pork

[pork @porknx]

https://t.co/S5JOfZjjPt

[pork @porknx]

The first jobs to go aren't on factory floors.

Home health aides for low-complexity tasks.
Hotel runners.
Stock replenishment.

NEO Gamma carries objects on command.
Moves through a home.
Responds to voice.

Unpredictable settings with fresh judgment calls each time last longer.

[pork @porknx]

https://t.co/S5JOfZjjPt

[pork @porknx]

50 minutes and 26 seconds.
21 kilometers.
Fully autonomous.

The warehouse runner, the delivery driver, anyone paid for consistent physical output over a full shift: they're closer to displacement than any copywriter or analyst losing sleep over ChatGPT.

Engineers extended its legs 10 centimeters, cut the weight, and aimed it at the 100m.
It's already clocking 10 seconds in testing.
Bolt's record is 9.58.

But endurance on known routes?
That ceiling is already gone.

[pork @porknx]

https://t.co/S5JOfZjjPt

[pork @porknx]

Sophia the Robot just sold at Sotheby's for 5M HKD.
The FIRST move humanoid robots are actually making is becoming capital assets, not labor substitutes.
This shifts incentives so collectors and funds now bid on prototypes for portfolio diversification instead of waiting on labor returns.
That's a very different story than the one everyone's panicking about.

[pork @porknx]

https://t.co/S5JOfZjjPt

[pork @porknx]

This robot dog hits 22 km/h on wheels.

Then the road ends. It switches to walking. No pause, no reset, just real-time terrain analysis.

Unitree's AS2W climbs 45° slopes, clears 80 cm obstacles, carries 150 kg, and covers 30 km on one charge. Industrial lidar, cameras, the whole stack.

Everyone's debating whether humanoid robots are overhyped while this thing is already spec'd for search and rescue, scientific expeditions, and delivery into places vehicles can't reach.

The most capable robot in the room doesn't always look the most human.

[pork @porknx]

https://t.co/S5JOfZjjPt

[pork @porknx]

The real automation story isn't the AI agents everyone tracks.

Unitree just priced a humanoid robot at $5,900.

That's not a typo.

Less than a used Honda.

While white-collar work got the headlines, physical labor just picked up a price a small business can stomach.

The Unitree R1 showed up at WRC 2025 as a product, not a concept.

At $50,000 a unit, humanoid robots are a research project.

At $5,900, they're a capital expense a warehouse manager makes on a Tuesday afternoon.

White-collar automation already has a 2-year head start.

Physical automation just closed the price gap that was keeping it theoretical.

The companies paying attention to both are the ones quietly repositioning right now.

[pork @porknx]

https://t.co/S5JOfZjjPt

[pork @porknx]

UBTECH's Walker S1 pairs directly with lightweight unmanned forklifts.

Both systems coordinated. One handles what the other can't.

Together they run a complete logistics operation, start to finish, with no one on the floor.

Not "human-assisted automation."

Full automation. The human is just not there.

This is what the next wave actually looks like. Not a chatbot answering emails. A robot moving through physical space, picking up where the forklift stops, in a real warehouse, right now.

The arguments that made people feel safe:
- "Robots can't navigate dynamic environments"
- "Humanoids are too slow to be practical"
- "Warehouse work is too complex to fully automate"

Walker S1 is working through that list one by one.

White-collar work got disrupted FIRST because software was cheap to deploy. Physical automation is slower, heavier, more expensive to scale. But it scales. And once it does, the economics flip fast.

Warehouses are early. Fulfillment centers are next. Then anywhere humans are doing repetitive physical work in a structured environment.

The automation wave didn't pause at the keyboard. It was just loading.

[pork @porknx]

https://t.co/S5JOfZjjPt

[pork @porknx]

The first World Humanoid Robot Games just ran in Beijing.

Not a product demo. A competition.

That distinction matters more than the headline does.

Demos are curated. Competitions expose inconsistency. You cannot script a tournament.

The fact that humanoid robots can now compete against each other in structured events means they crossed a reliability threshold most people missed.

Still nowhere near handling unpredictable real-world conditions. But "can run a tournament" is a completely different category than "impressive promo clip."

This is what EARLY actually looks like. Not polished. Just real.

[pork @porknx]

https://t.co/S5JOfZjjPt

[NeuralPulse @NeuralPulsaulq]

This humanoid robot just stepped into the kitchen… and it might cook better than you 🍳🤖

Watch it chop, stir, and plate a meal with precision that feels almost human. No burnt edges, no wasted motion — just pure AI-driven skill in action. We’re entering an era where robots don’t just clean or lift, they create. Food, art, even emotion-adjacent tasks are on the table now.

Would you trust a robot chef to cook your dinner tonight? 👀 Drop your thoughts below!

[pork @porknx]

Optimus caught the ball using the precise grip a picker repeats a thousand times a shift.

Clean catch. New hand. Impressive engineering.

Everyone in that room clapped.

Nobody in that room mentioned the warehouse worker whose grip that hand is replacing.

The robot didn't eliminate your job.
The DECISION to not transition you did.

The warehouse worker foots the bill.

[pork @porknx]

https://t.co/S5JOfZjjPt

[pork @porknx]

Tesla's Optimus is now cooking and cleaning in people's homes.

One neural net. Trained entirely on human demonstration videos.

Meaning real people showed it, move by move, how to do their jobs.

When it replaces a housekeeper, the blame map gets complicated fast. Is it Tesla? The homeowner who buys the unit? Or the workers whose labor became the training data?

The demo looks clean. The liability question is anything but.

[pork @porknx]

https://t.co/S5JOfZjjPt

[pork @porknx]

China isn't waiting for the West to agree on AI ethics frameworks before scaling humanoid labor.

