The Data Wall: Robotics Goes to Whoever Distills Humans Fastest

A robot is a few dozen numbers changing over time. Every arm, hand, wheel and leg boils down to two kinds of joint, one that rotates and one that slides, chained together by rigid links. The stream of numbers those joints produce is 0.004% of what the robot’s cameras take in.

So here’s my bet. Robotics isn’t a separate science waiting for its own Einstein. It’s a sub-problem of general intelligence, and the companies building general intelligence just walked through the door. Five claims, each with numbers behind it:

1. Action is just another modality. Whoever owns world understanding owns the robot. That’s the frontier labs.
2. General intelligence will crush every neo-lab selling a “disruptive” new learning algorithm. We already ran that experiment. In 2019, 13,000 years of simulated experience bought one Rubik’s cube at 20% success. In 2016, 3,000 robot-hours bought one skill: grasping.
3. One lever is left that a giant can’t simply buy: the human distillation interface. That’s the device and pipeline that turns human skill into training data.
4. You can measure how efficient it is, and the gaps are huge. Between today’s interfaces they run from 3x to over 100x.
5. Whoever maximizes it survives. Everyone else becomes a feature inside someone else’s model.

Start with the object itself. Here’s a whole robot arm. Six numbers. Hit play.

1. A robot is a chain of joints and links

Strip away the marketing and every robot is the same object: rigid links, connected by joints that either rotate or slide. A 7-DOF arm’s entire configuration is 7 numbers. A dexterous humanoid is about 50. The pose of the hand is just those joint values pushed through one rigid transform $A_i \in SE(3)$ per joint:

Everything a robot does is a trajectory $q(t)$ through that space. Wiping, folding, pouring, plugging in a cable. We can code any trajectory we want. The hard part was never the joints. It’s choosing the trajectory. And that needs knowing what a cup is, how a towel drapes, where you want things, and what happens if you push too hard. That’s world understanding, and world understanding is general intelligence.

The bandwidth makes the point brutal. Fifty joints at 50 Hz is 2,500 numbers a second. Three cameras at 512 × 512 RGB and 30 fps is 70.8 million numbers a second. The action is 0.004% of the stream.

It’s also a smooth, low-dimensional waveform, which is exactly why Physical Intelligence’s FAST tokenizer squeezes it with a frequency transform (the DCT), the same family of trick JPEG and MP3 use. Play with it below. Six coefficients out of fifty and you can barely tell the difference. To a frontier lab that already models speech and video, the action head is a rounding error.

2. General intelligence eats robotics

If the action is a rounding error, then the robot brain is just the world model plus a thin decoder. And the world model is the thing frontier labs already train on tens of thousands of GPUs and the biggest video corpora on earth. NVIDIA’s Cosmos alone was trained on 20 million hours of video (NVIDIA). In 2026 every major lab stopped watching and moved in:

Read OpenAI’s job posts and the strategy is plain. They’re hiring for collection facilities, operators, rigs, throughput and data quality, not just model researchers. OpenAI wound down its first robotics team in 2021, citing a lack of data. The second attempt starts by building a human data factory. The labs have already reached the conclusion I’m arguing for here.

3. Exploration is a GPU furnace, and we already watched it burn

Every year some new lab raises money on the same pitch: our new algorithm will let robots learn on their own. The field has run that experiment, at scale, with the best teams in the world:

Put the Rubik’s cube next to today’s best human-distilled model. 13,000 years is 114 million hours of experience, for one task, at 20%. Generalist’s GEN-1 was pretrained on 500,000 hours of human interaction across hundreds of tasks, and it adapts to a new robot and task with about one hour of robot data (Generalist). That’s 228x less experience, for a model that generalizes instead of memorizing one puzzle.

Why exploration loses: the search space. Take a 50-joint body, a 10-second task at 50 Hz, and only 10 possible values per joint per step. The number of possible trajectories is

There are about $10^{80}$ atoms in the observable universe. Random search in that space is hopeless. A human demonstration doesn’t search it at all. It points at the thin slice that works. Human distillation is importance sampling, with a billion years of evolution as the proposal distribution.

