Seattle, WA
- Tweets 11,848
- Following 1,296
- Followers 9,616
- Likes 17,716
Pinned Tweet
After 2+ years in the robotics data space, we are shutting @Eidon_AI down.
The thesis was right. But the business is brutally hard.
We close this chapter by open-sourcing everything we built and sharing lessons for anyone venturing into the space.
101
141
1,527
236,864
Sam Padilla retweeted
If you are a
- engineer with 2+ yrs of exp
- based in SF
- fascinated with robotics, infrastructure and energy
- FDE type beat
Exciting space co. might be ur next work home.
DMs open 👀
8
3
98
8,276
Data ops internalization
The data needed to scale a truly general purpose robots doesn't exist on the internet - it has to come from the real world
For the last 4 months, we’ve been building a Figure-exclusive pipeline to scale data collection at higher throughputs. Today, we’re coming out of stealth
5
1,158
Some shots of the great convo with Formant.
When @theSamPadilla shut down EIDON AI, a hardware and data startup building training data for physical AI, he and his co-founders open-sourced the company's data, hardware designs and firmware. Then Padilla published a post-mortem called Robotics Data Is a Broken Business that had a viral moment last week.
We invited Sam to share his hard earned wisdom on our podcast: piped.video/watch?v=cZAd8y5y…
4
721
Spoke with @FormantInc to expand on the lessons from two years in the Robotics Data industry.
Many thoughts I didn't get to cover on the original essay went here. Have been chatting with a lot of people in the space.
There are many interesting patterns appearing.
Yesterday: "There are ~20 people on Earth who can buy a $1M robotics dataset."
Today: The full breakdown is live 🎙️
@theSamPadilla & Henrik unpack physical AI data traps, supply chain chaos, and why VCs pass on data startups.
👇
#PhysicalAI #Robotics
1
1
50
7,413
Sam Padilla retweeted
Read Sam's full article here:
After 2+ years in the robotics data space, we are shutting @Eidon_AI down.
The thesis was right. But the business is brutally hard.
We close this chapter by open-sourcing everything we built and sharing lessons for anyone venturing into the space.
1
2
278
Fun times were had chatting with the Formant team about the ins and outs of the Robotics Data Market.
Full episode coming out soon!
There might be 20 people on Earth who can say yes to a $1M robotics data deal.
"Maybe 50." That's it.
@theSamPadilla built Eidon AI, open-sourced its data, and wrote "Robotics Data Is a Broken Business." He told Henrik why go-to-market in robotics data isn't really go-to-market at all. It's about access.
Full episode of Formant Verified drops Tue, Oct 13. 🎙️#PhysicalAI #Robotics
2
4
282
Thin, easy to wear, comfortable pressure sensing gloves are the way to go.
To teach robots touch, wear it first.
MX MATRIX built tactile data gloves that capture pressure, motion, and temperature straight from human hands as they feel the world.
They weave their own piezoresistive fabric in-house to get right the three things that matter: zero noise when untouched, sensitivity to the lightest press, and durability over thousands of uses.
6
676
In case anyone was curious about the massive subsidizing work going on
BREAKING: Anthropic is making a massive bet that AI will transform the global economy more profoundly than industrialization, electricity and the internet, per Reuters.
Anthropic reported a net loss of $42 billion in 2025, and plans to spend $518 billion on cloud, computing and infrastructure obligations in coming year, according to the prospectus.
1
2
1,279
Quoting to amplify
If you're a robotics researcher working with GPT 6 Astra to do interesting things with robotic control or agentic real2sim, and would like access to open source models (Kimi K3, Qwen 3.8 Flash Next) as a comparison, please DM me!
I'd love to provide you with free access to tokens to help you benchmark the agentic capabilities of these open models, and would be super eager to see what you can do with open source models in this domain.
6
1,702
48,000 downloads of @Eidon_AI's parting open-source tracker dataset on Hugging Face 🤗
Damn.
4
1
20
1,026
Some more thoughts on this:
ponderly.ai/sam/node/n_431df…
May turn it into an essay later.
ponderly.ai @ByCrux
Looking at this makes you wonder if the full humanoid embodiment is truly the best way to go.
A deep rabbit hole to explore: did evolution really optimize for the best possible design?
1
5
594
Claude, make this video go viral. Make no mistakes.
The most advanced AI video editor is coming.
Connect @claudeai or @OpenAI, drop in your clips, say what you want, then make it yours.
Plug in your @Kling_ai, @GeminiApp, and many more connectors.
ALL FOR FREE.
Join the waitlist: montaj.ag/waitlist
1
796
Looking at this makes you wonder if the full humanoid embodiment is truly the best way to go.
A deep rabbit hole to explore: did evolution really optimize for the best possible design?
9
1
35
7,729
"Interfaces"
This is a word we will start to hear a lot in the space. It is clear that visuospatial understanding of LLMs is improving as a function of scale and tooling.
Even setting aside the demos, I am sure anyone who's been working with these agents long enough has some anecdotal evidence of this improvement.
For me, it was the increased understanding of quaternion and Euler space while working on our inverse kinematics resolver: x.lingyaoai.com/theSamPadilla/status/2…
What these interfaces turn out to be is anyone's guess. It may be smaller end-effector VLAs, world models, or simple lower-level policies.
But the improvement in 3D understanding is undeniable. Now, getting that to be fast, cheap, and reliable enough to deploy to different embodiments is a whole other story.
More and more, a future where embodiments are driven by a generalist model plus tooling plus special policies seems very likely.
4
968