft, Nvidia, and Anthropic’s “dream come true” AI alliance tracks how frontier models are being scaled for enterprise workloads.
