The MFV team is excited today (February 21, 2025) to share our investment in OpenInfer, the leader in AI compute for edge devices.
We are delighted to join the company’s over-subscribed $8M seed round funding led by Cota Capital and Essence VC, with B5 Capital, Brave Capital, Future Fund, Machine Ventures, Pretiosum, SilverCircle, StemAI, Tau Ventures, and YG Ventures, besides other top practitioners and angel investors in the AI space.
AI is at an inflection point as it is shifting to inference from training. We are seeing AI applications expanding in various verticals, such as healthcare, automotive and manufacturing. This growth drives demand beyond traditional data centers – to devices and systems at the edge such as smartphones, wearables, self-driving cars and robots. As a result, there is a need for an “Always on Inference” with a large context, which requires low latency, privacy and connectivity resilience. Cost is also a key factor for adoption.
A report from Bloomberg shows the strong momentum in inference hardware – 48 percent CAGR until 2032. This is over 1.5x the rate of growth in the training segment. At this pace, inference will account for over a quarter of the hardware market in 2032. The chart below only captures the hardware part – the overall AI infrastructure market, including software and services is expected to grow to $1.3 Trillion in that period. The inference market size is estimated to be about $400 Billion by 2030.
OpenInfer targets the AI inference market for edge applications that require low-latency, privacy, and cost-efficiency. Its software platform provides breakthrough AI inference performance at the edge with innovative optimizations in quantized value handling, enhanced caching for memory access and model-specific tuning. OpenInfer achieves 2-3x faster inference compared to Llama.cpp and Ollama for distilled DeepSeek models.
By optimizing AI inference for edge devices, OpenInfer is making it possible to build efficient and responsive AI applications. One key feature of the solution is the flexibility of its API. OpenInfer Engine is designed as a drop-in replacement – a user can replace Ollama or OpenAI endpoints with OpenInfer by updating a URL and without changing the application code. The platform is also agnostic to the model architecture and existing agents and frameworks continue to function seamlessly, without any modifications.
As a deep tech fund, we look for companies in a large and growing market, with proprietary and hard-to-replicate technology that drives competitive advantage. Our long-held thesis is that true AI adoption and prevalence will happen at the edge. We had previously invested in Edge AI startups including Sagence AI, an edge silicon company for low-power inference, and several Robotics / Physical AI companies such as Agility Robotics, Chef Robotics, and Ati Motors.
We at MFV Partners continue to bet on the growth of AI at the edge and are hence thrilled to partner with OpenInfer which is designing inference from the ground up. Their solution enables running powerful models efficiently on the edge, and is a key component of the inference infrastructure needed for a transformative future.
Look forward to the exciting journey with the highly talented team and witnessing up-close the company’s impact on the AI landscape. Welcome to the MFV Portfolio, OpenInfer!