One arm is a research-lab masterwork that reads a wearer’s intent and, in some setups, lets them feel through it. The other was built at home by a high schooler on his sister’s 3D printer, for roughly a thousandth of the price. Both are controlled, in a sense, by thought. What separates them tells you a lot about where prosthetics really are right now, and where the money goes.

We are not clinicians, prosthetists, or biomedical engineers, and nothing here is medical guidance. This is our read of the research on two very different machines. The clinical studies we cite are small, and a striking demo is not the same as a proven solution for any individual patient.

The $300 build, and how it reads a brain without surgery

In 2022, Benjamin Choi, then a 17-year-old senior at the Potomac School in Virginia, was named a top-40 finalist in the Regeneron Science Talent Search for a mind-controlled prosthetic arm he built during the pandemic. The first version came off his sister’s $75 3D printer, strung with fishing line. The design went through more than seventy-five iterations, and the finished version costs about $300 to make.

The clever part is how it listens. Instead of implanted electrodes, Choi’s arm uses two ordinary EEG sensors that sit on the skin: one on the forehead, one clipped to the earlobe as a baseline. They pick up the faint electrical signals your brain gives off. That signal travels to a chip in the arm, and a small AI model turns the pattern into a movement. No incision, no operating room. Choi says his algorithm hits a mean accuracy of 95 percent. That figure is his own claim and as far as we can tell hasn’t been independently checked, so we’d hold it loosely.

What drove him was less the tech than the shut door. Watching a segment on brain-implant prostheses, he came away impressed and uneasy at once. Choi has said : “I was really, really amazed at the time because this technology was so impressive. But I was also alarmed that they require this really risky open brain surgery. And they’re so inaccessible, costing in the hundreds of thousands of dollars.”

The $500,000 arm and what it can do that the cheap one can’t

That expensive machine has a name: the Modular Prosthetic Limb, built by the Johns Hopkins Applied Physics Laboratory under a DARPA program that contracted more than $107 million. It is a full arm with 26 joints and hundreds of sensors, and it can curl up to 45 pounds. A clinical evaluation describes 26 controllable types of movement driven by 17 motors. As of 2015, one was reported at about $500,000.

The difference isn’t just polish. The MPL is built for fine, multi-joint dexterity, and in some setups it sends sensation back so a wearer can feel pressure through the fingertips. Getting there means surgery.

None of that comes cheap, and the weight and durability are real limits.

Off the skin or inside the skull

Underneath the price tag is a genuine engineering trade-off. Reading brain signals from inside the skull, with implanted electrodes, gets you a cleaner, richer signal, because the sensors sit close to the neurons doing the work. A case study on EEG-based control makes the point directly: implanted systems have better signal quality because the sensors are nearer the source, while sensors on the skin are portable and cheaper.

The cost is surgery, and everything that comes with it. Reading signals from the scalp is safer and far cheaper, but also noisier and less precise, which means rougher control. As the same study notes, the signal quality of scalp EEG may be inferior to what implants deliver. Put the sensor closer and you hear more clearly, but you have to open something to do it.

So Choi’s arm and the MPL aren’t really competing on the same terms. One favors access and safety and accepts a rougher signal. The other favors capability and accepts high cost and, often, an operating room.

Why the price gap matters, and what it doesn’t fix

Even a conventional, muscle-and-cable arm runs about $7,000, and thought-controlled options have long sat in a price range most people can’t touch. A $300 open build changes the conversation about who gets to try one at all.

We’d resist the tidy version of this, though, where the clever teenager slays the half-million-dollar giant. The $500,000 figure is a 2015 price for one specific advanced limb, not the going rate for every mind-controlled prosthesis, and it buys real capability the cheap arm doesn’t have: more joints, finer control, and in some setups the ability to feel. A $300 arm is a genuine opening at the low end. It is not a stand-in for a device engineered to lift 45 pounds and send sensation back to the wearer.

What the cheap build actually changes is the floor, not the ceiling. It shows that useful, non-surgical, thought-driven control can be built from cheap parts and good code, which matters most for the people priced out of everything above it.