Humyn Labs raises $47M Series A to feed robots real-world data
Bengaluru's Humyn Labs raised $47M ($26M equity, $21M debt) to supply frontier AI labs with real-world recordings from 20+ countries, booking $10M revenue in seven months.

Why it matters
- Humyn Labs raised $47M Series A on Nov 18, 2026: $26M equity led by Accel, $21M venture debt led by Lighthouse Canton Growth Debt Fund II.
- The company reported $10M revenue from April to October 2026, plus $20M in signed contracts and $20M in unfinalized deals.
- Humyn has 12 paying customers, with about 90% of revenue from the United States.
- Data collection spans more than 20 countries, with Brazil, India and South Korea as main hubs.
Bengaluru-based Humyn Labs has raised $47 million in Series A financing to build a business around the one thing AI labs can't easily synthesize: recordings of real people doing real tasks in real homes. TechCrunch reported the round on November 18, 2026. The money is split between $26 million in equity led by Accel and $21 million in venture debt led by Lighthouse Canton Growth Debt Fund II.
The equity portion drew a broad syndicate. Prosus Ventures, BeeGlobal, Bajaj Finance's Finserv Intelligence, Bajaj Alts AI Opportunities Fund, Dallas Venture Capital, Lighthouse Canton and Jump Trading all participated, alongside the lead. On the debt side, Trifecta Capital joined Lighthouse Canton. The structure matters as much as the size, and it says something specific about how this company burns cash.
Who is behind Humyn Labs?
Manish Agarwal and Ishank Gupta founded the company in February 2026. Agarwal was previously CEO of Indian gaming company Nazara Technologies; Gupta is a former AB InBev executive. The two had already founded one company together, Kratos Studios, now known as KGeN. That history isn't incidental. The founders said KGeN's contributor network across emerging markets is what helps Humyn get its recording equipment inside people's homes in the first place.
Humyn employs 51 full-time staff, most of them in Bengaluru, with local partners spread across its collection countries. In barely a year, the founders have moved from gaming networks to physical-world AI data.
What does the company actually sell?
Humyn supplies training data for AI and robotics models that operate in physical environments. Its own two-camera hardware records people performing everyday tasks. The company's software then processes and labels the footage — tracking hand movements, object interactions and other fine detail — to turn raw recordings into something a model can learn from.
The key design choice is authenticity. Rather than staging activities, Humyn records people doing their usual work or household chores, while local partners identify sites and supervise the process. Collection spans more than 20 countries across Latin America and Asia, with Brazil, India and South Korea as the main hubs, Agarwal said.
The platform's description lays out a pipeline that goes well beyond video. It spans four sensory channels:
- Sound: speech audio in WAV and FLAC formats
- Sight: first-person video
- Motion: traces from inertial measurement units
- Touch: grip-force data
All of it passes through validation, processing, labelling and human review before being combined into a synchronized training signal. Listed output formats include MCAP, RLDS, LeRobot v3, ROS Bag, WAV, FLAC and JSON — the file types that robotics researchers and RL pipelines actually consume. In plain terms, Humyn is selling synchronized, multi-sensor recordings of human behavior, packaged in the formats robot-learning teams already use.
How much demand is there?
Gupta said Humyn has 12 paying customers spanning frontier AI labs, robotics companies and developers of world models, and that roughly 90% of its revenue comes from the United States. He declined to name any specific customers.
The revenue picture suggests the demand is more than hypothetical. Agarwal reported:
- $10 million in revenue between April and October 2026
- $20 million in additional signed contracts
- $20 million more in deals not yet finalized
That last figure is prospective business, not part of the signed-contract total. Still, a company founded in February 2026 booking $10 million in seven months, with $20 million contracted on top, is an unusually fast ramp for a data-supply business. It also reflects where the market is heading: as labs push from chatbots into embodied AI, physical-world data has become a scarce, priced commodity.
Why did the round include $21 million of debt?
The venture debt isn't decoration — it solves a specific cash-flow problem. Agarwal said growth has increased Humyn's working-capital needs because the company pays its workers and collection partners daily or weekly, while its AI-lab customers typically take 30 to 60 days to pay.
Borrowing bridges that gap. And, as Agarwal framed it, debt causes less dilution than selling more equity to cover the same gap. For a founder-led company scaling fast, that's a defensible trade: use debt to finance the receivables, keep the equity cap table cleaner.
The mechanism is worth spelling out. Every new collection contract increases the mismatch — money flows out to collectors within a week, money flows in a month or two later. Faster growth means a wider gap, which means more borrowing. In plain words: the debt exists because Humyn pays people faster than its customers pay it.
Why does this round matter beyond Humyn?
The financing is a data point in a broader shift. Frontier labs have largely exhausted high-quality text and are now competing for recordings of the physical world — video, motion, touch — to train robots and world models. A company that can reliably collect labeled, consented, multi-sensor human-activity data from 20-plus countries sits on the supply side of that scramble.
It also shows how the data-labeling industry is evolving. Earlier generations of Indian data vendors handled annotation and QA for text and images. Humyn's model is vertically integrated: proprietary camera hardware, in-house processing software, a global collection network and standardized robotics output formats. The stakes are commercial as well as technical — whoever supplies the training corpus for embodied AI shapes what those systems learn about how humans actually move and work.
Humyn's central operating challenge, as its own numbers make clear, is sustaining a distributed collection network and its processing pipeline while covering the delay between paying for recordings and getting paid by customers. The reported revenue and contracts point to early demand; the payment cycle shows the cost of meeting it. Whether the $47 million bridges that gap long enough for the $20 million in unsigned deals to convert will define the company's next year.
Source: unite.ai
More from Sophie Lindqvist
Show full bio
Staff writer covering regulation, policy and their impact on artificial intelligence and machine learning at AI In Context.
749 articles
Related articles
- Isomorphic Labs raises $5B to push AI-designed drugs into trials
- Isomorphic Labs raises $5 billion, its second mega-round in six months
- Google's $15bn AI project draws land seizure claims in Andhra Pradesh
- Blackstone's $1.3 Trillion Bet: AI Debt Fears Are Overblown, Says Jas Khaira
- INFevo Raises Over RMB 100M to Build AI Virtual Cells and Robotic Labs