Infinity Aims to Disrupt Nvidia’s Dominance in AI Chip Development

In a significant move, Infinity, an AI infrastructure company, has secured a $15 million funding at a $100 million valuation from investors including Touring Capital, Principal VC, and researchers from OpenAI and Anthropic. This investment is a testament to the company’s ambitious vision to revolutionize AI chip development by providing a CUDA-alternative software stack.

The largest AI development frameworks, PyTorch and TensorFlow, have been built on top of CUDA, which allows developers to write their apps in popular languages like Python, use those major AI frameworks, and have their apps run on Nvidia chips by default. However, most app-level startups lack the resources or expertise to write their own kernels, the low-level software that operates chips, and port their apps to other AI chips.

Infinity is trying to bridge this gap by building a universal inference library that can run on any type of chip, including SRAM, GPUs, phone chips, and Systolic Arrays. This library will enable developers to create AI models that can run on a wide range of chips, reducing the dependence on Nvidia’s proprietary technology.

A Universal Inference Library for a New Era in AI Development

Infinity’s AI research agent, Ignition, is designed to write the low-level code needed for AI inference on Nvidia-alternative chips. It tests, debugs, and measures how fast the hardware performs with the code, and automatically rewrites the code if needed to improve performance. The system is self-optimizing, meaning it continuously learns and improves itself, adapting to different chip architectures regardless of proprietary designs.

Infinity’s CUDA-level software stack is a significant achievement, as it allows developers to write their apps in popular languages like Python and have them run on a wide range of chips. This is a game-changer for the AI development community, as it enables developers to create AI models that can run on any chip, reducing the dependence on Nvidia’s proprietary technology.

Customers and Partnerships

Infinity has already gained the attention of major players in the AI chip space, including the AI chip maker D-Matrix. The company is also in talks with other big chip and cloud companies, demonstrating the potential of its technology to disrupt the status quo.

In an interview with TechCrunch, Jeremy Nixon, the founder of Infinity, shared his vision for automated invention, where AI systems can generate new machine learning algorithms and automatically evaluate them in a feedback loop. This vision has led to the development of Infinity’s AI research agent, Ignition, which is designed to write the low-level code needed for AI inference on Nvidia-alternative chips.

Nixon’s success with his machine learning algorithm, Omega, has inspired him to explore other areas where automated systems can generate low-level code, like the kernels and so forth, needed to help run chips more effectively. Infinity’s AI research agent, Ignition, is a testament to this vision, as it continuously learns and improves itself, adapting to different chip architectures regardless of proprietary designs.

A New Era in AI Development

Infinity’s CUDA-level software stack is a significant achievement, as it allows developers to write their apps in popular languages like Python and have them run on a wide range of chips. This is a game-changer for the AI development community, as it enables developers to create AI models that can run on any chip, reducing the dependence on Nvidia’s proprietary technology.

The company’s self-optimizing AI research agent, Ignition, is a key component of its technology, as it continuously learns and improves itself, adapting to different chip architectures regardless of proprietary designs. Infinity’s universal inference library is designed to run on any type of chip, including SRAM, GPUs, phone chips, and Systolic Arrays, making it a game-changer for the AI development community.

With its $15 million funding, Infinity is poised to disrupt the status quo in AI chip development, providing a CUDA-alternative software stack that enables developers to create AI models that can run on a wide range of chips. This is a significant step towards a new era in AI development, where developers can create AI models that can run on any chip, reducing the dependence on Nvidia’s proprietary technology.