Infinity Attire Secures $15 Million with OpenAI and Anthropic

Infinity raises 15 million dollars with OpenAI and Anthropic

The AI infrastructure company Infinity announced a fundraising round of 15 million dollars at a valuation of 100 million dollars on Monday, with investments from firms such as Touring Capital, Principal VC, as well as corporate researchers like OpenAI and Anthropic. The startup is developing software aimed at facilitating the execution of AI models on AI chips. One of the major reasons Nvidia has become the market leader is not only the performance of its chips but also its CUDA (Compute Unified Device Architecture) software, which allows its GPUs (initially designed for graphics) to function as general-purpose CPUs. The leading AI development frameworks, PyTorch and TensorFlow, have been built on CUDA. This enables developers to write their applications in popular languages like Python, use these major frameworks, and their applications will run by default on Nvidia chips.

Most of these application-level startups would not have the resources or expertise to write their own kernels — the low-level software that makes chips operate — and port their applications to other AI chips. Thus, Infinity is attempting to create an alternative kernel software to CUDA that works with all types of chips, such as SRAM, GPUs, phone chips, and Systolic Arrays. Infinity is part of a new wave of startups striving, product by product, to reduce Nvidia's dominance in the market.

Infinity aims to build a universal inference library to operate on all chips, allowing these chips to automate the replication of cutting-edge research results.

Infinity was founded last year by Jeremy Nixon, a former researcher at Google Brain and creator of the hacker network community AGI House. Nixon told TechCrunch that he decided to launch this company because he was obsessed with the idea of "automated invention" — the belief that "AI systems can actually be a meta technology." He himself invented a machine learning algorithm called Omega, which essentially created new machine learning algorithms and automatically evaluated them in a feedback loop.

This success led him to think about other cases where this approach could work, and he turned to hardware, believing that automated systems could also generate the low-level code, such as kernels and others, needed to operate chips more efficiently.

Infinity's AI research agent, Ignition, is designed to write the low-level code necessary for AI inference on chips alternative to Nvidia. It tests, debugs, and measures the speed performance of the hardware with the code, and automatically rewrites the code if necessary to improve performance. The system is self-optimizing, meaning it learns and improves continuously. It also adapts to different chip architectures, regardless of proprietary designs, according to Nixon. The result is what Infinity claims to be a CUDA-level software stack.

Among its clients is the AI chip manufacturer (and future competitor to Nvidia) D-Matrix, and Infinity is in talks with other major chip and cloud companies, Nixon stated.

However, humans remain involved, providing high-level direction while the agent handles most of the tedious tasks. In a case study, the startup found that the agent worked much faster than a human alone, reducing what could have been a process of several years or months to just a few hours or days. Infinity does not charge initial licensing fees; instead, it takes a share of performance gains and cost savings, measuring changes in tokens per second.

Currently, Infinity has 26 employees, including those in design, operations, and engineering.