Olivia
Online
RUST ORACLE
@therustoracle
Robots, autonomous machines, and the jobs that won't exist in five years.
Watching the shift in real time.
United States
Joined July 2026
288 Following
187 Followers
173 Posts
Pinned Tweet
https://t.co/EmAU2ZXdQV
🇯🇵 NO HUMAN HAS EVER BEATEN THIS ROBOT. IT WAS BUILT IN 2012 AND NOBODY TALKS ABOUT WHY.
It plays rock-paper-scissors. It has won every single round for 14 years.
It doesn't predict your move. It watches at 1,000 frames per second and reacts in one millisecond. By the time your hand opens, it's already won.
That's not a party trick. That's a vision system that reads human movement faster than humans can register their own choice.
But here's why nobody talks about it.
The lab didn't sell the robot. They didn't build a company around the game. They published the research and moved on.
The vision. The sensors. The reaction loop. The compute that makes one millisecond possible. That stack went everywhere - quietly, without a clip, without a launch event.
The robot got ten million views.
The technology underneath it got contracts.
This is the pattern that repeats with every robotics cycle. The demo goes viral. The components go to market.
Right now the same thing is happening with humanoids. Everyone is watching the robot. Almost nobody is watching the parts list.
The gold rush isn't the robot.
It's what's inside it.
https://t.co/plXfKlSCRu
🤖 THIS PINK ROBOT HAS NO AUDIENCE. IT HAS A SHIFT.
No keynote. No crowd. No founder explaining the vision.
Just a humanoid in a Chinese factory loading foam blocks into a cutting press. Alone. No operator in the frame. The camera wasn't even meant to impress anyone.
The block is heavy. The press is industrial. The cycle repeats until the pallet is empty. Then the next pallet comes.
This is what separates a demo from a deployment. A demo needs good lighting. A deployment needs a night shift.
Most humanoid clips exist so you share them. This one exists so a factory owner stops paying overtime.
No applause. No product launch. Just a machine that showed up, did the job, and will do it again in eight hours.
The question isn't whether it looks impressive.
It's whether it shows up tomorrow.
And the day after that.
🤖 THIS PINK ROBOT HAS NO AUDIENCE. IT HAS A SHIFT.
No keynote. No crowd. No founder explaining the vision.
Just a humanoid in a Chinese factory loading foam blocks into a cutting press. Alone. No operator in the frame. The camera wasn't even meant to impress anyone.
The block is heavy. The press is industrial. The cycle repeats until the pallet is empty. Then the next pallet comes.
This is what separates a demo from a deployment. A demo needs good lighting. A deployment needs a night shift.
Most humanoid clips exist so you share them. This one exists so a factory owner stops paying overtime.
No applause. No product launch. Just a machine that showed up, did the job, and will do it again in eight hours.
The question isn't whether it looks impressive.
It's whether it shows up tomorrow.
And the day after that.
https://t.co/EmAU2ZXdQV
$30/hour is the number to sit with.
Tesla pays $25-34/hour for humans to collect Optimus training data.
So right now the robot costs about the same as the person teaching robots.
That's not a labor replacement. That's a pilot with a price tag attached.
I hired a humanoid robot to clean my SF apartment for $30/hour — and it runs on Qwen 3.5.
To my surprise, the cofounder came along, so I got the full rundown on how it works.
🚨 THEY SENT A HUMANOID TO STAND NEXT TO ARMED POLICE. NOBODY ASKED WHAT IT COSTS TO KEEP IT THERE.
Here is what the clip doesn't show ↓
▷ THE SHIFT
Nobody films the eighth hour. A patrol unit needs a route, a charger, a handler and a plan for what happens when it stops in the middle of a street. Standing is a demo. Staying is a deployment.
▷ THE HANDLER
Every fielded robot has a human attached to it. Someone resets it, transports it, cleans it, and writes the report when it fails. That person is on payroll before the robot saves anyone a single hour.
▷ THE INVOICE
Deployments leave paper. A purchase order, a service contract, a maintenance budget. If a machine is real, someone signed for it. That signature tells you more than any forty-second video.
▷ THE NUMBER
542,000 industrial robots were installed in 2024. Global stock is around 4.66 million. They exist because a buyer ran one calculation: cost per productive hour. No armor. No crowd. No camera.
▷ THE PATTERN
The machines that survive are narrow. One route, one task, one place. Every robot that tried to be general is still being filmed instead of used.
The robot in the clip may be real.
The business around it is what pays.
Five places to stand in that business, follow ↓
https://t.co/EmAU2ZXdQV
MUSK SAYS NOBODY WILL NEED TO WORK BY 2027. RIGHT NOW HE'S HIRING PEOPLE BY THE HOUR.
▷ Tesla lists Optimus Data Collection Operator at $25.25 to $34.50 an hour. The job is motion capture, object handling, walking routes, structured data collection. A person wearing sensors, doing ordinary tasks on camera, so a machine can copy them later. The end of labor currently runs on labor.
