Humyn Labs Funding: $47 Million Series A for Robot Training Data
Humyn Labs funding: the Bengaluru physical AI startup raised $47 million, with $26 million in equity led by Accel and $21 million in venture debt.

Quick Take
- Bengaluru-based Humyn Labs raised $47 million in a Series A round: $26 million in equity led by Accel and $21 million in venture debt led by Lighthouse Canton.
- It films people doing real work in factories, warehouses, farms and homes, and turns the footage into training data for robots and AI models. About 90% of its revenue comes from the US.
- Next: the Humyn Labs funding goes into capture hardware, its data engine and robot deployment infrastructure in Brazil, South Korea and India.
In this story4 sections
Humyn Labs, a Bengaluru-based startup that supplies real-world training data for robots, has raised $47 million in a Series A round. The Humyn Labs funding is made up of $26 million in equity led by Accel and $21 million in venture debt led by Lighthouse Canton.
Prosus Ventures, BeeGlobal, Bajaj Finance’s Finserv Intelligence, Bajaj Alts AI Opportunities Fund, Dallas Venture Capital, Lighthouse Canton, Jump Trading and angel investor Anita Pramoda joined the equity round, according to TechCrunch and Business Standard. The debt came from Lighthouse Canton Growth Debt Fund II, with Trifecta Capital taking part.
Humyn Labs says it will spend the money on its own capture hardware and data engine, and on deployment-readiness infrastructure in Brazil, South Korea and India. It also plans “robot gyms”, centres for testing models and robot hardware in real-world settings, Business Standard reported, along with a bigger presence in San Francisco, London and Seoul.
What Humyn Labs Does
Humyn Labs records how people do specific jobs in factories, warehouses, farms and homes, based on what AI labs and robotics companies need. Its software turns those recordings into labelled, training-ready data covering sight, sound, motion and touch.
The company films people with its own two-camera devices while they do their usual work or household chores, not staged tasks, TechCrunch reported. Its software tracks hand movements and how people handle objects, and local partners find collection sites and supervise the recording.
Also read
It collects data in more than 20 countries across Latin America and Asia, with Brazil, India and South Korea as its main hubs. It has 51 full-time employees, most of them in Bengaluru.
“A robot does not learn a job from the internet.”
Ishank Gupta, co-founder, Humyn Labs, as quoted by Sahyadri Startups.
Humyn Labs Funding, Revenue and Customers
Humyn Labs says it has 12 paying customers, including frontier AI labs, robotics companies and developers of world models. About 90% of its revenue comes from the US, Gupta told TechCrunch. He did not name the customers.
The startup earned $10 million in revenue between April and October, and it has $20 million in signed contracts, with another $20 million in deals not yet finalised, according to TechCrunch. The company also told Business Standard that its annualised revenue run rate crossed $50 million within six months of starting operations.
Nearly half of the $47 million is debt. Co-founder Manish Agarwal told TechCrunch that the debt pays for working capital: Humyn Labs pays its workers and data partners daily or weekly, but AI lab customers usually take 30 to 60 days to pay. He said borrowing is less dilutive than raising more equity.
Pratik Agarwal, a partner at Accel, said “data and deployment are where the real moat will be built.”
Who Is Behind Humyn Labs
Manish Agarwal is the former chief executive of listed gaming company Nazara Technologies. Ishank Gupta was an executive at brewer AB InBev. The two earlier started Kratos Studios, now called KGeN, whose network of contributors in emerging markets helps Humyn Labs collect data inside people’s homes, TechCrunch reported.
Humyn Labs was founded this year, and in April it announced a $20 million commitment to build human data infrastructure for physical AI companies. In October, StartupFeed reported that Indian physical AI startups were weighing a common pay rate of $7 to $10 an hour for people who collect robot-training data, with Agarwal among the founders asking India to set rules for the work early.
Also read
A Crowded Race for Robot Training Data
Robots learn physical tasks from examples of people doing them, and that kind of data is scarce online. TechCrunch named Mecka AI, XDOF and MicroAGI as rivals collecting real-world robotics data, and India-based Human Archive and BuildAI as part of the same wave. It also noted the privacy and consent questions raised by recording people at home.
Agarwal told Business Standard that India is not an immediate market for commercial robot deployment, because costs and real-world reliability do not yet work for many uses. Humyn Labs plans to start in San Francisco, followed by London and Munich, with Seoul and Tokyo also on its list.
“Data collection is just the input side,” Agarwal told Business Standard. “You need to start deploying these models and hardware in the real world.”
What this means for you
If you build robots or physical AI in India, note where the Humyn Labs funding is going: into data and testing infrastructure, not into building robots. If you can run data collection, labelling or real-world test sites, buyers in the US are paying for that work now. And if your customers pay slowly, this round shows how venture debt can carry working capital without giving up more equity.
StartupFeed Insight
The number to watch is not $47 million. It is how much of the $50 million run rate turns into collected cash. Humyn Labs earned $10 million from April to October and holds $20 million in signed contracts, but its revenue comes from 12 customers, mostly in the US. If two or three of those labs change how they buy data, the numbers can swing fast. The debt fits a business that pays workers every week and gets paid in 60 days. The bigger test is the next step: moving from selling footage to running the robot gyms it plans, where models and hardware are tested in real settings. That is the harder business, and the one that would set it apart.
By Harshvardhan Kothari, Technology and Policy Correspondent
Have a tip? Write to us at editorial@startupfeed.in.
