India’s Humyn Labs raises $47M to feed U.S. robotics companies the data they need

Startups & Funding

Humyn Labs Raises $47M to Feed U.S. Robotics Firms Real-World Data

Humyn Labs raised $47M to collect and label real-world human activity data for U.S. robotics firms, booking $10M revenue since April across 20-plus countries.

By
Rebecca Stone
Filed
Channel
Startups & Funding
Read
3 min read

Bengaluru-based Humyn Labs has raised $47 million to solve one of robotics' most stubborn bottlenecks: getting enough real-world footage of humans doing everyday tasks to train machines that must operate in physical spaces.

The round, announced now, splits into two pieces. Accel led $26 million in equity, with Cactus Partners, Prosus Ventures, BeeGlobal, Bajaj Finance's Finserv Intelligence, Bajaj Alts AI Opportunities Fund, Dallas Venture Capital, Lighthouse Canton and Jump Trading joining. Lighthouse Canton Growth Debt Fund II led another $21 million in venture debt, with Trifecta Capital participating.

That debt is not decoration. It's working capital. The startup pays its data collection workers and partners daily or weekly, while its AI lab customers typically take 30 to 60 days to pay up. Borrowing bridges that gap and dilutes less than selling more equity, co-founder Manish Agarwal told TechCrunch.

Who is behind Humyn Labs?

Agarwal knows scale from a previous life — he was CEO of Indian gaming company Nazara Technologies. He founded Humyn Labs in February with Ishank Gupta, a former executive at brewing giant AB InBev. The pair had already built Kratos Studios, now known as KGeN, which runs a contributor network across emerging markets. That network now does double duty: it helps Humyn Labs collect data inside people's homes.

The company keeps just 51 full-time employees, most in Bengaluru, plus local partners across its collection countries.

How does the data operation work?

Humyn Labs built its own two-camera hardware to record people performing everyday tasks. Software then processes and labels the footage — tagging what happens in each clip so AI models can learn from it. The labeling goes deep: the company tracks hand movements and object interactions, turning raw recordings into structured training data.

Collection spans more than 20 countries across Latin America and Asia. Brazil, India and South Korea serve as the main hubs, Agarwal said.

Not everyone records staged activities. Humyn Labs films people doing their usual work or household chores, working with local partners to identify sites and supervise the process. "The worker is a worker who is incentivized, but he needs to be monitored and supervised for the right input," Agarwal said.

What do the numbers say?

The commercial traction is concrete. Humyn Labs generated $10 million in revenue between April and October, Agarwal said. It holds another $20 million in signed contracts, with a further $20 million in deals yet to be finalized. Twelve customers pay today — frontier AI labs, robotics companies and developers of world models — and about 90% of revenue comes from the U.S., Gupta said, though he declined to name clients.

"Scale, diversity, and quality are important aspects, and they need to work in tandem," Agarwal said.

Is India becoming a physical AI data hub?

Over recent months, India has emerged as a center for collecting training data for so-called physical AI. Startups including Human Archive and BuildAI tap the country's large workforce to record everyday human activity. Globally, Humyn Labs competes with Mecka AI, XDOF and MicroAGI, which also collect real-world data for robotics training.

The practice raises real questions. Recording people inside their homes touches on privacy and consent — issues that will only grow as the market does. Agarwal's answer is the supervised, incentivized-worker model, but regulators and customers will keep pressing.

With $47 million in hand, daily payout obligations and a customer base concentrated in U.S. AI labs, Humyn Labs now has the runway to chase the $40 million sitting in signed and pending contracts — if it can keep quality and scale moving in tandem.

Source: techcrunch.com

Share this article:

More from Rebecca Stone

Rebecca Stone

Show full bio

Correspondent covering technology and product news in Semiconductors, chip manufacturing, hardware, AI compute at Chip Dispatch.

1,104 articles

Related articles

« Previous articleNext article »