TechCrunch: Recursive Superintelligence launches with $650M to build self-improving AI

On May 14, 2026, TechCrunch reported on the launch of Recursive Superintelligence, a startup founded by Richard Socher — previously founder of You.com and Chief Scientist at Salesforce — with $650 million in funding. The company’s goal is to build AI that can autonomously improve itself: identifying its own weaknesses and redesigning its architecture without human intervention.

What the company is building

Socher describes the approach as distinct from simple automated research. The intended process is end-to-end: ideating research directions, implementing new approaches, and validating results — all without a human in the loop. The team includes Peter Norvig, former Director of Research at Google, and Tim Shi, previously of Cresta. The company plans to release products within quarters, not years, despite the ambitious research scope.

Socher acknowledged in the interview that compute allocation will become a meaningful strategic question as AI systems gain more ability to direct their own development: decisions about which problems receive computational resources — disease, energy, infrastructure — take on a different character than a conventional product roadmap.

Why it matters for product managers

This development sits at the far end of the current AI trajectory, but has near-term implications. The framing Socher uses — AI that can take a research goal and execute it end-to-end — is a sharper version of what autonomous coding agents and multi-agent systems already do in product development workflows today. The distinction between AI that assists a task and AI that runs a task is one that product teams building AI-powered features are already navigating in practice.

The $650M raise also signals continued investor confidence in frontier AI research as a product category, which affects the competitive environment for any product built around AI capabilities. For PMs tracking the direction of AI tooling, this launch is a signal about where the next wave of capability is expected to come from.