Watch it happen in a toy. Red is random exploration trying to get a hand through a slot to the cup. Teal is a policy fit to five human demos. Then crank the sliders on the calculator and watch the exponent explode.

The AlphaGo Zero exception proves the rule. Also in 2017, DeepMind showed that pure self-play beat human data at Go. It worked because Go meets three conditions: the simulator is the real game, experience is free and infinitely parallel, and the reward (win or lose) is exact. The physical world fails all three. Simulators drift from reality, real experience costs money and breaks things, and “tidy the kitchen” has no clean reward.

So my verdict on the neo-labs is simple. A lab whose edge is a new learning algorithm is proposing to rerun 2016–2019 with newer GPUs. If the idea works, it gets published or replicated within months, and the giants run it on 100x the compute. RL still has a job, but as polish, not as the engine. It fine-tunes a policy that human data already put in the right place, which is exactly how post-training works for language models.

4. The only lever left: the human distillation interface

If models commoditize and exploration is a furnace, one variable decides who wins. How efficiently can you turn human skill into robot-usable data? Call the device and pipeline that does this the distillation interface: a teleop rig, a handheld gripper, a sensor glove, a head-mounted camera. Its efficiency is

where $r$ is capture speed relative to a bare human hand, $f$ is fidelity (how much of the signal a robot can actually use: hand pose, contact, force), $d$ is diversity (distinct homes, objects and tasks per thousand hours), and $c$ is cost per captured hour.

The speed term has actually been measured. In the UMI paper, people arranging cups produced about 231 demonstrations in 15 minutes with bare hands, 111 with the UMI handheld gripper (about \$371 including its GoPro), and 35 by teleoperation (UMI). On a dynamic tossing task, teleop produced zero successful demos.

The cost column assumes a \$25-an-hour operator and hardware amortized over 4,000 hours. Per \$1,000 of hardware, UMI yields about 171x more demonstrations per hour than teleop. Feed the 3.8x data advantage through the scaling law in section 5 (error ∝ D^−0.5) and the same check buys about 2x lower error. No algorithm paper delivers 2x with that kind of certainty.

Drag the budget around. The glove’s speed is the one number nobody has published, so it’s a slider. Set it to whatever you believe.

The industry has already voted with its capex:

• Sunday trains on about 10M episodes from 500+ homes, captured with a \$200 skill-capture glove. Its ACT-1 model uses “zero robot data” (eWeek).
• Generalist pretrains GEN-1 on 500,000+ hours with no robot data, captured with low-cost wearables. Back at GEN-0 it was already growing by about 10,000 hours a week (Generalist).
• Tesla moved Optimus training in 2025 from motion-capture suits and teleop to workers wearing five-camera helmet-and-backpack rigs (eWeek).
• OpenAI is hiring to run data-acquisition stations and operator workforces at multi-site scale.

The critics’ best objection is about the $f$ term. Video alone misses the physical interaction, the push and resistance that teleop captures (eWeek). Fair. That’s exactly why the race is in the top-right corner of that scatter: wearables that record hand pose, contact and eventually force, at human speed, in real homes. Whoever pushes η furthest owns the scarcest input in robotics.

5. What the data wall costs

Why does the interface matter this much? Because the bill is enormous. Two numbers frame it: robotics has about 0.5M hours of real manipulation data, and Waymo alone has 11–18M hours of fully driverless driving (271.3M rider-only miles through June 2026, at 15–25 mph; Waymo). Driving outputs two numbers, steering and acceleration. A dexterous body outputs 40–50, and every one of them makes and breaks contact. And driving still isn’t considered solved.

