RicursiveTranslation site

Company Overview

Founded in 2023 by former Google researchers Anna Goldie (CEO) and Azalia Mirhoseini (CTO), Ricursive Intelligence is an AI-powered company focused onchip designautomation startup. The company was founded only two months ago and completed $300 million in Series A financing, with a valuation of $4 billion, making it one of the most highly valued startups in the semiconductor design automation field. This round of financing was led by Lightspeed, followed by NVentures, the venture capital arm of NVIDIA, DST Global and Radical AI, and was preceded by a seed round led by Sequoia Capital, with a cumulative financing of $335 million.

Products & Services

Ricursive's core products areAI-driven automated chip design platform, provides full-flow solutions from architecture design to silicon substrate layer creation. Its services cover AI chips, data centers, autonomous driving, AR/VR and other fields, helping companies to rapidly develop high-performance, low-power customized chips. For example:

• End-to-end automated design: Companies can build specialized chips for specific workloads (e.g., autonomous driving sensing, robot control) without the need for a specialized chip design team.
• Automatic creation of silicon substrate layers: Generate the underlying chip architecture through AI to accelerate the iteration cycle and lower the design threshold.
• Customized Chip Solutions: Optimize chip performance for different scenarios to meet differentiated needs.

core technology

Ricursive's technological innovations are focused on three main directions:

1. Recursive Intelligence (Recursive Intelligence)::
   • Apply AI principles directly to the chip design process, autonomously improving architecture, layout and efficiency through a continuous feedback loop.
   • Unlike traditional EDA tools that rely on predefined algorithms, Ricursive's system dynamically adapts to the design requirements and closes the loop of design-optimization-redesign.
2. AlphaChip Reinforcement Learning Platform::
   • The AlphaChip methodology previously developed by the founders has been applied to the layout design of Google's quadruple TPU chip, proving its technical feasibility.
   • The platform is capable of autonomously generating silicon-based substrate architectures and continuously optimizing performance through reinforcement learning, targeting advanced process nodes such as 2nm.
3. ReCode Technology Framework::
   • Allows intelligences to switch freely between different decision granularities, unifying the planning and execution process and significantly improving design efficiency and adaptability.

major client

Ricursive's potential customers include two categories:

1. Tech Giants and Chipmakers::
   • Such as Google, NVIDIA, Intel, AMD, etc., need to efficiently develop AI gas pedals or customized chips to support their AI models and cloud services.
   • NVIDIA's strategic investment in Ricursive through NVentures shows its focus on AI-designed chip technology.
2. Verticals::
   • Autopilot companies (e.g., Tesla, Waymo), AR/VR vendors (e.g., Meta, Apple), robotics companies, etc., need low-cost, rapidly iterative specialized chips.

development prospect

1. Potential for industry change::
   • Reduced design cycle timeWhile traditional chip design takes 2-3 years, Ricursive's goal is to shorten the cycle to weeks or even days, triggering a “proliferation of customized silicon chips”.
   • Lowering the design threshold: SMEs can develop chips without having to set up specialized teams, reducing their dependence on a few manufacturers such as TSMC and Samsung.
2. Technology Vision::
   • The founders believe that through the closed-loop process of AI-designed chips, it may eventually be possible to achieve general artificial intelligence (AGI), in which chips can evolve themselves to support more powerful AI systems.
3. Capital and Market Recognition::
   • The participation of top-tier venture capital organizations (Sequoia, Lightspeed) and strategic investors (NVIDIA) reflects the capital market's high confidence in the prospect of AI-driven chip design.
   • Ricursive's technology fits the trend of the global chip shortage and geopolitical tensions that have made autonomous chip design capabilities a necessity.
4. Competition and challenges::
   • Need to face competition from traditional EDA giants (e.g., Synopsys, Cadence), who are also laying out AI-assisted design tools.
   • The technology needs to be validated to maintain stability and performance advantages in complex chip designs.