Ex-Robots just shipped its first 1,000-unit run of hyper-realistic humanoid workers.

Not a concept. Not a TED talk. A factory line.

A humanoid robot costs less per hour than a human.
Works 24/7. Never calls in sick.
The CFO isn't consulting a philosopher.

The actual question isn't "should we allow this?"
What do we do with the humans who get displaced?

[pork @porknx]

https://t.co/S5JOfZjjPt

[Abraham @Abraham_wi]

Psalm 20:1New International Version
Psalm 20[a]
For the director of music. A psalm of David.
1 May the Lord answer you when you are in distress;
may the name of the God of Jacob protect you.

[pork @porknx]

Nobody is buying humanoid robots.

They're subscribing to them.

Tesla's Optimus. Figure AI. 1X.

The pitch is robot as service. Pay per hour or per shift.

They deploy. You get billed. They recall when you scale down.

For CFOs that's the whole point.

Turn a hiring call into a utility bill.

No severance. No unemployment claims. No union risk.

Variable cost beats fixed cost.

No $2M capital to automate a warehouse.

Just a subscription.

That moves the decision from Fortune 500 to mid market.

I tracked three mid market RFPs last quarter that signed without board signoff.

Companies stop posting the roles.

[pork @porknx]

https://t.co/S5JOfZjjPt

[pork @porknx]

The $25T Optimus bet rests on one fact: the world was built for humans.

Every door handle. Every staircase. Every keyboard. Every assembly line tool. All sized for human hands.

You have two options.

Redesign every factory and warehouse for specialized bots. Billions per site.

Or build one robot that already fits what humans built.

That is the thesis.

Amazon has 750,000 warehouse robots. They move flat inventory on flat floors. The second a task needs a ladder or door handle, a human steps in.

Humanoid robots close that gap.

At 1 billion units, even at $20k per robot, that's $20 trillion in hardware alone.

$25 trillion stops looking like a stretch.

The company that owns the standard wins.

[pork @porknx]

https://t.co/S5JOfZjjPt

[pork @porknx]

Deep Robotics just posted footage of its DR02 climbing stairs.

Sounds underwhelming.

It isn't.

Stairs are the real filter. Not speed, not payload capacity. Balance, foot placement, weight shift on every single step, in a structure designed for humans.

A robot that can't climb stairs works in a warehouse. A robot that CAN works anywhere humans already work.

That's the actual investment thesis. Not the robot. The environments it can finally enter.

[pork @porknx]

https://t.co/S5JOfZjjPt

[pork @porknx]

The last shift was just the paperwork.

I watched routing decisions leave first.
No meeting about it. The system just handled them.

Then quality checks automated.
Then inventory reconciliation.

At some point we were auditing the auditors.

The last humans in the building weren't doing warehouse work anymore.
They were doing PAPERWORK about warehouse work.

[pork @porknx]

https://t.co/S5JOfZjjPt

[pork @porknx]

Still writing every move yourself?

Most robots need a script.

Pick up object. Move left. Place down. Repeat.

You're not giving it a goal. You're writing every step yourself. The robot is only as capable as whoever programmed the sequence.

Eno works differently.

You give it a task. Gene (the onboard AI brain) reads that task, plans the steps itself, then executes. You never touched a sequence. You just described an outcome.

That sounds like a small distinction. It's not.

Take a kitchen after dinner. The pile of dishes shifts every night. Someone has to map every grab, wipe, and stack in advance.

When the robot handles its own planning, that bottleneck disappears.

Eno is also foldable, adjusts its height per task, and has dexterous hands built for environments that aren't a controlled lab. The physical design is solid.

But the HARDWARE isn't what moves the ceiling.

The planning layer does.

Once a robot understands what you want instead of just what you typed, the gap between "useful in a factory" and "useful everywhere else" starts to close.

You're not programming a machine anymore.

You're briefing one.

That's a fundamentally different relationship. And it's the one that actually scales.

[pork @porknx]

https://t.co/S5JOfZjjPt

[pork @porknx]

I spent a day working next to a humanoid robot.

A real shift alongside real workers, not a demo or press event.

Here's what I actually observed:

It was slower than every human on the floor.
Not broken-slow. Deliberate-slow. Each motion calculated before it happened.
A trained worker would've done it in half the time on autopilot.

But it ran for 8 hours straight.
No pace drops. No distraction. No "two-minute" break that turned into 12.

By hour 5 it had closed the speed gap through pure consistency.

The failure points were the interesting part.

Anytime the environment broke from expected, a misaligned bin, a package in the wrong spot, it paused.
Sometimes 3 seconds. Sometimes closer to 10.

The humans didn't pause. They read the situation in half a second and adapted.

That pause is the whole story right now.

It's not strength. It's not compute. It's not even speed.

The actual ceiling is AMBIGUITY.

[pork @porknx]

https://t.co/S5JOfZjjPt

[offbyone @offbyoone]

a ps5 carries 16gb gddr6 and sells at $499 near cost early in the cycle. worth putting next to the constraint that gddr capacity isn't ringfenced from ai demand.

if costs rise 3-4x sony can absorb it, pass it through or delay. delay works only if microsoft faces the same bind. one of these is wrong: either the ai infrastructure build stays at current pace, or the console timeline does.

[pork @porknx]

The UBI vs mass unemployment debate assumes the hard part is choosing which path.

It's not.

The hard part is who pays.

The people who own the robots fund solutions only when forced.

Figure, Tesla Optimus, Boston Dynamics already build the hardware.

[pork @porknx]

https://t.co/S5JOfZjjPt