Picture a humanoid robot about as tall as a kitchen chair seat, walking across a classroom floor, tripping, and getting right back up. Now picture it costing about as much as a high-end laptop. That’s the pitch behind the FF Nova humanoid, which Faraday Future launched on 15 October.
Yes, that Faraday Future, the electric car company. It’s now selling a $2,490 walking robot to developers and schools, plus a $299 mini version for kids. So is this the moment humanoid robots get cheap enough for everyone? Or is it a clever press release? Let’s dig in.
Wait, a humanoid robot for $2,490?
Faraday Future and its sister company, FF EAI Robotics Ecosystem (FFR), launched two robots together. Nova itself costs $2,490 and targets developers, universities and schools. The smaller FF Nova Mini costs $299, and its audience is children aged eight and up.
Both companies call Nova the “first embodied AI humanoid robot sold in the US for under $2,500.” Embodied AI simply means AI that controls a physical body, like a robot, instead of living only on a screen. The Next Web, though, points out that it hasn’t independently verified that “first” claim.
FFR’s chief executive, Jerry Wang, called the launch “an important step for FFR in bringing Physical AI from the laboratory to real-world applications.” That’s the big idea here. Instead of showing off a robot on stage, they want robots in the hands of people who’ll actually tinker with them.
What you actually get for the money
First, set your expectations on size. Nova is 49.7 cm tall and weighs 2.8 kg, so think tabletop teaching robot, not factory worker. Humanoid.guide calls it a “small research and teaching platform,” which feels about right.
It has 20 degrees of freedom, meaning 20 joints that can move: two in the head, six in the arms and twelve in the legs. It can walk, turn, wave and stand back up after being knocked over. For sensing, it uses a top-mounted 2D LiDAR, a wide-angle camera, a 3D time-of-flight depth sensor and a three-microphone array.
Inside, there’s 10 TOPS of onboard computing, 8 GB of RAM and 32 GB of storage. TOPS is a rough measure of how many AI calculations a chip can run per second. Don’t expect speed, though. Nova walks at 0.1 metres per second. If you’re in a hurry, it can ride a detachable wheeled base at 0.8 metres per second instead.
Developers get ROS 2 support, a widely used robotics software framework. A developer software package costs an extra $390.
Why the FF Nova humanoid is really about software
Here’s the part we find most interesting. The hardware is modest, but the plan around it isn’t. FF is offering a training programme built on Nvidia’s Isaac platform. Developers teach the robot new skills in simulation first, then test them on the real Nova.
People call this approach sim-to-real. In other words, you let a virtual robot fall over again and again in software, so the physical one doesn’t have to. The workflow includes parallel simulation training, reward tuning and domain randomization. That last one means shuffling things like lighting and surfaces in simulation, so the robot copes with a messy real world.
FF also plans a Skills Store, where developers can share skills they build and possibly earn money from them. If that takes off, Nova becomes a bit like an app platform for robot moves.
We’ve written about how physical AI learns to act in the real world, and this is a small, affordable version of the same idea. Because bigger humanoids cost far more, a cheap body to practise on matters for students and indie developers.
The fine print, and why we’re a little skeptical
Now for the honest bit. The first phase is limited to 100 units for invited developers and schools, and only 20 are ready to ship right away. Wider developer sales come next, then a full consumer launch, but neither has a date.
There’s also the corporate backstory. FFR agreed to buy Faraday Future’s robotics business in an all-stock deal valued at about $200 million. After closing, Faraday Future would own about 65.7% of FFR, and the companies expect the deal to close by the end of October. FFR, by the way, went by the name AIxCrypto Holdings until 30 September.
Faraday Future only began its robotics pivot earlier this year. Still, it isn’t starting from zero. The Next Web, citing Investing.com, says the robotics unit had shipped 817 units by the end of September.
So we’d treat the specs as promises until independent testers get their hands on one. Most of the numbers come from the company’s own product pages.
Okay, but what does the FF Nova humanoid mean for classrooms?
This is where it gets exciting for us. At EIF, we care a lot about tech education, and the price of hardware is often what holds schools back. Bigger humanoids cost a lot more. Our guide to what humanoid robots really cost shows how wide that range is.
A $2,490 robot that runs ROS 2 and trains in simulation is a different story. University labs and tech clubs, including those in Armenia and other emerging markets, could actually budget for one. Students would then learn on the same kind of tools the industry uses.
The Nova Mini is aimed at younger kids. According to Unite.AI, its $499 course runs 16 live small-group classes over eight weeks, with no coding experience required. It ends with students choreographing a robot dance. Honestly, that sounds like more fun than most coding classes we took.
Our honest take
The FF Nova humanoid won’t carry boxes or fold your laundry. But that was never the point. It’s a cheap, programmable body for learning physical AI, and that’s something the field badly needs.
Yet a launch isn’t a track record. We’d wait for real reviews, real shipping dates and the deal to close before ordering a classroom set. So here’s our question for you: if a walking robot cost the same as a laptop, what would your students build first?