Infinity Raises $15 Million in Seed Funding to Build the Software Layer That Makes Any AI Chip Inference-Ready

Infinity, developing a generative optimization engine that automatically improves computational kernels to accelerate AI training and inference, led by Jeremy Nixon, has raised $15 million in seed funding at a $100 million post-money valuation. The round included significant participation from Touring Capital, along with Principal VC, executives at major chip companies, researchers from OpenAI and Anthropic, and other prominent angel investors.

The funding will support the expansion of Infinity’s automated AI research platform and its autonomous AI agent, Ignition, which generates inference code for emerging AI chips. The company will also use the capital to grow its engineering team and accelerate collaborations with semiconductor partners, including d-Matrix.

Infinity is already generating millions of dollars in annual recurring revenue (ARR) through its chip design partnerships, addressing a critical challenge for both established and emerging silicon companies: developing the software required to run AI inference workloads on their hardware.

Jeremy Nixon, Founder and CEO of Infinity, said:

“The AI industry has operated under the assumption that only a small number of chips could run AI workloads effectively, largely because NVIDIA spent decades building the software ecosystem required to optimize performance. Ignition is designed to remove that constraint.

The next era of AI will be shaped not only by advances in chip design, but by the ability to make diverse hardware platforms run state-of-the-art models at high performance. For semiconductor companies, the difference between having an optimized software stack and lacking one can determine whether hardware reaches its full potential.

Infinity’s goal is to help every hardware partner unlock that potential.”

NVIDIA’s CUDA ecosystem has played a central role in shaping the AI accelerator market, with the company holding an estimated ~80% share of data center AI accelerators. While hardware platforms from companies such as AMD, Qualcomm, and AWS can compete with or exceed industry performance benchmarks in certain workloads, the development of production-ready inference software stacks has historically been a significant challenge.

Infinity addresses this gap by helping semiconductor providers accelerate the deployment of optimized inference software for their hardware platforms. Solving this software bottleneck has become a major priority across the chip industry, as inference is expected to account for a growing share of AI compute spending, reaching approximately two-thirds of total AI compute expenditure in 2026.

Ignition: Any Chip. Any Model. In Days, Not Years.

Infinity’s flagship product, Ignition, is an AI research agent that automates the generation, testing, and optimization of compute kernels, the low-level software that determines how efficiently AI models run on different chips.

The platform combines autonomous code generation with human-guided architectural input, continuously improving through performance feedback loops. By automating chip optimization, Ignition enables semiconductor companies to accelerate deployment of new hardware platforms while reducing reliance on manual software tuning.

Key capabilities include:

• Autonomous Kernel Generation: Creates, tests, and optimizes compute kernels with human oversight.

• Recursive Self-Improvement: Uses real-world performance data to continuously improve generated code.

• Cross-Architecture Optimization: Adapts across different chip architectures and instruction sets.

• Proven Performance: Infinity reports a 34% improvement in Qwen3-8B inference throughput in one day and achieved up to 92% of theoretical peak performance on d-Matrix’s Corsair chip within 10 hours of hardware access.

• Aligned Partnerships: Shares in performance gains and cost savings generated for chip partners rather than charging traditional licensing fees.

Investor Perspectives.

Sid Sheth, Founder and CEO of d-Matrix, said:

“Infinity’s AI-driven approach to model enablement has the potential to significantly reduce the time required to bring new AI compute architectures into production. The company’s work can help accelerate deployment and time-to-first-revenue for next-generation AI hardware.”

Songyee Yoon, Founder and Managing Partner at Principal Venture Partners, said:

“Infinity is addressing one of the defining challenges of the AI era: making powerful AI more accessible and affordable at scale. The company is building critical infrastructure that can help extend the benefits of AI to a broader set of users and industries.”

Samir Kumar, General Partner at Touring Capital, said:

“As AI models evolve rapidly, hardware companies face the challenge of continuously adapting their software stacks. Infinity addresses this bottleneck by using AI agents to automate kernel development and build optimized inference libraries for different hardware platforms, helping new silicon achieve peak performance faster.”

Infinity was founded by Jeremy Nixon, a former Google Brain researcher and co-founder of AGI House, a San Francisco-based artificial general intelligence community and hacker network that has supported hundreds of startups and research projects.

Nixon is also a recognized voice on the future of AI, with his views and commentary featured in publications including The New York Times and Forbes.