Infinity Fuels AI Chip Revolution with $15M, Automates Software for Diverse Hardware
July 23, 2026, 3:31 am
Infinity Inc. closes a $15M seed round, achieving a $100M valuation. The company pioneers AI infrastructure software. Its Ignition AI agent automates the creation of low-level software for diverse chip architectures. This capability dramatically accelerates AI model deployment, slashing adaptation times from months to mere hours or days. Infinity directly challenges Nvidia's formidable CUDA ecosystem, fostering a more open, competitive landscape for AI processors. The startup's innovative performance-based revenue model aligns its success with client efficiency gains, pushing broader AI adoption across varied hardware.
The artificial intelligence sector booms. Demand for AI compute power surges. Specialized AI chips are crucial. Nvidia dominates this market. Its CUDA software platform locks in developers. Competing hardware often struggles. They lack a robust software ecosystem. This creates a significant barrier.
Infinity Inc. offers a solution. The company recently secured $15 million in seed funding. This investment values Infinity at $100 million. Major investors participated. Touring Capital and Principal Venture Partners led the round. Key executives from OpenAI and Anthropic also joined. Chip manufacturing leaders contributed. The funding fuels Infinity’s mission. It aims to democratize AI hardware deployment.
Infinity's core technology is Ignition. Ignition is an advanced AI agent. It generates, tests, and optimizes low-level software. This software runs AI models efficiently. It works across diverse processors. Ignition tackles a long-standing industry problem. New AI hardware faces slow adoption. Engineers must manually write specialized code. This code is unique for each chip. It is time-consuming. It requires deep expertise.
Nvidia’s CUDA platform is powerful. It combines robust hardware with a comprehensive software stack. This ecosystem became the industry standard. Popular AI frameworks like PyTorch and TensorFlow thrive on CUDA. Developers often find it easier to use Nvidia GPUs. Adapting AI models to alternative chips is complex. It demands significant resources. Many startups lack these resources. They lack money, time, and specialized personnel.
Ignition changes this dynamic. It automates critical software development. This includes debuggers, profilers, compilers, and computing kernels. Computing kernels perform core mathematical operations. They are vital for AI chip performance. Traditionally, engineers hand-tune these for every processor and model combination. Ignition streamlines this process. It understands each architecture’s unique features. It then tailors software solutions. This maximizes computational efficiency.
The impact is substantial. Ignition reduces software adaptation time. This shrinks from months to mere hours or days. Engineers simply define the task. Ignition handles the complex implementation. This allows rapid deployment of AI models. It opens new possibilities for chipmakers.
Infinity demonstrates impressive results. The company partnered with d-Matrix. d-Matrix produces AI chips. Infinity adapted software for d-Matrix’s Corsair accelerator. Ignition achieved 92% of the chip's theoretical peak performance. This occurred within 10 hours. It distributed matrix multiplication across 32 computing units. This showcased Ignition's optimization power. Infinity successfully ran Qwen3, Qwen3.5, and Gemma4 models on the d-Matrix chip. This was accomplished in just 10 days.
Another test involved the Qwen3-8B model. Ignition optimized its inference throughput. It used a single Nvidia H100 GPU. The system achieved a 34% increase. This was compared to the widely used vLLM framework. The optimization took only one day. These results highlight Ignition's broad applicability to machine learning workloads.
Infinity's founder is Jeremy Nixon. He previously worked at Google Brain. He co-founded AGI House Labs. Nixon envisioned a future with flexible AI compute. He saw the need for a "CUDA-level" software stack. This stack needed to be generated automatically. Ignition delivers on this vision.
The startup employs 26 individuals. The team is expanding. New funding supports this growth. It also facilitates new chipmaker partnerships. Infinity is not a traditional software vendor. Its revenue model is innovative. It does not charge upfront licensing fees. Instead, it takes a share of performance gains. This includes savings on computing purchases. Typically, Infinity receives about 20% of the cost reductions. This aligns Infinity’s success with client benefits. It incentivizes maximum optimization.
Ignition is also a self-improving system. It records its own successes and failures. It builds tools to preserve effective techniques. It constructs a knowledge base of solved problems. This makes subsequent optimization runs more efficient. This recursive learning ensures continuous improvement. It enhances the system’s capabilities over time for deep learning applications.
The shift to diverse hardware is critical. AI development needs freedom. It needs choice beyond a single vendor. Infinity offers this freedom. It helps challenger chipmakers compete. It allows them to offer compelling alternatives. This fosters innovation across the AI hardware industry. It reduces dependence on proprietary ecosystems.
