Ricursive Intelligence Aims to Slash Chip Design Cycles with Self-Improving AI

At the heart of modern artificial intelligence lies a bottleneck: the slow, complex process of designing the hardware that powers it. While AI models advance rapidly, chip development traditionally spans two to three years. Ricursive Intelligence, a startup backed by former Google researchers, is working to collapse that timeline to mere weeks by deploying AI systems capable of designing their own hardware.

Co-founders Anna Goldie and Azalia Mirhoseini will address this challenge on the TechCrunch Disrupt Stage during the 2026 event. Their session, titled “When AI Starts Designing Its Own Hardware,” focuses on closing the feedback loop between AI capabilities and chip development, potentially unlocking faster iterations of both software and silicon.

The company’s foundation rests on prior groundbreaking work at Google. Before launching Ricursive, Goldie and Mirhoseini co-led AlphaChip, an AI system that could generate chip layouts in hours—a task that previously required significantly more human effort. Their algorithms contributed to the design of multiple generations of Google’s Tensor Processing Units (TPUs).

Ricursive Intelligence seeks to expand upon this success by creating tools that not only automate component placement and design verification but also learn across different projects. The goal is a self-improving cycle where designing one chip informs and enhances the creation of the next, accelerating the development of more efficient architectures.

Launched in late 2025, the startup attracted substantial investment quickly, raising $335 million within four months to reach a $4 billion valuation. The funding round, which included a $300 million Series A, attracted prominent investors such as Nvidia. Goldie serves as CEO, holding a Ph.D. from Stanford and recognized by MIT Technology Review as one of its 35 Innovators Under 35. Mirhoseini acts as CTO and is an assistant professor at Stanford, where she also founded the Scaling Intelligence Lab.

Both founders bring extensive experience from Google DeepMind, where they served as senior staff research scientists, and Anthropic, where they were early employees. They also co-founded Google’s ML for Systems team.

TechCrunch Disrupt takes place from October 13–15 at Moscone West in San Francisco, featuring over 200 sessions, 250 speakers, and 300 exhibiting startups. The event expects more than 10,000 attendees, including founders, investors, and technology leaders.