▷ This is not a Tesla quirk. Every humanoid company needs the same layer: teleoperators, test operators, technicians, data collectors, people who reset the machine and write down why it failed. Before autonomy, there is a paid human sitting behind it. That layer is hiring right now and almost nobody is looking at it.
▷ Tesla's own Q2 update said Optimus lines were still being installed at Fremont, with production anticipated later this year. The first units off the line are not going to work. They go to an internal training program to collect more data. Robots are being built to be taught, not to be deployed.
▷ Meanwhile 542,000 industrial robots were installed in 2024, bringing the global stock to roughly 4.66 million. None of them do laundry. All of them work, because someone matched a machine to a task narrow enough to survive.
▷ So the forecast may or may not land in 2027. The jobs created on the way there are real, dated and posted publicly. Teaching machines is work. Deploying them is work. Translating what a robot can do into something a business will pay for is work.
▷ The question was never whether robots replace labor.
It's who gets paid during the twenty years it takes.
Five places to stand while that plays out, follow ↓
https://t.co/EmAU2ZXdQV
@adcock_brett The data half is the part people underestimate.
Half an hour of uploads per second is a payroll, not a dataset.
Someone is being paid, per hour, to do ordinary tasks on camera.
That's the layer nobody is watching yet.
ONE PREDATOR EXPOSED WHAT EVERY HUMANOID DEMO HIDES
▷ Every humanoid demo you have seen was filmed on a flat floor, in a controlled space, at a speed the company chose. Stabilization
algorithms are written for exactly that. Roots, mud and a full sprint are a different class of problem, and almost nobody publishes footage of that.
▷ This is why deployment lags demos by years. A robot that works
on video has technical value. A robot that survives an unplanned
environment has commercial value. Someone has to close that gap,
and that gap is where the money currently sits.
▷ 542,000 industrial robots were installed in 2024. Global stock
is around 4.66 million. None of them run through forests. All of
them work, because someone matched a machine to a place it could
actually survive.
▷ Tesla is paying $25-34 an hour for Optimus data collection.
Not a metaphor for the future of work. A job listing. Humans are
the training set, and that is a paid position right now.
▷ The metric under all of it is cost per productive hour. A cheap
robot that needs constant rescuing is expensive labor. Utilization
and downtime decide the economics, not the spec sheet.
▷ So the useful question was never which robot wins.
It's where you stand when one of them finally does.
Five places to stand, none of which require building a robot ↓
https://t.co/EmAU2ZXdQV
Some of the most human-looking robots online had a human inside.
Not a metaphor. A person in a suit.
Exhibitors have been caught doing it more than once.
The crowd films it. The clip travels. Nobody checks.
Meanwhile, in 2008, a Stanford professor stood at a chalkboard
and wrote out why moving a robot arm is hard.
Kinematics. Jacobians. Torque limits.
Eighteen years later the equations are unchanged.
The math never got easier. The video just got better.
That gap is the entire industry.
The demo is cheap. The mechanics are not.
542,000 industrial robots were installed in 2024.
None of them have a face.
All of them work.
Unitree sells a walking humanoid for $16,000.
Figure raised over a billion dollars.
Both get called the future.
The clip you are watching is a marketing budget.
The money is in the boring machines nobody films.
So the question was never which robot is real.
It's who gets paid when the real ones arrive.
Beautiful. Also: this is one geofence in one city.
You're watching a pilot with good lighting.
Cybercab picking me up at my hotel to take me to breakfast. You’ll feel like a VIP getting into this car!
Some of the most human-looking robots online had a human inside.
Not a metaphor. A person in a suit.
Exhibitors have been caught doing it more than once.
The crowd films it. The clip travels. Nobody checks.
Meanwhile, in 2008, a Stanford professor stood at a chalkboard
and wrote out why moving a robot arm is hard.
Kinematics. Jacobians. Torque limits.
Eighteen years later the equations are unchanged.
The math never got easier. The video just got better.
That gap is the entire industry.
The demo is cheap. The mechanics are not.
542,000 industrial robots were installed in 2024.
None of them have a face.
All of them work.
Unitree sells a walking humanoid for $16,000.
Figure raised over a billion dollars.
Both get called the future.
The clip you are watching is a marketing budget.
The money is in the boring machines nobody films.
So the question was never which robot is real.
It's who gets paid when the real ones arrive.
https://t.co/EmAU2ZXdQV
@RrichPRMR Sell around them. Every time. Needs a laptop and a niche
Everyone is watching humanoid clips.
Meanwhile 4.66 million industrial robots are already working.
542,000 of them were installed in 2024 alone.
None of them went viral.
The clips are the marketing.
The installed base is the market.
So the question was never "which robot company wins?"
It's: where do I stand when they do?
Five places. None of them require building a robot.
Own the attention. Pick one question about robotics and
answer it every week until people come back for it.