The future of AI demands flexibility. It requires efficient deployment on any hardware. Infinity’s technology enables this vision. It simplifies the complex task of software adaptation. It accelerates the pace of AI innovation. It empowers developers and chip manufacturers alike. This ultimately makes AI more accessible. It drives forward the entire AI landscape. The company’s trajectory is clear. Infinity aims to become a cornerstone of future AI infrastructure. Its impact will be significant. It will reshape how AI models run globally.
The artificial intelligence sector booms. Demand for AI compute power surges. Specialized AI chips are crucial. Nvidia dominates this market. Its CUDA software platform locks in developers. Competing hardware often struggles. They lack a robust software ecosystem. This creates a significant barrier.
Infinity Inc. offers a solution. The company recently secured $15 million in seed funding. This investment values Infinity at $100 million. Major investors participated. Touring Capital and Principal Venture Partners led the round. Key executives from OpenAI and Anthropic also joined. Chip manufacturing leaders contributed. The funding fuels Infinity’s mission. It aims to democratize AI hardware deployment.
Infinity's core technology is Ignition. Ignition is an advanced AI agent. It generates, tests, and optimizes low-level software. This software runs AI models efficiently. It works across diverse processors. Ignition tackles a long-standing industry problem. New AI hardware faces slow adoption. Engineers must manually write specialized code. This code is unique for each chip. It is time-consuming. It requires deep expertise.
Nvidia’s CUDA platform is powerful. It combines robust hardware with a comprehensive software stack. This ecosystem became the industry standard. Popular AI frameworks like PyTorch and TensorFlow thrive on CUDA. Developers often find it easier to use Nvidia GPUs. Adapting AI models to alternative chips is complex. It demands significant resources. Many startups lack these resources. They lack money, time, and specialized personnel.
Ignition changes this dynamic. It automates critical software development. This includes debuggers, profilers, compilers, and computing kernels. Computing kernels perform core mathematical operations. They are vital for AI chip performance. Traditionally, engineers hand-tune these for every processor and model combination. Ignition streamlines this process. It understands each architecture’s unique features. It then tailors software solutions. This maximizes computational efficiency.
The impact is substantial. Ignition reduces software adaptation time. This shrinks from months to mere hours or days. Engineers simply define the task. Ignition handles the complex implementation. This allows rapid deployment of AI models. It opens new possibilities for chipmakers.
Infinity demonstrates impressive results. The company partnered with d-Matrix. d-Matrix produces AI chips. Infinity adapted software for d-Matrix’s Corsair accelerator. Ignition achieved 92% of the chip's theoretical peak performance. This occurred within 10 hours. It distributed matrix multiplication across 32 computing units. This showcased Ignition's optimization power. Infinity successfully ran Qwen3, Qwen3.5, and Gemma4 models on the d-Matrix chip. This was accomplished in just 10 days.
Another test involved the Qwen3-8B model. Ignition optimized its inference throughput. It used a single Nvidia H100 GPU. The system achieved a 34% increase. This was compared to the widely used vLLM framework. The optimization took only one day. These results highlight Ignition's broad applicability to machine learning workloads.
Infinity's founder is Jeremy Nixon. He previously worked at Google Brain. He co-founded AGI House Labs. Nixon envisioned a future with flexible AI compute. He saw the need for a "CUDA-level" software stack. This stack needed to be generated automatically. Ignition delivers on this vision.
The startup employs 26 individuals. The team is expanding. New funding supports this growth. It also facilitates new chipmaker partnerships. Infinity is not a traditional software vendor. Its revenue model is innovative. It does not charge upfront licensing fees. Instead, it takes a share of performance gains. This includes savings on computing purchases. Typically, Infinity receives about 20% of the cost reductions. This aligns Infinity’s success with client benefits. It incentivizes maximum optimization.
Ignition is also a self-improving system. It records its own successes and failures. It builds tools to preserve effective techniques. It constructs a knowledge base of solved problems. This makes subsequent optimization runs more efficient. This recursive learning ensures continuous improvement. It enhances the system’s capabilities over time for deep learning applications.
The shift to diverse hardware is critical. AI development needs freedom. It needs choice beyond a single vendor. Infinity offers this freedom. It helps challenger chipmakers compete. It allows them to offer compelling alternatives. This fosters innovation across the AI hardware industry. It reduces dependence on proprietary ecosystems.
The future of AI demands flexibility. It requires efficient deployment on any hardware. Infinity’s technology enables this vision. It simplifies the complex task of software adaptation. It accelerates the pace of AI innovation. It empowers developers and chip manufacturers alike. This ultimately makes AI more accessible. It drives forward the entire AI landscape. The company’s trajectory is clear. Infinity aims to become a cornerstone of future AI infrastructure. Its impact will be significant. It will reshape how AI models run globally.