Compute. Training cost is roughly $C \approx 6ND$, where $N$ is parameters and $D$ is hours × tokens per hour. A lean policy model sees about 3.5M tokens per hour of three-camera video. A rich world model at 512 px and 30 fps sees about 83M. On 20,000 B300s, one epoch over 10M hours of rich video takes about a month. At 100M hours it takes about eleven months and \$0.8–1.4B. And if model size grows with data, compute grows with the square of data. Try it:

Reliability. A task with $n$ steps and per-step success $p$ succeeds with probability $p^n$. Sunday’s table-to-dishwasher demo had 33 distinct interactions (68 in total): 72% end to end at 99% per step, 97% at 99.9%. A full “clean the kitchen” is hundreds of interactions. Here’s what that feels like when you actually run the episodes:

The scaling law turns that into a data bill. Generalist’s GEN-0 post shows loss falling as a power law in pretraining data, and one secondhand report puts the exponent at about 0.5, i.e. $L \propto D^{-0.5}$ (AIbase). I haven’t seen Generalist state that number itself, so treat it as a working guess. Then the big assumption, stated out loud: if per-step failure rate falls the same way, each 10x cut in per-step error costs 100x more data:

This is why η is the whole game. Human-level household reliability sits about two orders of magnitude of data beyond today, and the compute bill grows faster than the data. A distillation interface that’s 4x more efficient isn’t a 4x edge. It moves the finish line by years.

6. Who survives: the owners of distillation

Sort every robotics AI company by one question: does it own a distillation interface that compounds faster than a lab can buy one?

Dead on arrival: the algorithm neo-labs. Their asset is an idea, and ideas leak in months. Best exit is an acqui-hire.

Squeezed: the model-only “robot brain” companies. They sell what the giants will give away. And a general robot API fails the revenue math on the robots that actually exist today:

Physical Intelligence clearly sees this. Its Pi API, launched in beta on October 13, is a fine-tuning service, not a chat-style endpoint: upload your robot data, fine-tune on PI’s compute, run hosted inference (PI). Smart, because it pulls customer data back toward PI. But it’s services-shaped, and PI has reportedly acknowledged network-connectivity problems in some deployments.

Survivor type A: own the interface and the home (Sunday). A \$200 glove instead of a ~\$20,000 teleop rig, worn in 500+ real homes, producing about 10M episodes. Sunday raised \$165M at a \$1.15B valuation and plans to put 50 Memo beta units into homes by Thanksgiving (SiliconANGLE). Its product sense about which chores people will pay to escape is the one thing a lab won’t build fast.

Survivor type B: own the interface at industrial scale (Generalist). From 270,000 hours (November 2025) to 500,000+ (April 2026): about 11,000 hours a week of real interaction, valued at \$3B by August. A lab can copy the GEN architecture in a quarter. It can’t copy two years of operational know-how about running human capture at that rate.

Both will be courted by the giants, because both make the one input the giants can’t print.

7. Distill humanity first, then augment it

Phase 1: human distillation. Robots learn by compressing human physical behavior, and the interface sets the speed. The data that matters most is deep. Not factory loops of the same pick, but wild, messy, in-home trial and error: the over-grip, the dropped sock, the wrong drawer, and the recovery. Exploration still happens. It’s just that humans do the exploring, at human speed, with human judgment about what’s worth trying. That’s the cheapest exploration on earth, and it’s the opposite of burning GPUs on random actions.

The gate: data becomes a byproduct of revenue. A hand-built Memo costs about \$20,000 today (Sunday). Spread over three years at eight working hours a day, that’s about \$2.30 per robot-hour of capital, against \$25 for paid human capture. Once a robot’s useful work covers its own running cost, every deployed hour is free data that someone paid you to collect. Waymo hit that moment with paid rides.

Phase 2: human augmentation. Past the gate, robots work alongside people, and their own deployments become the main data source. This is where RL finally earns its keep: on top of a distilled prior, on real outcomes, with human corrections. Humans move from demonstrators to supervisors.

Below, the top half is one robot’s hourly ledger, and the bottom half is a deliberately toy timeline. Push the success rate up until the gate opens, then move the gate year and see when fleet hours overtake human hours.

