Sam Rubinstein (@sirubi_eth) header
Sam Rubinstein (@sirubi_eth) profile picture

Building @THORMachines | Robotics data × Bittensor | Prev Score & Bitcast

Hong Kong
Joined August 2015
Photo by Sam Rubinstein (@sirubi_eth)Photo by Sam Rubinstein (@sirubi_eth)Photo by Sam Rubinstein (@sirubi_eth)Photo by Sam Rubinstein (@sirubi_eth)Photo by Sam Rubinstein (@sirubi_eth)Photo by Sam Rubinstein (@sirubi_eth)Photo by Sam Rubinstein (@sirubi_eth)Photo by Sam Rubinstein (@sirubi_eth)Photo by Sam Rubinstein (@sirubi_eth)Photo by Sam Rubinstein (@sirubi_eth)
> robotics data has always been a quality problem 100%. And from what I can tell the industry is realising the bottleneck has been shifting to verification and QA over just collection.
Robot data has always been a quality problem more than a quantity problem. 3 things decide whether data is usable for deployment: - Was it captured on hardware built for it - Can you track hands accurately outside a lab - Was it validated before anyone trained on it @humynlabs is building all three in-house and that’s impressive. 

Happy to see this shipping from India for the world!
1
100
Been nice seeing some raises in past week or so from the robotics data world: @MosaicoLabs (an open-source robotics data platform) earlier this week, @GenrobotAI last week and now @humynlabs. Def no shortage of capital tackling the data bottleneck in robotics
We're building the infrastructure that takes Physical AI out of the lab and into the real world. And we've raised $47M to make that happen. Physical AI models are getting better every month. But getting robots to work reliably in real-world environments, across full shifts and changing conditions, is a different challenge altogether. The missing piece isn't just better models. It's deployment readiness. At Humyn Labs, we're building the research and deployment layer to close that gap. Here's how: - Enrichment Engine: Our end-to-end data engine transforms real-world work into labelled, training-ready signals with verified provenance. - Deployment Readiness: We work with labs and site operators to deploy robots in those same environments, identify where they fall short, and feed those gaps back into the next generation of models. - Capture Hardware: Our own capture devices across sight, sound, motion and touch record how people actually perform a job, built to each lab's specifications. We've made significant progress since launch: - $50M+ in annualised revenue within six months. - Capture and enrichment operations across 20+ countries in Latin America, East Asia, South Asia and Southeast Asia. Today, we’re announcing $47M in Series A funding: - $26M in equity, led by @Accel with participation from @Prosus_Ventures , @Bajaj_Finserv Intelligence through Bajaj Finance, Bajaj Alts AI Opportunities Fund, BeeGlobal, @wearedallasvc, @insights_by_LC , @jumptrading and a marquee angel investor. - $21M in venture debt, led by @LighthouseCanton with participation from @TrifectaCapital. With this funding, we're expanding our deployment readiness infrastructure in Brazil, South Korea and India, scaling our capture hardware and enrichment engine, and growing our teams across San Francisco, London and Seoul. Our mission is simple: accelerate the deployment of Physical AI in the real world. The next generation of robotics won't be built on models alone. It'll be built on the infrastructure that gets those models working in the real world. Learn more: humynlabs.ai
4
188
It's been great following all the developments this past week from CoRL Some recurring themes that I came across: data you can collect is getting cheap, and checking whether it's any good is the hard part A few talks and papers that stuck with me below Clip is SuperMimic from @qiayuanliao and @berkeley_ai, awarded with CoRL's Best Paper and now open source. It beat the previous best on less than half the motion data, trained on 1 GPU in 3 hours"
1
6
325
My takeaway from the week Collecting robot data is turning into a race to zero. Knowing which data is actually good is where the value is
1
29
Thanks to @sippeyxp for building CoRL PostGame. It's the easiest way to catch up on the panels you missed Link: sippeyxp.github.io/ xitter.cf/sippeyxp/status/212098…
