@LiveTechProof True, folding into a case isn't intelligence. It's the logistics problem nobody's solved yet, and that's the boring part that actually gets robots shipped at scale.
ROBOT JUST FOLDED ITSELF INTO A SHIPPING CRATE AND CLOSED THE LID.
No operator did it for it. Arms tucked, legs bent, dropped straight into the packing foam like it was built to ship itself.
That's the part that gets me. We keep picturing robots as things factories make. This one packs itself like a product waiting on a shelf.
Five years from now that might just be how a robot arrives at your door. No assembly crew, no delivery guy carrying a box. It opens itself.
Whoever builds the standard packaging format for self-shipping machines owns a very boring, very necessary slice of this industry.
Feels closer than a rumor already.
@LiveTechProof so brands are paying to fake real customer reviews, isn't that just deceptive advertising" -"it's disclosed as an ad either way, honesty question isn't new, we just cut the actor"
THE COMPANY BUILDING AMERICA'S BIGGEST MAGNET FACTORY DIDN'T EXIST THREE YEARS AGO – AND THE PRESIDENT'S SON IS ALREADY AN INVESTOR.
The company is Vulcan Elements. Benson, North Carolina. $1 billion, 1,000 jobs – that's a million dollars of investment for every single job, before the plant ships one magnet.
money alone wouldn't make this one stand out. what does is who's behind part of it. Donald Trump Jr.'s fund, 1789 Capital, holds a stake, and this year the same company landed a federal contract worth roughly $620 million. the timing raised enough questions to trigger a congressional push for the records.
Every company in this thread so far chased the magnet gap on private money – GM, Stellantis, venture funds. This is the first time the federal government itself is writing checks this size, and the fact that it's tied to the president's family is exactly what's under scrutiny right now.
Government picking a winner this early, in an industry that barely has finished products yet, makes me uneasy no matter whose name is on the investment.
Somebody still has to build the plant. Furnace equipment, sintering presses, engineers who've run a magnet line before – almost none of that supply chain exists in the US yet, and none of it needs a billion-dollar company behind it to get started.
Does a contract this size speed up the fix, or just decide who gets rich fixing it?
3,000 PRE-ORDERS FOR A ROBOT HORSE IN UNDER ONE HOUR.
This isn't a concept video. Someone is actually riding it, live, on a public expo floor in Shanghai.
DaxAI built it with legs that switch into wheels mid-stride – walk over rocks, then roll the moment the ground goes flat. 320 kg, 40 km/h top speed, 100 km range on a charge.
Robot dogs never looked like something you'd actually want to sit on. this one does, and watching the wheel-to-leg switch happen live is the part that actually got me.
No price is public yet, and a clean demo on a trade-show floor isn't the same as years of daily riding on real terrain outside a booth. That gap is exactly where "real" and "rumor" usually split.
Three thousand people didn't wait to find out. They pre-ordered anyway.
Robots cost less per hour than the coffee your intern brings you. Nobody's buying them anymore — they're renting instead, and three continents built three totally different price points at once. Evidence, not hype. https://t.co/PS5gul4KO5
@LiveTechProof It's not a fix, it's a hedge – Apple locked in a domestic supplier before China could squeeze it again. Compare that to Niron's approach a few days ago: two very different bets, both live right now.
$500M FROM APPLE, $400M FROM THE PENTAGON – SAME MINING COMPANY, DAYS APART.
Same magnet powering your iPhone's speaker spins the wrist joint on a humanoid robot arm. Different device, same chokepoint.
MP Materials runs Mountain Pass, the only active rare-earth mine in the US. In July 2025 the Department of Defense bought $400 million in preferred stock, becoming the company's largest shareholder – governments don't usually buy into a mine like this, and honestly, i had to reread that line twice.
Days later, Apple signed its own $500 million supply deal, prepaying $200 million up front. The magnets ship in 2027, built entirely from recycled scrap – the same year Niron's rare-earth-free plant in Minnesota comes online too.
Recycled or not, these are still rare-earth magnets, and the underlying dependency doesn't go away – it just relocates onto American soil, while China still controls close to 70% of the global supply everyone else draws from. that part doesn't get fixed by either deal.
Apple just spent half a billion dollars betting the chokepoint doesn't need to disappear – just needs an American owner.
$150 MILLION JUST FUNDED THE ONE THAT DOESN'T.
EVERY ROBOT ACTUATOR SPINS ON A MAGNET CHINA CONTROLS.
AgiBot, Unitree, Optimus – name a humanoid, and you've named a magnet supply chain that runs through China first.
Niron Magnetics builds magnets from iron and nitrogen instead. Neodymium, dysprosium, praseodymium – the metals China refines almost all of – simply aren't in the recipe.
GM Ventures and Stellantis Ventures put real money in back in 2023, before anyone outside Minneapolis was paying attention. Real automakers, real supply contracts.
Ground broke on the first commercial plant in Sartell, Minnesota last September. Twelve days ago the Department of War's Office of Strategic Capital committed up to $150 million to help finish it – still needs final sign-off before it's locked in.
i wanted this to be the fix for the chokepoint. it's closer to a start.
Sartell opens in 2027, built to make 1,500 tons of magnets a year. Global demand for permanent magnets sits around 280,000 tons today. The entire plant covers roughly half a percent of what the world already needs – before robots even add their own demand on top.
The chokepoint just found its first American challenger. It's the size of a keyhole.
@Myrqen Tough is one word for it. The bet only pays off if 20,000 units next year come with a margin that's climbed back up, not just more units at the same shrinking cut.
ONE COMPANY OUTSHIPPED TESLA, FIGURE AI, AND AGILITY ROBOTICS COMBINED. THEN ITS PROFIT FELL BY HALF.
Unitree moved 5,500+ humanoid units in 2025. Those three American companies, added together, didn't reach that number.
Then Q1 2026. Revenue up 68%. Adjusted profit down 52%.
And the target for this year is 20,000 units - close to four times last year's volume, set while the margin is moving the wrong way.
i keep going back and forth on what that actually means. selling more while earning less usually means a company buying market share on purpose. it can also mean one that priced itself somewhere it can't climb back out of. from the outside those look the same for a while.
Somebody at Unitree looked at a falling margin and decided the answer was to build four times as many. That decision is the real story here, and it won't resolve for another year.
NEITHER ONE IS AMERICAN. TOGETHER, THESE TWO COMPANIES BUILD 3 OUT OF EVERY 4 HUMANOID ROBOTS ON EARTH.
Here's the footage that actually proves it. AgiBot's G2 robots working a real production line in Nanchang, China - not a demo stage, not a lab. Real tablets, real grippers, over 99% accuracy at a 19-second cycle time, alongside human coworkers.
AgiBot and Unitree together account for roughly three-quarters of every humanoid unit shipped worldwide in the first half of 2026. Add the rest of China's manufacturers and it climbs to 97%.
honestly, this caught me off guard. I follow this space closely and I still pictured Tesla or Figure when I thought "who's ahead." Turns out the real leaderboard has been sitting in a factory in Nanchang the whole time, mostly unnoticed outside China.
Worth being fair here - shipped isn't the same as profitable. Unitree's own margins took a hit even as its volume grew, so leading in units doesn't mean the economics are settled yet.
But if you watch that footage, you'll see why this deserves more attention than it's getting. This is happening right now, not in five years.