8. Endgame: open weights, closed humans

To feed their core models, the giants will push robot models toward open release. Any company that tries to hold a closed model moat gets replicated, often with better results, and the replica gets released. You can already see the pattern:

• NVIDIA released Cosmos under a license that permits commercial use, and published an open reference humanoid on a Unitree chassis (The Decoder, Forbes).
• Physical Intelligence ships openpi with released weights.
• DeepSeek showed in early 2025 that a frontier reasoning model can be replicated and released openly within months, and it now has a stake in Unitree.
• The economics favor it. For a lab selling general intelligence, the robot model is a complement, and commoditizing your complement grows demand for your core product.

Now look at what nobody opens. Cosmos released weights, not its 20M hours of video. No lab has opened a large real-world manipulation corpus. OpenAI is building closed collection facilities, and Waymo never open-sourced its driver.

So the endgame is pretty precise: weights converge to free, and human experience stays private. A closed robot model is the most exposed asset in the stack. A closed distillation interface, with the network of homes and hands behind it, is the least exposed. Same line that separates the dead from the survivors in section 6.

9. What would prove me wrong

A thesis worth publishing should name its own failure conditions. Any one of these breaks part of the argument:

1. A simulator closes the contact gap. If physics simulation becomes faithful for deformable, contact-rich manipulation, the AlphaGo Zero conditions come back and self-play wins again. Breaks claim 2.
2. An algorithm wins on a data diet. Some lab hits 99.9% per-step reliability on long household tasks with under 10% of the leaders’ human data. Breaks claims 2 and 5.
3. The exponent is much steeper than 0.5. If error falls close to linearly with data, the data wall shrinks by orders of magnitude and the interface edge matters less. (Slide α to 1 in the reliability demo and watch 5B hours become 50M.)
4. A giant opens a large real-world manipulation dataset. That breaks “closed humans” and commoditizes the survivors’ moat.

What I’m watching over the next 12 months:

• Sunday’s 50 beta homes: reliability on long-horizon chores, not single demos
• Generalist’s hours per week, and any published scaling exponent beyond GEN-0
• Capture-speed numbers for gloves and wearables, the missing x-values in the interface demo
• Pi API pricing and paying-customer counts
• The first force or tactile corpus measured in hundreds of thousands of hours
• The first fleet whose own working hours outnumber human-captured hours: the phase-two gate

10. Further out: closed hive versus open hive

Now the speculative part. Take it as a hunch, not a forecast.

Section 8 ended on a moat: weights go free, human experience stays private. Push the logic one step further and that moat dissolves too. A distillation interface is a machine for making tacit labor legible. Every glove, wearable and camera rig turns know-how that was never written down into data. Point the same machinery at materials, processes and supply chains. Once AI can reverse-engineer matter and not just code, secrecy stops being a moat and becomes a liability. The rational response isn’t higher walls. It’s radical openness.

From the outside, that will look a lot like an open-source religion whose god isn’t a person but a protocol. It’s less of a metaphor than it sounds. AI is already a kind of swarm mind: one main brain emerging out of distributed models, data, sensors and agents. Once physical production becomes legible, the only thing that can stand up to a proprietary hive mind is an open hive of equal or greater scale.

Humans don’t become obsolete in that world. They become the hive’s hands: improvising, negotiating, repairing, handling the environments too messy for machines alone. Phase 2 in section 7 is the first draft of this, with humans as supervisors and correctors in the loop. Scale it up and flexible human execution becomes the frontier’s immune system, the part that handles the long tail no dataset has covered yet.

So the real fight isn’t AI versus humanity. It’s closed hive versus open hive: corporate feudalism against a distributed, post-secret civilization.

The open side wins only if it pulls off something very hard: turning humans into agents of a collective mind without extinguishing the human capacity to refuse. If it can, it’s the seed of something much bigger than robotics. If it can’t, it’s digital feudalism with better liturgy.

That’s the stake behind every glove, every dataset and every open-weights release of the next five years. A robot is joints and links, and its brain belongs to general intelligence. What stays scarce is the skill in human hands. Whoever pours that skill into data fastest survives. What they pour it into decides which hive we end up living in.

Appendix: assumptions behind the math

Every demo on this page computes from the rows above. Nothing is a product photo; the arm, the workspace and the timeline are drawings.

Sources