AI Software Powerhouse Infinity Secures $15M to Unlock Chip Performance
July 27, 2026, 3:38 pm
Location: United States, California, San Francisco
Employees: 201-500
Founded date: 2015
Total raised: $155.07B
Infinity.inc secures $15 million in seed funding at a $100 million valuation. Its AI platform, Ignition, radically changes AI chip inference. Ignition uses autonomous AI to build optimized low-level software, making new hardware production-ready in days. This approach eliminates the critical software bottleneck, significantly boosting chip performance. It transforms theoretical compute power into real-world efficiency. Key investors support this vision. The company's technology accelerates market entry for specialized AI processors. It promises massive speed improvements, like 14x inference throughput gains. This initiative opens the AI accelerator market to more players. It capitalizes on the growing inference spending trend, democratizing high-performance AI.
Infinity.inc has made a significant move. The company secured $15 million in seed funding. This round values Infinity at $100 million post-money. Major investors participated. Touring Capital led the round. Principal Venture Partners joined. Chip industry executives provided support. Researchers from leading AI labs also invested. Angel investors contributed.
This funding fuels a critical mission. Infinity aims to revolutionize AI chip deployment. Its focus: software for AI inference workloads. New AI chips often struggle. They offer great theoretical power. Real-world performance lags. Optimized software is missing. This gap can take months or years to close. Infinity shrinks this timeline. It promises readiness in days.
The core product is Ignition. It is an autonomous AI research agent. Ignition builds low-level software kernels. These kernels are vital. They make AI models run efficiently on specialized chips. They control computational operations. They optimize processor architecture use. Ignition removes a major bottleneck. It ensures optimal silicon performance.
NVIDIA dominates AI computing. Its CUDA ecosystem is key. CUDA took decades to build. It enables efficient processor use. Rival chipmakers need similar software stacks. Without them, hardware adoption stalls. Infinity levels this playing field. It offers a path to competitive AI accelerators.
Ignition automates development. Human engineers provide direction. Ignition writes code. It measures performance. It then improves subsequent versions. This process is recursive. Software continuously tunes itself. It adapts to target processors. Hardware engineers focus on design. They avoid manual software optimization. This accelerates deep learning hardware deployment.
The results are dramatic. Ignition enhanced Qwen3-8B inference throughput. It increased from 1,400 to over 20,000 tokens per second. This happened in one day. It represented a 14-fold boost. The framework surpassed vLLM software by 34%. These are major gains for AI computing.
Infinity also partners with d-Matrix. It optimized software for the Corsair chip. AI agents reached 92% of peak theoretical performance. This was achieved in only 10 hours. Within 10 days, Qwen3, Qwen3.5, and Gemma4 models ran end-to-end. Every model layer was custom-written. This showcases Ignition's adaptability and machine learning deployment capabilities.
The platform supports diverse instruction sets. It handles various memory architectures. This helps emerging chip companies. They can support major AI models. Even if their hardware was not in training data. This democratizes AI accelerator market access. It fosters innovation among semiconductor developers.
Infinity’s business model is innovative. It ties revenue to performance gains. It focuses on cost savings for partners. It shares in created economic value. It moves beyond upfront licensing fees. This aligns incentives. It promotes long-term partnerships in AI software optimization.
The market opportunity is immense. AI computing spending shifts. More funds go to inference. Model operation in production is key. Inference will represent two-thirds of AI compute spending by 2026. This is a massive segment. It underscores the importance of efficient AI inference solutions.
Jeremy Nixon founded Infinity in August 2025. He is a former Google Brain researcher. He co-founded AGI House. The team is growing. Research engineers are being hired. Hardware enablement specialists are sought. Developers for inference libraries are needed. These roles are critical for scaling advanced AI software.
Infinity believes in automated development. It can unlock competition. More semiconductor companies can enter the AI accelerator market. The goal is clear. Make specialized processors useful faster. Remove the software bottleneck. Bridge chip design to customer deployment. This changes the AI hardware landscape. It promises a new era of diverse and powerful AI silicon. Efficient AI compute becomes accessible. This fosters innovation across the industry. The future of AI inference is accelerating.
Infinity.inc has made a significant move. The company secured $15 million in seed funding. This round values Infinity at $100 million post-money. Major investors participated. Touring Capital led the round. Principal Venture Partners joined. Chip industry executives provided support. Researchers from leading AI labs also invested. Angel investors contributed.
This funding fuels a critical mission. Infinity aims to revolutionize AI chip deployment. Its focus: software for AI inference workloads. New AI chips often struggle. They offer great theoretical power. Real-world performance lags. Optimized software is missing. This gap can take months or years to close. Infinity shrinks this timeline. It promises readiness in days.
The core product is Ignition. It is an autonomous AI research agent. Ignition builds low-level software kernels. These kernels are vital. They make AI models run efficiently on specialized chips. They control computational operations. They optimize processor architecture use. Ignition removes a major bottleneck. It ensures optimal silicon performance.
NVIDIA dominates AI computing. Its CUDA ecosystem is key. CUDA took decades to build. It enables efficient processor use. Rival chipmakers need similar software stacks. Without them, hardware adoption stalls. Infinity levels this playing field. It offers a path to competitive AI accelerators.
Ignition automates development. Human engineers provide direction. Ignition writes code. It measures performance. It then improves subsequent versions. This process is recursive. Software continuously tunes itself. It adapts to target processors. Hardware engineers focus on design. They avoid manual software optimization. This accelerates deep learning hardware deployment.
The results are dramatic. Ignition enhanced Qwen3-8B inference throughput. It increased from 1,400 to over 20,000 tokens per second. This happened in one day. It represented a 14-fold boost. The framework surpassed vLLM software by 34%. These are major gains for AI computing.
Infinity also partners with d-Matrix. It optimized software for the Corsair chip. AI agents reached 92% of peak theoretical performance. This was achieved in only 10 hours. Within 10 days, Qwen3, Qwen3.5, and Gemma4 models ran end-to-end. Every model layer was custom-written. This showcases Ignition's adaptability and machine learning deployment capabilities.
The platform supports diverse instruction sets. It handles various memory architectures. This helps emerging chip companies. They can support major AI models. Even if their hardware was not in training data. This democratizes AI accelerator market access. It fosters innovation among semiconductor developers.
Infinity’s business model is innovative. It ties revenue to performance gains. It focuses on cost savings for partners. It shares in created economic value. It moves beyond upfront licensing fees. This aligns incentives. It promotes long-term partnerships in AI software optimization.
The market opportunity is immense. AI computing spending shifts. More funds go to inference. Model operation in production is key. Inference will represent two-thirds of AI compute spending by 2026. This is a massive segment. It underscores the importance of efficient AI inference solutions.
Jeremy Nixon founded Infinity in August 2025. He is a former Google Brain researcher. He co-founded AGI House. The team is growing. Research engineers are being hired. Hardware enablement specialists are sought. Developers for inference libraries are needed. These roles are critical for scaling advanced AI software.
Infinity believes in automated development. It can unlock competition. More semiconductor companies can enter the AI accelerator market. The goal is clear. Make specialized processors useful faster. Remove the software bottleneck. Bridge chip design to customer deployment. This changes the AI hardware landscape. It promises a new era of diverse and powerful AI silicon. Efficient AI compute becomes accessible. This fosters innovation across the industry. The future of AI inference is accelerating.