Recursive Superintelligence, an AI startup that emerged from stealth in May with $650 million in funding, has signed a $410 million multi-year compute deal with Amazon Web Services. The company is focused on building open-ended, self-improving systems—a research approach that demands massive computational resources. The deal represents the bulk of Recursive’s fundraising to date, but CEO Richard Socher told TechCrunch he expects it to be the smallest of many such future agreements.

Unlike traditional AI labs that allocate budgets to headcount and operations, Recursive funnels most of its capital into compute. The company aims to automate its own product development process, meaning far fewer human engineers and far more software agents. “For us, it’s less about headcount and more about agent count,” Socher said. The deal does not include an investment from Amazon—contrary to hybrid deals common among major labs—but the scale allows AWS to commit significant resources to co-develop infrastructure tailored to Recursive’s unique needs. Jason Bennett, AWS’s VP for startups and venture capital, noted that the partnership could help attract other foundation-level AI companies.

Recursive’s self-improvement approach, known as recursive self-improvement (RSI), has long been seen as a potential inflection point for AI. Some researchers predict a sudden explosion of progress once systems can improve without human intervention; others see it as a gradual continuum. For Recursive, the goal is practical: use RSI to build real products. Socher expects to release the first examples by the end of the year. “You will see them within a few months, not within a few quarters or years,” he said. “In October or so, you’ll see some actually tangible, useful things that you’ll be able to play around with.”

Source: TechCrunch