The Data Gold Rush: Mecka AI Nears $500M Valuation in Race to Train the Next Generation of Humanoids

In the rapidly evolving landscape of artificial intelligence, the next great bottleneck isn’t compute power or algorithmic sophistication—it is high-quality, physical-world data. Mecka AI, a startup carving out a niche in the robotics data ecosystem, is reportedly in advanced talks to secure a new funding round led by Sequoia Capital. Sources familiar with the deal suggest the financing could value the fledgling company at approximately $500 million, a staggering figure for a firm that only emerged on the scene in 2024.

This potential capital injection arrives a mere three months after Mecka announced a $60 million raise led by Framework Ventures, with participation from marquee names like Menlo Ventures, SV Angel, and Kindred Ventures. While the precise size of the new round remains undisclosed and terms are subject to final negotiation, the interest from a top-tier firm like Sequoia underscores the white-hot demand for the specific type of data Mecka provides: the nuanced, complex, and messy reality of human movement.

A Strategic Shift: From Fintech to Robotics

The ascent of Mecka AI is a testament to the versatility of the current founder-led startup wave. The company was co-founded by a quartet of entrepreneurs: Josh Gao, Mogen Cheng, Jason Chong, and Duy Nguyen. Interestingly, the founders arrived at the robotics industry without traditional backgrounds in mechanical engineering or robotics research.

Gao and Cheng previously navigated the competitive waters of the restaurant fintech sector, while Chong brought experience from the high-stakes world of cryptocurrency, having joined Coinbase following the acquisition of his own crypto exchange. Nguyen, the operations lead, rounds out the team. Despite their non-traditional entry, the group identified a critical friction point: while large language models (LLMs) have been successfully fed on the internet’s vast repository of text, physical robots lack a similar "internet" of human motion data.

The name "Mecka" is a nod to the "mecha" genre of science fiction, which features giant robots operated by human pilots. By focusing on the human-to-machine interface, the founders aim to do for robotics what Scale AI did for LLMs—provide the fuel that allows artificial intelligence to perceive and navigate the physical world.

Chronology of Rapid Growth

The timeline of Mecka’s expansion has been nothing short of aggressive:

• Early 2024: Mecka AI is founded by Josh Gao, Mogen Cheng, Jason Chong, and Duy Nguyen. The company pivots toward "egocentric" data collection—recording human actions from a first-person perspective.
• June 2026: Mecka announces a $60 million fundraise. During this announcement, CEO Josh Gao projects an annual run rate of $100 million by the end of 2026, signaling immense confidence in the market demand for their datasets.
• September 2026: Reports emerge that the company is in late-stage talks for a new round led by Sequoia Capital, potentially pushing the valuation to the half-billion-dollar mark.

This rapid-fire progression from inception to unicorn-adjacent valuation in less than three years reflects a broader trend in the venture capital market: investors are no longer waiting for long-term product maturity if a startup can prove it holds a "data moat" in a mission-critical sector.

The Mechanics of "Egocentric" Data

To understand why Mecka is commanding such a valuation, one must understand its methodology. Training a humanoid robot to perform a task—such as folding laundry, fixing a car engine, or brewing a cup of coffee—requires more than just static images. It requires "embodied intelligence," which is best learned from watching humans perform these tasks.

Mecka pays individuals to wear body sensors and carry smartphones to record their daily activities from a first-person perspective. This is known as "egocentric" data. By capturing the minute adjustments, micro-movements, and sensory cues that humans use to manipulate objects, Mecka creates a library of motion that robotics labs can use to train neural networks.

This approach bypasses the limitations of teleoperation—the practice of a human remotely controlling a robot to "teach" it—which is slow and expensive. By crowdsourcing human movement at scale, Mecka is effectively creating a massive, diverse training set that allows humanoid robots to generalize their skills across different environments and tasks.

Official Responses and Market Silence

As is often the case with high-stakes financial maneuvers, the parties involved are exercising caution regarding public disclosure. Mecka AI did not respond to requests for comment regarding the rumored valuation or the specifics of the Sequoia-led round. Sequoia Capital, maintaining its traditional stance on pending deals, declined to provide a comment.

The silence is standard procedure in the venture world, yet the lack of denial from either party is frequently interpreted by market observers as a strong signal that the deal is indeed moving toward completion.

Implications: The Data Bottleneck

The implications of Mecka’s success—and the broader trend it represents—are profound for the future of automation. We are currently witnessing a "gold rush" for physical-world data.

The Competition

Mecka is not alone in this race. The market for robotics data is heating up rapidly. XDOF, another player in the space, was recently reported to be in talks for a round that would value it at $1.2 billion, despite being only three months out of stealth mode. Meanwhile, established giants like Scale AI, which built its reputation on text and image data, are aggressively expanding into the physical robotics domain.

The Impact on Robotics Development

For robotics startups, the availability of high-quality, ready-to-use motion data could be the difference between a project that stalls in the lab and one that reaches commercial viability. If startups like Mecka can successfully commoditize human motion, it lowers the barrier to entry for building general-purpose humanoid robots.

However, this reliance on human-provided data also raises questions about ethics and standardization. How does one ensure the data is representative? Is the compensation model for the human "data providers" sustainable? And perhaps most importantly, will this data actually translate into reliable, safe performance in the real world?

The "Robot Winter" vs. The "Robot Boom"

For years, the robotics industry struggled with the "last mile" problem—the inability of robots to function in unstructured environments like homes or messy workshops. By utilizing large-scale human data, the industry hopes to finally overcome this hurdle. If Mecka hits its $100 million revenue run rate target, it will serve as empirical proof that there is a massive commercial appetite for the data required to teach robots how to be human-like.

Looking Ahead: A New Era of Embodiment

The valuation of Mecka AI, alongside its peers, suggests that investors believe we are on the precipice of a shift from "digital AI" to "physical AI." While the last decade was defined by chatbots and image generators, the next decade is likely to be defined by embodied systems that can perform work in the physical world.

As Sequoia and other firms pour hundreds of millions of dollars into the "data layer" of robotics, they are betting that the startup which controls the training data will ultimately control the development of the hardware. For the four founders of Mecka, the journey from restaurant fintech to the center of the robotics revolution is an unlikely but logical progression: they saw the bottleneck, understood the value of the data, and built a platform to bridge the gap between human capability and machine execution.

Whether Mecka achieves the $500 million valuation or scales beyond it, the company’s trajectory is a clear indicator that the race to build the first truly autonomous, general-purpose humanoid robot is moving from the realm of science fiction into the realm of high-stakes industrial strategy. The era of the "mecha" may still be in its infancy, but the foundation of its brain—the data—is being laid today, one coffee-making recording at a time.