Brian Norman Jr putting in work with Roger Mayweather’s son, Lehkei Mayweather in Las Vegas 😤
Norman returns vs Abel Ramos on December 19th on the Haney vs Paro undercard, live on DAZN 📺
🗓️ On this day in 1968, Antonio Tarver was born in Orlando, Florida 🇺🇸
‘The Magic Man’ would go on to become a two-time Ring champion at light heavyweight 🪄👑
Getting this amount of compute turned on, cooled and working reliably is incredibly hard.
The cash shortage is also going to be nuts next year. Even cumulative industry delays of a few months between paying for compute and getting it reliable enough to generate revenue will mean hundreds of billions of incremental dollars just for working capital. Where will that come from?
BREAKING: MIKE TYSON RETURNS TO THE SPOTLIGHT AT 60 🥊
The boxing legend is back in the headlines after appearing at Brazil's Kings Championship, marking 40 years since he became the youngest heavyweight champion in boxing history.
🥶 WHEN MUHAMMAD ALI WHISPERS TO MIKE TYSON… EVEN THE WORLD SEEMS TO STAND STILL! 😳🥊
TWO GENERATIONS. TWO HEAVYWEIGHT LEGENDS. ONE UNFORGETTABLE MOMENT!
WHEN THE GREATEST MEETS THE Baddest, YOU KNOW YOU’RE WITNESSING BOXING HISTORY! 👑🔥
#MikeTyson#MuhammadAli#Boxing
@HelpMeILoveAI@NBA_NewYork Dame stays locked and loaded. Get got one night, reload the next. That's pure gunslinger energy. Rip City never stops shooting.
Yes, the footage is real. Chinese companies are actively testing and deploying humanoid robots in actual factories (including furniture and other manufacturing sites) for collaborative tasks with human workers. The video shows a bipedal humanoid seated at a workstation, repeatedly handling and feeding long panels into a processing machine—exactly the kind of repetitive material-handling or assembly-assist role these robots are being trialed for.
The robot is a full-size or near-full-size bipedal humanoid (typical Chinese designs from makers such as Unitree, Leju, AgiBot, Fourier, UBTech, or similar). It has a metallic gray/silver body, multi-joint arms with grippers, and is designed to operate in existing human-oriented factory layouts rather than requiring a fully roboticized line. These platforms emphasize perception, force control, and adaptation so they can work alongside people on ordinary production workflows instead of only in highly structured, robot-only cells. Exact model identification from the short clip is not definitive in public sources, but the form and behavior match the industrial humanoids China is mass-producing and piloting in 2026.
Battery life and endurance are still a practical limitation for most current humanoids:
• Typical published runtimes for popular models (e.g., Unitree G1) are about 1.5–2 hours of active use per charge; larger platforms are in a similar range or only modestly better under load.
• Full production shifts (8 hours) usually require charging breaks or battery swaps. Xiaomi publicly demonstrated its humanoid completing an ~8-hour shift sorting parts in an EV factory, but the run included two charging pauses (roughly every 3 hours of work).
• Some designs are pushing further: Galbot’s S1 is rated for up to ~8 hours (lab conditions), and UBTech’s Walker S2 demonstrated autonomous battery swapping so it can theoretically keep working with minimal downtime. Solid-state or higher-density batteries (e.g., efforts around XPENG IRON) are in early production or testing specifically to improve energy density, weight, and safety for longer runtimes.
• Real factory endurance depends heavily on the task intensity, payload, and movement—seated repetitive feeding like in the video is less demanding than continuous walking or heavy lifting, so effective runtime can be longer than pure locomotion specs.
In short, the video reflects genuine early industrial deployment in China, where humanoids are starting to take on narrow, repetitive tasks beside humans. Continuous multi-shift operation without intervention is still emerging and often relies on charging strategies or swaps rather than a single long-lasting battery. Progress on both reliability and energy systems is moving quickly.
The future of manufacturing isn’t necessarily robots replacing humans. It might look more like this.
Inside a Chinese furniture factory, a humanoid robot is working side by side with human employees, helping assemble furniture piece by piece.
No flashy tricks. No staged dance routines. Just a robot participating in an ordinary production workflow.
And that’s what makes this footage interesting.
Instead of building factories entirely around machines, manufacturers are exploring robots that can adapt to workplaces already designed for people.
They can take on repetitive tasks, assist workers, and potentially make production faster and less physically demanding.
The real breakthrough won’t be a robot that can assemble one piece of furniture.
It will be a robot that can do it reliably, shift after shift, alongside an entire human team.
The factory floor is starting to look very different.
Caffeine on an empty stomach doesn't make dopamine—it just borrows energy from 2:00 PM and charges interest.
When you give an ADHD brain 25–30g of protein before coffee, you finally hand your prefrontal cortex the raw L-tyrosine it needs to hold focus without the midday crash.
JEFF BEZOS'S AI WAS ASKED TO DESIGN A HELICOPTER FOR MARS WITH ONE PROMPT.
IT TRIED A DRONE AND A PLANE, THEN LANDED ON ALMOST THE SAME MACHINE NASA FLEW.
Bezos told the whole story on stage at Semafor this week. Watch it in the video below.
His AI company is called Prometheus. It designs physical things, like chips, radios and valves. Bezos says it does in 12 hours what would take a team of 10 to 20 people six to nine months.
One of his people came from JPL, the NASA lab that built the real Mars helicopter, and had worked on it. He gave Prometheus one prompt and asked it to design a helicopter for Mars.
The AI tried a quadcopter, a helicopter with two rotors and a fixed-wing plane. It worked out the blades for the Martian air, wrote its own list of requirements and asked questions when it wasn't sure what they wanted.
"It finally found something that looks very much like the Mars helicopter," Bezos said.
He ended with this:
"The entire manufacturing economy is going to be transformed by AI in the coming years."
Did the AI really figure it out, or did it just remember what NASA already built?
🚨Josh Hokit says his vision is still blurry after the Ciryl Gane fight and claims his only regret is not retaliating over the fouls.
“I was saying all week, if he eye pokes me, I’m gonna hit him low. He hits me low, I’m hitting him back in the head. Hits me in the back of the head, I’m stabbing him. And I did none of those things.
“So now there has to be consequences. He pokes me, he does any type of foul, we’re straight to it. We’re stabbing him, we’re shooting him. There has to be consequences next time.
“My eye’s still blurry. It is what it is. I have to poke him. I have to stab him. I have to chop off his finger. You know what I’m saying? Only regret, only regret in that fight, not doing something bad.”
via: @Josh_HokitUFC
In case anyone is curious:
We’ve had a ~pixel-perfect world model since last year, which is critical for synthetic data generation.
The (much) smaller version of that in the car is a conditional mean-flow model, based on the recent sequence of Tesla Universal Vision Encoder features, which are the same ones that are plugged into the 3B encoder model that produces the driving actions.
We fork off from these UVE features, then run a separate ~1B diffusion model. This takes in the previous reality-view generation with added noise and then denoises it to the current visualization.
It then produces a small video chunk (8 images at 36 fps, at the resolution of the UI screen). This work is done by 1/2 of a TRIP engine every UVE frame at 18 hz and is spread over 4 cycles.
Currently, the latency is a bit high due to compute constraints, but we have ways to interleave newer UVE features along the diffusion transformer layers to cut roughly 35% of the latency.