Mecka AI, a startup that gathers and analyzes human-motion data for robotics models, is reportedly nearing a Sequoia Capital-led financing at a valuation of about $500 million. The deal’s size has not been reported, and its terms are not final, according to TechCrunch.
If completed, the round would arrive only three months after Mecka announced a $60 million financing led by Framework Ventures. The speed of the prospective follow-on deal is notable, but the larger signal is about where robotics investors see a constraint: not simply in model architecture or robot hardware, but in access to useful physical-world training data.
The data problem is becoming a robotics business
Mecka pays people to record themselves doing everyday work and tasks—such as making coffee or fixing cars—using body sensors and smartphones. This produces what is often called egocentric data: recordings that capture how a person sees and moves through an activity.
The company’s premise is that broad-purpose robots, including humanoids, need many more examples of people manipulating objects and completing real-world tasks. Unlike text and image datasets, physical interaction data is difficult to source at scale. It must account for motion, context, object variation, environments and the sequence of decisions involved in getting work done.
That makes collection an operations-heavy business. A useful dataset depends on recruiting contributors, designing task protocols, managing hardware or capture tools, checking quality, labeling data where needed and meeting customer requirements around coverage. For enterprises building robots, that work may be too specialized and slow to establish in-house for every use case.
Mecka has not publicly named customers. But its approach fits a wider push to supply robotics companies and AI labs with data captured from the physical world, alongside methods such as teleoperation.
Why investors are moving quickly
The prospective valuation follows a broader rise in funding for companies positioned around robotics data. TechCrunch recently reported that XDOF, another real-world data-collection startup, was discussing a round at a $1.2 billion valuation. Meanwhile, human-data platforms associated with large language models, including Scale AI and Micro1, are expanding their attention beyond language-model workflows.
For investors, these businesses offer a familiar framing: become a critical supplier in a fast-growing AI stack. The comparison Mecka itself draws is to companies that built human-data infrastructure for LLMs. But robot data may be harder to standardize. A task dataset that helps a robot in a warehouse may not transfer cleanly to a restaurant, repair shop or home. The value is likely to sit in data quality, task coverage, collection reliability and the ability to turn raw recordings into training-ready inputs.
Mecka co-founders Josh Gao, Mogen Cheng and Jason Chong previously worked outside robotics, while co-founder Duy Nguyen focuses on operations. That background may be consequential: scaling contributor networks and repeatable collection programs is as much a marketplace and execution challenge as a research challenge.
What operators should watch
For robotics builders, the immediate question is whether third-party data suppliers can provide reliable task-specific coverage faster and more cheaply than internal collection. Procurement teams should look beyond dataset volume: capture modality, annotation standards, rights to use the data, geographic coverage, task diversity and refresh cadence will affect whether data can support deployment.
For founders, Mecka’s reported deal suggests that the robotics opportunity is widening beyond robot makers themselves. Tooling, evaluation, simulation, teleoperation and data operations may all become strategic layers if physical AI adoption accelerates.
The key test for Mecka will be conversion from collection capacity to durable customer value. The company projected in early June that it would finish 2026 with a $100 million annual run rate, according to comments Gao made to Fortune. Whether it reaches that target—and whether the reported financing closes—will indicate how quickly customers are turning the demand for robot-training data into recurring spend.