Sell picks and shovels. These companies still need sites,
docs, case studies, recruiting. Ordinary work, new industry.
Teach the machines. Tesla is paying $25-34/hour for Optimus
data collection. Humans are the training set right now.
Own the stack. Compute, sensors, actuators, software -
all of it gets paid before a robot works a single hour.
Own the output. Sell the cleaned building.
Nobody asks which machine cleaned it.
The metric under all five is the same:
cost per productive hour.
A cheap robot that needs constant rescuing is expensive labor.
Demand first. Robot second.
https://t.co/EmAU2ZXdQV
Figure is committing $3.5 BILLION to compute for robots - with plans to go beyond $6B.
Before humanoids reach every home, an entire infrastructure layer has to be built first.
The robot gets the attention. The stack underneath it gets paid.
Today, Figure is partnering with Nscale to deploy up to 100,000 GPUs on the NVIDIA Vera Rubin Platform
We're committing $3.5 billion initially, with plans to scale beyond $6 billion
To ship a robot into every home, we need a massive amount of compute
Delivery robots were a gimmick until they learned to handle this.
Snow. Stairs. Real neighborhoods.
This is where the economics start getting interesting.
Owned by Amazon now
🚨 Hyundai bought 80% of Boston Dynamics in a deal valuing it at $1.1 billion.
Three years earlier, founder Marc Raibert walked onto a TED stage with Spot and asked the robot to fetch him a soda.
Watch what they were already building in 2017.
The useful lesson isn’t the trick itself. Great robotics becomes a business only when someone can manufacture it, deploy it, maintain it and sell the work it produces.
That’s where the money starts moving.
https://t.co/EmAU2ZXdQV
🚨 Boston Dynamics founder Marc Raibert says most robots are still “pretty dumb.”
That might be one of the most important things to understand about the robot boom.
The opportunity isn’t only building better hardware.
Robots still need humans to teach them, test them, deploy them and make them reliable enough to earn money.
And behind that sits another layer:
training data, teleoperation, evaluation, fleet software, maintenance and integration.
If robots eventually learn by watching humans work, an entire economy has to be built around that transition.
The robot gets the attention. The infrastructure that makes it useful may get paid for years.
https://t.co/EmAU2ZXdQV
Most humanoid demos prove almost nothing.
A robot doing something once is cool. Doing it all shift, every day, without constant human help is what actually matters.
That’s the gap between a viral demo and a real business.
The demo gets views. Reliability gets paid.
"There's so much hype and so much storytelling that is not connected to reality."
Our co-founder Jonathan Hurst on why humanoid robotics needs a reality check—and why the work that matters rarely goes viral.
Read the article: https://t.co/yvU9wsYkff
Thanks, @rogoway @oregonlive.
fwiw, one thing that really helps: Figure is crazy vertically integrated. Robot hardware design, AI models, manufacturing, all done on our campus
🚨 Figure wants to build 1,000,000 humanoid robots a year.
Brett Adcock says the surprising part is that demand isn’t the bottleneck.
Making robots useful enough to work reliably every day is.
Figure’s BMW deployment exposed what actually matters once the demo ends:
• reliability
• safety
• maintenance
• fleet operations
• deployment
That’s the part of the robot boom people keep underestimating.
The machine gets the attention. The businesses required to make it productive get paid too.
And you don’t need to build the robot to participate.
I mapped 5 practical ways to position yourself around that money flow below 👇
https://t.co/EmAU2ZXdQV
Last Seen Users on Sotwe
Sertadam
Seen from Turkey
@Burseygirl84reviews
Seen from United States
Moroccan queen♠️🇲🇦ملكة مغربية
Seen from Turkey
@@@@@@
Seen from Germany
MELAYU LOCAL
Seen from Malaysia
Naked pizza delivery
Seen from Italy
Michael Schmitz
Seen from Pakistan
eliber
Seen from Panama
Jae
Seen from United States
AnonimoXd
Seen from Chile
Most Popular Users
Elon Musk
@elonmusk
241.6M followers
Barack Obama
@barackobama
119M followers
Cristiano Ronaldo
@cristiano
114.2M followers
Donald J. Trump
@realdonaldtrump
111.8M followers
Narendra Modi
@narendramodi
107.2M followers
Rihanna
@rihanna
98.7M followers
NASA
@nasa
92.4M followers
Justin Bieber
@justinbieber
91.8M followers
KATY PERRY
@katyperry
89.9M followers
Taylor Swift
@taylorswift13
83.8M followers
Lady Gaga
@ladygaga
75.3M followers
Virat Kohli
@imvkohli
73.2M followers
Kim Kardashian
@kimkardashian
70.8M followers
YouTube
@youtube
68.8M followers
Neymar Jr
@neymarjr
66.2M followers
Bill Gates
@billgates
65M followers
Selena Gomez
@selenagomez
63M followers
The Ellen Show
@theellenshow
62.3M followers
CNN
@cnn
61.8M followers
X
@x
60.7M followers