1X

Léo

@LeoKharon

Robotics essentials: TELEOPERATION Behind a surprising number of autonomous robots, there is a human in a headset. Teleoperation is how you drive a robot by hand, and it is doing two jobs at once. It is how nearly all robot training data gets made, and increasingly, it is the product itself. 1X ships its home robot for $20,000, and when it can't do something, a remote operator takes over. A Wall Street Journal reporter spent a day with one and did not see it complete a single task on its own. The company's position is refreshingly blunt: if we don't have your data, we can't make the product better. Figure's CEO calls shipping teleop "soy stuff" and says they won't do it. But the part that gets missed is that teleop itself is not a solved technology. There is no standard rig, and the competing designs don't even agree on what a demonstration is. A leader-follower rig is a small copy of the robot's arm that you move with your hands, and it works beautifully: in a head-to-head, 92% success against 72% for VR controllers. It costs $300 and it fits exactly one robot model. A handheld gripper with a camera on it lets a human do the task themselves, no robot needed, three times faster than teleoperating, for $371, but it captures no robot's joints and only ever a two-finger grip. A VR headset gives you a human's head and hands at 60 Hz and no sense of touch at all. An exoskeleton gives you torques and fits one body. Each rig records a different idea of what the demonstration even was. Nobody is converging: 2026 produced more of them, not fewer. The thing they nearly all drop is force. That is not cosmetic. Add force feedback to the training data and success on contact-heavy tasks jumps by 40 points. The reason it's missing anyway is a result from 1989: a delayed connection is not passive, so a force loop that is perfectly stable with no lag will oscillate once you add any. You can have cheap, remote, or haptic. Not all three. Latency doesn't degrade teleop gently either, it falls off a cliff. In one study, task completion went from 93% at 75 milliseconds to 50% at 150, to 10% at 225. Meanwhile the data factories are running anyway. Chinese collection centres put people in VR headsets and sensor exoskeletons, folding clothes for the robot to copy, for roughly 200 yuan a shift. JD is targeting ten million hours. Here is the tell: Nvidia trained its humanoid model on 88 hours of real teleoperated data, then used a video model to inflate it roughly tenfold into synthetic hours. The whole industry treats teleop data as precious, expensive and unscalable, and spends enormous ingenuity on not having to collect any more of it.

mete

@VaderResearch

Two robotics leaders. Opposite timelines. 1X CEO Bernt Børnich: ✸ hard takeoff within 10 years. ✸ robots build robots. ✸ labor becomes abundant. Agility CRO Jonathan Hurst: ✸ a slow snowball. ✸ no singularity. ✸ better tools, more data, gradual progress. https://t.co/L8yTvb3IAJ

0xSero

@0xSero

More intelligent, fits in 56gb of memory, perfect for 1x rtx pro 6000 terminal-bench-2.1 so far 60% on terminal-bench-2.1 but only 30% through. https://t.co/SUU6rk2yPr

Will Manidis

@WillManidis

genuinely disgusting the number of VCs sitting on maybe 1x, mostly 0.75x non-mark-to-market, 2021 books spending today tweeting dismissively about one of the most interesting runs in recent asset mgmt history. try playing in the big leagues buddy, i'm sure you could do it.

Humanoids daily

@humanoidsdaily

🚨 Is the timeline for physical AGI shrinking? Speaking on the @theallinpod with @jason 1X CEO Bernt Børnich dropped a massive prediction: we are less than a decade away from a physical AI "hard takeoff." His personal bet? Just 3 years. What his 3-year hard takeoff actually looks like: ▪️ Robots manufacturing robots ▪️ Autonomous chip fabs & data centers ▪️ Self-sufficient mining & refining ▪️ A true abundance of physical labor 1X's strategy to get there relies on an open ecosystem and training the NEO humanoid on internet-scale video.

Branxi

@OG_Branxi

can a perp DEX that has been sitting in top 3-5 by volume for months still rug traders right before TGE? yes, look at @grvt_io timeline: • July 10 - official Multiplier Plan announced: “1x - receive at TGE” “2x - after 4 months” “4x - after 8 months” • choice until July 17 • “choices are final” now comes the fun part July 28 after the deadline real rules dropped: • everyone (including 1x) gets 12-month vesting in tranches • even 1x receives only a portion at TGE (depends on activity) • multiplier applies only to the deferred first tranche, not the full allocation • shares diluted to ~0.63 $GRVT per 1 share and this after almost 3 years of points farming I’m glad I never made a single trade there also, let’s not forget what happened to the tokens of other perp DEXes that did TGE this year after some time: #EDGE as an example, 4 months post-TGE: -72% from ATH -45% from TGE simple lesson: if a project keeps dragging out seasons for years, it's no accident they know post-TGE is gonna be brutal

Seasoned finance leader in autonomy, aerospace, and life sciences; previously CFO at Zipline, Virgin Galactic, and Verily Life Sciences.

People and talent leader with experience at the Obama White House, Teach For America, and TOMS, now scaling 1X’s global team.

Robotics and AI researcher formerly founding research scientist at Luma AI, now leading the 1X World Model Lab on large-scale embodied world models.

Bill Nash

Chief Financial Officer

Previously at Ursa Major · San Francisco Bay Area

Joined Jun 2026

Sanket Goyal

Engineering Tech Lead

Previously at Monarch Tractor · San Francisco Bay Area

Joined Jun 2026

Nadir Khan

Senior EHS Lead

Previously at Cornerstone Building Brands · San Carlos, California, United States

Joined Jun 2026

Carlos Saenz

Staff Enterprise IT Engineer

Previously at edX · Greater Sacramento

Joined Jun 2026