Why Tesla Rebuilt the Optimus Hand From Scratch

[00:00:01]There is one part of the Optimus robot that decides whether the whole program works, and it isn't the legs, the battery, or the brain.

[00:00:09]It's the hands. Here's how we know.

[00:00:12]In mid-2025, Tesla stopped building Optimus for roughly 2 months.

[00:00:17]Not for software, not for the body, but to tear up and redesign the hand and forearm.

[00:00:23]At the same time, the head of the Optimus program, Milan Kovac, left the company.

[00:00:29]Think about what that tells you.

[00:00:31]A humanoid robot that can't manipulate objects with its hands is a very expensive mannequin.

[00:00:36]Everything Optimus is supposed to be worth, doing physical work people don't want to do by the millions, routes through one subsystem, the hand.

[00:00:46]So, this is the leg of the Optimus story everything else leans on.

[00:00:49]If the hand is dexterous enough to be useful and cheap enough to build at scale, Optimus becomes a product. If it isn't, it stays a demo.

[00:00:58]This one redesign decides which.

[00:01:00]And the redesign is a genuine engineering swing.

[00:01:03]Let me explain what Tesla actually changed.

[00:01:06]Look at your own hand for a second. The muscles that move your fingers aren't in your fingers. They're mostly up in your forearm.

[00:01:12]Long tendons run down through your wrist and pull each finger like a puppet string.

[00:01:17]Your fingers themselves are mostly bone and tendon, light, thin, and strong.

[00:01:23]Tesla's new hand copies that.

[00:01:25]In the older design, the little motors that move the fingers, the actuators, were packed inside the hand itself.

[00:01:30]The new architecture pulls those actuators out of the fingers and moves them up into the forearm. The forearm does the pulling. Thin tendons run down to the fingers and move them.

[00:01:40]Why go to all that trouble? Three reasons, and they're the reasons this could be the unlock.

[00:01:46]First, the fingers get light and slim.

[00:01:48]Nothing heavy inside them, so they can be thinner and more human-shaped, which matters for reaching into tight spaces and gripping small objects.

[00:01:55]Second, heat. Actuators are electric motors, and motors get hot when they work.

[00:02:00]Cram a stack of them into a hand and you got a heat problem in a tiny sealed space.

[00:02:04]And overheating motors were already a documented issue on the earlier design.

[00:02:09]Move them into the roomy forearm and it's far easier to keep them cool.

[00:02:12]Third and maybe most important for Tesla, manufacturing.

[00:02:16]A forearm is a bigger, simpler box than a fingertip. Motors you can reach, service, and bolt in by machine are motors you can build by the million.

[00:02:25]That's the difference between a hand you assemble by hand and one a factory can stamp out.

[00:02:30]But moving everything to the forearm creates a new hard problem and Tesla's own engineers named it.

[00:02:36]When the motor sits inches away from the finger pulling on a tendon, getting fine, precise control of that fingertip is genuinely difficult.

[00:02:44]The engineer leading the program described exactly that challenge, achieving delicate tendon control once all the actuation had been moved up into the forearm.

[00:02:52]Picture threading a needle using strings tied to your fingers from across the room.

[00:02:56]That is a core engineering bet.

[00:02:59]So how much more capable is the new hand?

[00:03:01]The honest answer, we can measure the floor, not yet the ceiling.

[00:03:05]The way you count a hand's dexterity is degrees of freedom, the number of independent ways it can move.

[00:03:11]Each joint that moves on its own is one degree of freedom.

[00:03:14]More degrees of freedom means more ways to bend, pinch, and grip.

[00:03:18]Robo's apps, which tracks Optimus teardowns, pegs the previous Gen 2 hand at 11 degrees of freedom.

[00:03:25]And that's the baseline the new hand's doubling gets measured against.

[00:03:29]The Gen 3 hand is reported to roughly double it.

[00:03:32]But here's the honest part.

[00:03:34]Tesla hasn't officially shown the new hand or confirmed its final numbers.

[00:03:38]The exact degrees of freedom and how many actuators drive them are the very things the public reveal is supposed to confirm.

[00:03:45]That's the missing number at the center of this whole story.

[00:03:48]And it's worth separating two Two people mix up. Degrees of freedom is how many independent ways the hand to move.

[00:03:54]Actuators is how many little motors do the moving.

[00:03:57]Those aren't the same figure. A tendon design can use several motors to drive one motion or one motor to drive several.

[00:04:04]So a big actuator count is about how the hand is powered and how durable it is, not just how dexterous it is.

[00:04:10]Now the strongest counter, and it's a fair one, doubling the degrees of freedom on paper is not the same as a human hand.

[00:04:16]A spec sheet doesn't tell you whether the grip is strong, precise, or long-lasting.

[00:04:21]Tesla's own feedback to its suppliers flagged real trade-offs in this tendon design.

[00:04:25]The transmission parts had a short lifespan and the dexterous hands had low load capacity.

[00:04:30]Meaning they couldn't yet hold much.

[00:04:32]Put simply, the earlier hand can move in a lot of ways, but it wore out fast and couldn't grip hard.

[00:04:38]The whole point of the redesign is to fix exactly that, durability and strength, while keeping the dexterity.

[00:04:44]Which brings us back to the uncomfortable fact, the redesign is also the thing that stopped the production line in the first place.

[00:04:51]So does the timeline confirm the leap or the trouble? Because the reveal has already slipped. On its first quarter earnings call this year, Tesla pushed the full Gen 3 reveal and body production ramp back to late July or August. It had been expected sooner, so the calendar has already moved once.

[00:05:08]Here's how to read that fairly.

[00:05:10]Tesla keeps two very different track records. On hardware, when Tesla says it has built a thing, the thing tends to be real. The actuators, the chips, the drive units usually show up and usually work.

[00:05:21]On timelines, Tesla is late almost as a rule. The date slips, the hardware eventually doesn't. So a slipped reveal is entirely consistent with a real hard engineering problem being solved slowly, which is exactly what a from-scratch hand redesign looks like. It is not, on its own, evidence the leap isn't coming.

[00:05:38]It's evidence it's hard. A 2-month production halt to rebuild the hand tells you the same thing.

[00:05:44]So the honest answer to how big this leap is, on paper, potentially a doubling of the hands dexterity in a design finally built to be durable and mass-producible. But, on paper is the operative phrase because Tesla hasn't shown it yet. Two things turn this from promise into proof, and both are datable. First, the reveal itself, expected late July or August. Watch whether Tesla shows the hand doing fine, delicate manipulation, not just waving.

[00:06:09]That's the tendon control problem, solved or not. Second, and this is the one most people will miss, watch whether the robots ship with the hands attached at volume. A great hand on the spec sheet means nothing until it's on thousands of bodies coming off the line.

[00:06:22]If the forearm redesign delivers both, the dexterity and the durability and the low cost to build, it's exactly the unlock the whole program has been waiting for.

[00:06:31]The hand stops being the bottleneck and becomes the reason Optimus works. We'll know a lot more in a matter of weeks.

[00:06:37]That's the Tesla brief for today. If it's been useful, please subscribe.