#CoRL2026 ends soon. I missed many sessions, doing work, chatting with others, etc. CoRL recordings is uploaded. But it is super inconvenient to navigate a 8hr long videos. Entering CoRL PostGame, which give you more powerful navigation sippeyxp.github.io/corl26_po… More👇
32
Sam Rubinstein retweeted
THOR (SN112) signed their first client. Demand for robotics training data is going to be yuge - Thormachines is positioned. Are you?
$20K. THOR’s first customer deposit landed onchain before work began. This deposit was matched within our whitelisted miner set for annotation work. Miners are paid as work passes verification. Buyers pay only for accepted units.
1
9
442
Here's to the first of many. Very excited for annotation work on THOR to kick off and for us to contribute to the space and help labs get access to the data they need.
$20K. THOR’s first customer deposit landed onchain before work began. This deposit was matched within our whitelisted miner set for annotation work. Miners are paid as work passes verification. Buyers pay only for accepted units.
5
219
Sam Rubinstein retweeted
$20K. THOR’s first customer deposit landed onchain before work began. This deposit was matched within our whitelisted miner set for annotation work. Miners are paid as work passes verification. Buyers pay only for accepted units.
2
5
3
23
1,683
Insightful piece on where robotics is at now - aligned with what I've seen so far. Demand shifting to quality of data over just quantity Collection is getting easier and it's a race to 0, so the bottleneck shifts to verification of existing data.
60 conversations in SF later, here is everything I learned, and the hot takes most of the people I talked to would argue with. Data, models, deployment, hardware, money, and where the bottleneck goes next.
2
1
3
192
We launched THOR after a lot of conversations with robotics labs about what they actually need from data. Those conversations are now signed contracts. Next is delivering on them. More on this soon.
A data network should start with customer demand. THOR’s first jobs come from signed robotics labelling contracts. Customers fund each job upfront into an onchain deposit contract and pay only for accepted units. A share of those payments funds SN112 alpha buybacks.
1
10
354
Sam Rubinstein retweeted
A data network should start with customer demand. THOR’s first jobs come from signed robotics labelling contracts. Customers fund each job upfront into an onchain deposit contract and pay only for accepted units. A share of those payments funds SN112 alpha buybacks.
2
4
1
15
1,030
So much is happening across Bittensor, it's hard to keep up between: - @reaxlabs topping Hugging Face's Decision Index, - @lium_io GPU rentals up 74% in October to $2.3M and Tencent collaborating with @chutes_ai - hard not to be optimistic about where Bittensor is heading.
NEWS: Tencent, whose Weixin and WeChat platforms reach 1.439 billion monthly users, has selected Chutes, a Bittensor subnet, as a launch partner for its new flagship AI model, Hy4. The partnership puts @opentensor's decentralized AI infrastructure alongside one of the world's largest technology ecosystems. Chutes offers competitive inference pricing at $0.525 per million input tokens and $1.5625 per million output tokens. This collaboration could bring greater visibility to decentralized AI networks and highlight their growing role in supporting large scale AI applications.
1
36
961
Pretty cool Some parts are sped up significantly, and tasks cherry picked. Don't think we can conclude robotics is solved from this, but the progress is really exciting xitter.cf/claudeai/status/211986…
Replying to @claudeai
Fable 5.2 shows substantially improved skills in controlling physical robots, even on tasks it has never explicitly been trained to do. On RoboDojo, a benchmark with a simulated two-arm robot, it outperformed other models and most AI systems built to control robots.
4
178
Sam Rubinstein retweeted
16 seconds of someone unloading a dishwasher. A THOR miner turns that footage into a structured record: eight actions, each with a start and end time. At scale, labels like these become training and evaluation data for robots. Action labelling is THOR’s first task type. Every THOR task is built to the customer’s brief, and work is checked before delivery.
1
1
8
596