After three incredible years at Apple, I’m heading back to Bengaluru to try and build something exciting. If you’re passionate about AR, let’s connect. I would love to explore opportunities to collaborate!
Stay hungry, Stay foolish!
@SkildAI Genuinely cool. I'd argue the harder problem isn't teaching a robot to play soccer, it's bringing a legend to life in front of you. That's an AR and physical-response problem more than a robotics one.
AI won’t decide humanity’s future. It will reflect it. Feed it fear, get fear. Feed it hope, get abundance.
I’m an optimist. If I’m right, we thrive. If I’m wrong, nobody’s around to gloat.
Can’t lose 😆
I want to play football with Messi in my backyard.
Not on a screen. Not in VR. The actual ball at my feet moving like he actually kicked it.
Here’s the physics problem standing between me and that, and why I think it’s solvable in 12 months 🧵
1/ A pro kick = ~25 joules of energy, ball exits at ~25 m/s, contact lasts ~10 milliseconds.
To make a ball move like Messi kicked it, you need to deliver that exact impulse from inside the ball. On command. In any direction. In under 50ms.
Newton doesn’t care if the kick came from a foot or a spring. He just cares about momentum.
2/ So the ball becomes the entire project.
Inside a regulation 450g sphere, you need:
- stored mechanical energy (springs, gas, or a flywheel)
- a 2-axis gimbal that vectors in <20ms
- sensors that know when a real foot hit it
- a radio link to the field’s edge compute
- enough battery for ~10 min of play
Then you swap balls like F1 tires until the tech improves.
3/ “Why not just VR?”
Because the moment your foot expects contact and gets nothing, the illusion dies. The magic isn’t in the headset. It’s in the ball doing exactly what your eyes told you it should.
This is the difference between watching football and playing it.
4/ The hard part isn’t the AR. The AR is solved. The AI Messi is solved (give it 20 years of his match footage).
The hard part is one mechatronics problem: build a sphere that can kick itself with conviction, survive being kicked back, and fit a battery.
It’s basically a tiny, very polite combat robot.
5/ The closest existing expertise is:
- @SpaceX attitude control (reaction wheels, fast actuators in tight mass budgets)
- Missile guidance and fighter jet control systems
- BattleBots (storing energy, releasing it explosively, surviving impact)
The Venn diagram has maybe 50 people in the world.
6/ Hey @elonmusk could I borrow one of your attitude control engineers for a year? I promise to return them with stories.
Will trade: equity, the satisfaction of inventing a new category, and a goal assist from Messi.
7/ Long-term, the football is just the iPhone.
If a ball can be jointly kicked by a person and a ghost, then you can rally with Nadal on a tennis court at sunset. You can face Michael Holding’s bouncer and feel why batters of that era didn’t sleep. You can take a free throw from MJ’s shoulder.
Every great human who ever lived becomes available to learn from, physically.
8/ We call this Curios Exploration. Each project starts with one question.
This one’s question was: what if distance, time, and death stopped deciding who gets to stand next to greatness?
The thesis is simple: the boundary between digital and physical should disappear, starting with the objects we already love. The ball is just where we start, because every kid on earth understands what it should do.
9/ If you’re a mechatronics engineer who reads this and your first reaction is “this is crazy” followed by “…but here’s how I’d do it”, that’s exactly who we’re looking for. DMs open.
If you’re @elonmusk and you read this: I was serious about the engineer. Also serious about the Messi assist.
The modern "Simulation Hypothesis" seems just like ancient Hindu philosophy with updated tech vocabulary.
The Simulation = Maya (projected illusion)
Base Code = Brahman (pure consciousness)
The Experiment = Leela (divine play)
The Player = Atman (the base reality in disguise)
We traded the analogy of a dream for a video game, but the core realization is millennia old.
Would like to get your thoughts on what the core difference is in your analogy.
@elonmusk@grok What’s the best way to import all the past conversations from ChatGPT to Grok? Historical context is the reason why I go back to ChatGPT. If we can seamlessly transfer those conversations to Grok, I don’t think I would ever need to go back.
After three incredible years at Apple, I’m heading back to Bengaluru to try and build something exciting. If you’re passionate about AR, let’s connect. I would love to explore opportunities to collaborate!
Stay hungry, Stay foolish!
Hi @elonmusk I am not good at physics. Need your help.
1. Is it possible to build a camera technology that takes pictures and videos in 3D?
2. Is it possible to build a wearable device that allows you to experience those images and videos in 3D?
Look forward to your guidance.