Mecka AI, a New York startup that captures human movement to train robots, has landed $60 million in fresh capital from Sequoia Capital and a roster of heavyweight investors that includes NVIDIA, Microsoft's M12 venture fund, Qualcomm Ventures and Samsung. The company disclosed the Series B round on October 7, bringing its total raised to roughly $128 million since launching less than two years ago.
The funding comes as the robotics industry races to solve what many consider its thorniest problem: how to teach machines to move and manipulate objects with anything approaching human dexterity. Mecka's answer involves strapping sensor rigs to people and having them perform everyday tasks, then translating that motion data into training material for artificial intelligence models.
Sequoia partners Anas Biad and James Flynn led the investment, joined by returning backers Kindred Ventures, Framework Ventures and Neo. The round also attracted operator-angels with serious robotics credentials, among them DoorDash founder Tony Xu, former ServiceNow and Snowflake chief Frank Slootman, and Milan Kovac, who previously oversaw Tesla's Optimus humanoid robot program. Christopher Payne, who served as president and COO at DoorDash, and Jonathan Chadwick, a former VMware executive who now sits on Databricks' board, also participated.
TechCrunch reported in September that Mecka was seeking funding at a valuation around $500 million, though the company declined to confirm that figure when announcing the raise.
The startup claims its revenue trajectory has been exceptional. It says it crossed $100 million in annualized run-rate revenue in June and expects to reach $300 million by year-end, though those figures are company-stated projections that remain unverified by external auditors. "This is the fastest-growing revenue company that we've ever invested in," Framework Ventures cofounder Vance Spencer told Fortune over the summer, adding that "the proof is in the business itself."
Whether that growth holds remains to be seen, but Mecka has clearly struck a nerve with investors betting on embodied AI. The company positions itself as the infrastructure provider for what it calls "physical AI," building both the data pipeline and the deployment operations that robotics companies need but often struggle to assemble themselves.
The business model has two main components. First, Mecka designs multi-sensor hardware that people wear to capture body and hand movements in granular detail—the company says it achieves sub-centimeter precision on hand-pose tracking even in uncontrolled environments. It operates what it describes as global capture fleets, paying contributors to wear these sensors while performing tasks in homes, warehouses and other real-world settings.
Once captured, that motion data gets processed through computer vision systems and multimodal AI models, then packaged into training datasets. Mecka sells access to these datasets, along with evaluation tools and APIs, to robotics labs and companies. But it doesn't stop there. The startup also deploys robots directly for customers, acting as what CEO Josh Gao calls a "modern robotics integrator" that continuously retrains models based on how machines actually perform in the field.
Mecka says it works with several of the leading frontier robotics labs and multiple so-called Magnificent Seven technology companies, though it won't name clients. The company has collaborated with researchers at Georgia Tech, Stanford, UC San Diego, ETH Zürich, MIT and Meta on academic projects, including the EgoVerse dataset published earlier this year, which focused on transferring human motion patterns to robotic systems.
The founding team brings an eclectic mix of backgrounds. Gao and cofounder Mogen Cheng previously sold a restaurant payments technology startup in 2023, according to Fortune. CTO Jason Chong came aboard after selling his own company to Coinbase, while cofounder and COO Duy Nguyen built what Fortune described as a multimillion-dollar sneaker resale operation. None of them came from traditional robotics, which perhaps explains the company's unconventional approach to the problem.
Mecka started small. It raised an $8 million seed round led by Neo in August 2025, according to Silicon Legal, which advised on the deal. A $25 million Series A from Framework Ventures followed in November 2025, then a $35 million extension in mid-2026 from Framework, Menlo Ventures, SV Angel and Kindred Ventures. The company had grown to roughly 40 employees by June, Fortune reported at the time.
The startup made one acquisition along the way, picking up Docula, a three-person Vancouver team specializing in high-volume data processing. BetaKit reported the deal earlier this year, though terms weren't disclosed.
Mecka offers an iOS app for data capture and web-based tools for browsing egocentric video and sensor data. It maintains what it calls enterprise evaluation and deployment programs for larger customers, according to company materials.
The fresh capital will fund expansion of Mecka's data infrastructure and research operations, as well as build out the on-site robot deployment business. The company is hiring across research, hardware and operations roles in New York and Toronto.
Other startups chasing the robot training data opportunity have also attracted investor interest recently. XDOF is reportedly in funding talks at a $1.2 billion valuation, according to TechCrunch, while Scale AI has been pushing into robotics alongside its core data labeling business. Dealroom's humanoid robotics tracker shows continued momentum across US and Chinese startups in the sector.
Gao told Fortune in June that "useful robots can be deployed today, and immediately." Whether the technology is truly ready for prime time at scale remains an open question, but Sequoia's bet suggests at least one premier venture firm thinks the moment has arrived. "That is why we are excited to back Josh, Jason and their fast-growing team," Biad and Flynn wrote in a blog post announcing the investment.
For now, Mecka occupies a curious position in the robotics stack: neither a hardware manufacturer nor a pure software play, but rather the connective tissue trying to bridge human capability and machine learning. It's a bet that the path to useful robots runs through millions of hours of human motion data, captured one sensor-wearing contributor at a time.
