Mecka AI Eyes $500M Valuation as Sequoia Bets on Robot Training Data
Mecka AI is closing a Sequoia-led round at roughly $500M valuation, just three months after a $60M raise, as demand for humanoid robot training data surges.
Mecka AI, a two-year-old startup that pays people to record everyday movements for robot training, is close to completing a Sequoia Capital-led funding round at a valuation of approximately $500 million, according to two people familiar with the deal. The raise follows a $60 million round from just three months ago and positions Mecka inside a fast-moving race to supply the physical-world data that humanoid robot developers cannot yet generate on their own.
What happened
| Detail | Fact |
|---|---|
| Reported valuation | ~$500 million |
| Lead investor | Sequoia Capital |
| Previous round size | $60 million (June 2026) |
| Previous round lead | Framework Ventures |
| Previous round participants | Menlo Ventures, SV Angel, Kindred Ventures |
| Projected 2026 ARR | $100 million (as of early June) |
| Founded | 2024 |
Mecka AI collects human motion data to train humanoid robots and other robotic systems. The company pays participants to film themselves performing ordinary tasks, such as making coffee or repairing cars, using body sensors and smartphones. This approach, known in the industry as “egocentric” data capture, records the world from a first-person perspective and is used alongside other methods like teleoperation (where a human remotely controls a robot to generate movement data).
The company was founded in 2024 by four co-founders. Canadians Josh Gao and Mogen Cheng previously built a restaurant fintech startup. Jason Chong joined Coinbase after it acquired his crypto exchange. Duy Nguyen, the only non-Canadian, handles operations. None of them came from a robotics background.
The round’s precise size has not been confirmed, and terms are not yet final. Neither Mecka AI nor Sequoia responded to comment requests, according to TechCrunch.
Why it matters
Humanoid robots need enormous volumes of real-world physical data to learn generalizable behaviours. Synthetic data and lab simulations have limits. Companies like Figure, Physical Intelligence, and others building general-purpose robots depend on firms like Mecka to fill that gap, the same way large language model developers leaned on Scale AI, Mercor, and Surge for text and image labelling.
The speed of Mecka’s successive raises tells you something about supply pressure. Three months between a $60 million round and a potential $500 million valuation is fast by any standard. Competitor XDOF was reportedly approaching a new round at a $1.2 billion valuation, reported by TechCrunch the week before this story broke. Scale AI and Micro1, originally known for LLM data, are also expanding into physical robot training data.
Gao told Fortune in early June that Mecka expected to finish 2026 at a $100 million annual run rate. If that target holds, the ~$500 million valuation implies a roughly 5x revenue multiple, which is reasonable for a high-growth data infrastructure business but leaves little margin for a revenue miss.
Our take
The Mecka story is a clean example of a pick-and-shovel play: rather than betting on which humanoid robot wins, investors are funding the company selling the training data every competitor needs. That logic is sound, and the founding team’s willingness to chase a problem outside their background (restaurant fintech to robot data) is a reasonable bet when the market is still wide open.
What we would watch closely: the $100 million ARR projection assumes sustained demand from robot labs that are themselves burning significant capital. If the broader humanoid sector hits a funding slowdown, demand for training data contracts fast. Mecka’s valuation at that point would look a lot less comfortable.
For businesses tracking AI infrastructure, this round is further evidence that data pipelines, not just model architectures, are attracting serious capital. The same pattern played out in language AI, where data labelling firms scaled well before the models themselves reached commercial viability. You can follow more stories in this vein on our AI news coverage. If you are thinking about how AI integration touches your own operations, our AI Integration service is a practical starting point.
What to watch next
- Watch for Mecka to publicly confirm the round size and any named customers, which would clarify how diversified their revenue actually is.
- Track XDOF’s rumoured $1.2 billion round. If both close at those valuations, expect more entrants and possible price compression on data collection contracts.
- Monitor Scale AI and Micro1’s robotics data offerings. Incumbents with existing enterprise relationships could undercut newer entrants on price.
Frequently asked questions
What does Mecka AI actually do?
Mecka AI pays people to record themselves doing everyday tasks, like making coffee or fixing cars, using body sensors and smartphones. That motion data is then used to train humanoid robots and other robotic systems.
How much has Mecka AI raised in total?
Mecka raised $60 million in a Framework Ventures-led round announced in June 2026. It is now reportedly closing a new Sequoia-led round at roughly a $500 million valuation, though the new round size has not been disclosed.
Who founded Mecka AI?
Mecka AI was co-founded in 2024 by Josh Gao and Mogen Cheng (who previously built a restaurant fintech startup), Jason Chong (who sold a crypto exchange to Coinbase), and Duy Nguyen, who handles operations. None have a robotics background.
What is egocentric robot training data?
Egocentric data is footage and sensor readings captured from a first-person point of view, as if seen through the eyes of the person performing a task. Robot developers use it to teach models how humans interact with objects in real environments.
