Startup Ricursive to Create an End-to-End AI Model for Chip Design

2026-06-10 · Source: Big Data & AI News - EE Times · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Emerging Technologies & Innovation, Hardware Design & Optimization · Depth: Advanced, medium

Summary

Startup Ricursive, founded by former Google AlphaChip leads Anna Goldie and Azalia Mirhoseini, aims to develop an end-to-end AI model for chip design, having raised \$335 million largely for GPU compute hours. The company focuses on hardware workload co-optimization, targeting workload-specific chip design for third parties. Ricursive plans a three-phase rollout: initially accelerating physical design and verification for chip companies, then combining design stages into an end-to-end model generating GDSII files, and ultimately co-designing frontier AI models with custom hardware. This approach seeks to democratize custom chip access for companies with at-scale workloads, enabling significant performance improvements and new applications beyond what off-the-shelf chips offer. The founders emphasize their independence from Google IP and traditional EDA toolchains.

Key takeaway

For AI Architects evaluating custom hardware solutions, Ricursive's approach signals a shift towards highly optimized, workload-specific silicon. You should consider how an end-to-end AI design platform could dramatically reduce the time and cost of developing custom chips for your specific models, potentially enabling performance gains and new applications previously deemed unfeasible. This could accelerate your team's ability to deploy frontier AI models by co-evolving hardware and software.

Key insights

Ricursive is building an end-to-end AI for chip design to co-optimize hardware and workloads, accelerating custom silicon development.

Principles

• Custom compute architectures yield massive performance gains.
• Rapid implementation is crucial for architecture relevance.
• Co-evolving chips and models accelerates AI development.

Method

Ricursive's three-phase rollout involves accelerating physical design, then end-to-end GDSII generation from workloads, finally co-designing frontier AI with custom hardware.

In practice

• Accelerate physical design and verification for faster market entry.
• Create custom AI accelerators for specific large-scale workloads.
• Enable new applications with stringent latency/power needs.

Topics

• AI Chip Design
• Hardware-Workload Co-optimization
• Reinforcement Learning
• Custom Silicon
• Frontier AI
• GDSII Generation

Best for: Investor, CTO, VP of Engineering/Data, AI Scientist, AI Architect, Director of AI/ML

Related on AIssential

• Meta Is Developing 4 New Chips to Power Its AI and Recommendation Systems — Meta is developing custom MTIA chips to power its internal generative AI and content ranking systems.
• Extreme Co-Design for Efficient Tokenomics and AI at Scale — Extreme co-design across the entire AI stack is crucial for efficient, scalable real-time reasoning and MoE models.
• Cognichip wants AI to design the chips that power AI, and just raised $60M to try — AI can significantly accelerate and de-risk complex chip design by automating and optimizing engineering workflows.
• The Sequence Radar #861: Last Week in AI: IPOs, Interactive Models, and Recursive Dreams — AI's evolution emphasizes agentic capabilities, interactive models, and automated self-improvement, alongside critical hardware and market dynamics.
• Exclusive: Allen Wu on Disrupting $100M Cost of Building Custom AI Chips — Heterogeneous custom AI chip development can be made affordable and accessible using an open, modular RISC-V baseline.
• In another wild turn for AI chips, Meta signs deal for millions of Amazon AI CPUs — AWS secured a significant Meta deal for Graviton CPUs, intensifying cloud and custom AI chip competition.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Big Data & AI News - EE Times.