Midas Raises $10M in Funding to Expand AI Verification Services

There is a moment in every technology cycle where the noise drops out and you hear the math breathing underneath. That is where Midas shows up. A New York based AI security company born at Duke Kunshan University, now stepping into the public eye with a $10M raise and a very specific obsession: proof. Not vibes. Not probabilities dressed up as certainty. Proof. The kind that holds when the room goes quiet and the system actually matters.

Renzo Rodrigo Balcazar Tapia, CEO & Co-Founder, does not talk about intelligence like a parlor trick. He talks about evidence. About how law, finance, science, and infrastructure all run on things you can point to, test, and stand behind. Artificial intelligence skipped that step. Midas did not. Alongside Shalim Monteagudo, CTO & Co-Founder, the company built a verification layer that treats AI outputs the way engineers treat bridges. If it cannot be proven correct, it does not ship. That mindset alone explains why this company exists.

The funding round was co-led by Valor Equity Partners and Nova Global, and it reads less like a bet and more like a recognition. Valor’s John Stanton, VP, called it the missing layer, and that is not marketing language. It is structural language. When AI systems start touching biotech pipelines, defense infrastructure, hardware design, financial systems, and cloud architecture, correctness stops being a preference and becomes the job. This is where money gets serious, because failure gets expensive fast.

Midas is built by a team stacked with 10–11 IMO and IOI medalists, drawn from places like Cambridge, MIT, Princeton, Duke, and Stanford, with scars from Jane Street, Google, AWS, NVIDIA, and Mercor. That matters because formal verification is not a slide deck concept. It is deep work. It is slow thinking done at speed. Rodrigo Porto, Tech Lead, put it plainly at launch: verification cannot be bolted on after the fact. It has to live at the center, or it does not count.

The name Midas is not subtle, and it should not be. In the old myth, everything turned to gold. In this version, everything that passes through gets measured, tested, and proven before it is trusted. In a market drowning in confident outputs and fragile foundations, that is not just a product. It is a filter. And for anyone deploying AI where mistakes are not academic, that filter starts to feel less like insurance and more like gravity.