Midas Launches with $10M Funding to Mathematically Secure AI Systems

Backed by investors from OpenAI, Tesla, and SpaceX, Midas has launched with $10 million in funding to mathematically secure artificial intelligence systems. The new company, founded by 11 medalists from the International Mathematical Olympiad and Informatics, aims to move AI beyond simply appearing reliable to being demonstrably, provably correct. “Modern AI produces fluent, convincing answers, but it cannot prove they are correct,” explains Shalim Monteagudo-Contreras, President and Co-Founder of Midas. Midas intends to build the infrastructure for “proof-native” AI, targeting critical applications in biotech, defense, and finance where errors are unacceptable, and establishing a new standard for trust in a rapidly evolving technological landscape.

IMO/IOI Medalists Launch Midas for AI System Verification

A new company, Midas, is tackling the critical issue of reliability in artificial intelligence through rigorous mathematical verification, following a $10 million funding round led by Valor Equity Partners and Nova Global. The team’s expertise originates from institutions including Stanford, MIT, and Cambridge, and prior experience at companies like Jane Street, Google, and NVIDIA. Midas is designed to address a fundamental flaw in current AI development: the inability to guarantee accuracy.

The company’s approach focuses on embedding mathematical evidence at the core of AI systems, verifying outputs, data, and reasoning before deployment, particularly in high-stakes environments like biotech, defense, and financial systems. According to Renzo Balcazar, CEO and Co-Founder of Midas, “Every human institution, from law to science to finance, runs on evidence,” a standard currently absent in widespread AI application. John Staton, Vice President at Valor Equity Partners, emphasizes that “Verification is the final missing layer,” and highlights Midas’s unique culture, stating, “What sets Midas apart is its culture: a team trained to reject ‘almost correct’ answers and accept only what can be demonstrated.”

Formal Verification Replaces Probabilistic AI Outputs

Midas, a newly launched company, is pioneering a shift in artificial intelligence development, moving beyond systems that appear correct to those that are mathematically proven to be reliable. The core innovation lies in enforcing mathematical correctness before results are implemented, rather than retrospectively identifying errors. Rodrigo Porto, Tech Lead at Midas, emphasizes that verifying reasoning from the outset is essential for trust in increasingly complex systems.

Targeted Deployments: Biotech, Defense, and Financial Systems

Midas is swiftly targeting sectors demanding absolute reliability from artificial intelligence, initiating deployments within biotech, defense, hardware design, and critical financial systems—areas where demonstrable correctness isn’t merely desirable, but essential. The company’s approach centers on building “mathematical trust infrastructure” that verifies AI outputs, training data, and reasoning processes from the outset. Carlo Agostinelli, founder of Nova Global, believes this work positions Midas to become “a generational company,” tackling the fundamental challenge of trust in AI and establishing a “proof-native approach” to reliability.