On Tuesday, Recursive Superintelligence confirmed a $400 million compute agreement with Amazon Web Services.
The company, which stepped out of stealth in May following a $650 million funding round, plans to use this multi-year contract to scale systems designed to improve themselves without constant human input. This approach demands significant processing power. Recursive expects this deal to be the first of many, noting during a conversation with TechCrunch that the current announcement represents one of the smaller commitments they will sign over the coming years.
Spending on machines, not people
The $410 million total outlay covers the majority of funds raised so far. Traditionally, AI firms spend large sums on salaries and office operations. Recursive is shifting that capital directly toward compute to automate its own product development.
Socher explained the change in focus:
“For us, it’s less about headcount and more about agent count.”
The arrangement differs from standard industry practice because Amazon is not taking an equity stake. The sheer scale of the commitment allows AWS to dedicate resources to Recursive’s specific needs. Jason Bennett, VP for startups and venture capital at AWS, stated that part of the agreement involves co-developing infrastructure purpose-built for such companies.
Self-improvement in practice
Recursive self-improvement has long been viewed as a potential turning point for artificial intelligence. While some expect a sudden explosion of progress once machines can upgrade themselves, others see the capability as a gradual continuum. The specific requirements remain unclear across the sector.
Recursive aims to apply these powers to build actual products. Socher expects the earliest examples to appear before the end of the year.
“We are excited to build like really amazing products that people can use, and you will see those within a few months, not within a few quarters or years,” Socher says. “In October or so, you’ll see some actually tangible, useful things that you’ll be able to play around with.”
