About Anvil Robotics
We raised a $6.5m seed round 🎉
Hey there this is Mike Xia. Co-Founder and CEO at Anvil Robotics. I have several exciting announcements to make!
We're releasing the first end to end devkit for Physical AI, allowing any single developer to collect data, train a model, and deploy a real robot for under $8,000 USD.
To acclerate our vision, Anvil has raised $6.5m to build the open source platform for Physical AI. We're excited to be partnering with top investors across Silicon Valley and Asia.
Vijay Pradeep (one of the original developers of ROS and the PR2 robot) has also joined Anvil as Co-Founder and CTO to build a world class developer experience.
Our mission is to accelerate this wave of builders through a coherent ecosystem of hardware, software, and data solutions. Today, there are many challenges slowing down the development and deployment of Physical AI ranging from cost, fragmented software, supply chain access, lead times, and much more.
Misaligned Incentives
Robot hardware vendors today have critically mis-aligned incentives. This is systematic, across Asia and the USA.
Traditional hardware companies want their customers to buy the same SKUs. This is easier to mass produce, their costs go down, and service is simpler. But teams are targeting different tasks and different markets, so there's bound to be varying needs.
They succeed when you buy robots, not when you deploy robots. These vendors only make money when hardware is sold, so they need to maximum margins, and minimize support and warranty costs. What happens a year later is literally is not a variable in the equation.
This is an extremely short sighted value extraction mindset. Customer needs are diverse, and given the nascent stage of the market, they are likely to change quickly and need iteration.
Anvil is structured differently. We’re proposing a new model to better align with our customers for the long run, adopting what's worked in cloud computing and aligning around long term value add.
For early stage developers, our software stack is effectively free, and our hardware will be equal or better than alternatives. Worth a look, could save you weeks or months of pain.
But more importantly, as you scale up Anvil's business model adapts to you. We'll help you fork, customize, and manufacture new robots at thin margins. This is offset through long term RAAS licensing, warranty, maintenance services, and more.
This means Anvil only wins when you deploy, shipping you the first set of hardware is just the first step in a longterm partnership.
Democratize critical tools
The surface area of Physical AI tooling and hardware is huge. Everyone will hit every speed bump. Currently every team needs to re-learn, re-evaluate, and re-build all the core pieces of a robotic system.
Embodiments range from 300g to 10kg payload arms, 6 to 20+ degrees of freedom, etc.
Data collection can range from tele operations with GELLO & VR, or different types of UMI grippers, to just raw human demonstration.
Robot control can be buggy single threaded python scripts or highly performant kHz controllers.
Sensors range from camera modules, force torque sensors, tactile sensors, that all have nuances.
Anvil is building best-in-class tooling as modules in our platform. Developers no longer need millions in venture funding to re-build these internally. A grad student with minimal funding can soon access all the state of the art robotics systems designed specifically for building Physical AI.
We hope that through making these SOTA tooling generally available, smaller teams, will be able to accomplish bigger results, driving the entire industry forward. Just like how cloud platforms drastically simplified the deployment of software applications, Anvil Robotics aims to do the same for Physical AI applications.
Closing thoughts
Seed funding is only the start of a very long journey. If you were an early supporter of Anvil, investing in us or buying from us, or just sharing us with friends and colleagues, thank you. Your support got us here today.
It's been an intense 2025, lots of progress made, but we're only scratching the surface of where we want to be as a company and as an industry. We've come a long ways from 3 people in an apartment, but I know we can all do better and move faster. I hope our efforts in 2026 will bring us one step closer to what we aspire to be, as a company and as an industry.
Through adversity, to the stars.
Best,
Mike Xia & Vijay Pradeep
Last updated