Mecka AI Hits $500M Valuation as Sequoia Backs Robot Training Data Play
TL;DR
Mecka AI is raising Series B at ~$500M valuation led by Sequoia Capital, just three months after closing a $60M Series A. The startup captures egocentric motion data to train humanoid robots, positioning itself as “Scale AI for robotics” in a market racing to solve the physical-world data bottleneck.
The Core Play: Physical Data as the Limiting Factor
Mecka AI’s valuation sprint—from $60M Series A to $500M Series B in 90 days—reflects market conviction that real-world training data is now the primary constraint for deployed robotics. While LLMs saturated on text datasets, humanoid robots face an inverse problem: insufficient egocentric footage of humans performing everyday tasks.
The startup’s model is straightforward. It pays crowdsourced workers to record themselves via body sensors and smartphones performing activities like coffee-making or automotive repair. This data feeds directly into robotics labs and AI companies building general-purpose embodied systems. By June 2026, Mecka was projecting $100M annual run rate for year-end—a metric that likely accelerated the Sequoia extension.
Background: The Robotics Data Supply Chain
Founders Without Robotics DNA
Mecka’s co-founders—Josh Gao, Mogen Cheng (both ex-restaurant fintech), and Jason Chong (ex-Coinbase, crypto background)—arrived at the problem through operational bottleneck analysis rather than domain expertise. Their insight was simple: roboticists need what data engineers already built for LLMs—a systematic pipeline for labeled, real-world information.
Competitive Landscape Heating Up
According to TechCrunch’s reporting, XDOF is nearing $1.2B valuation for similar data collection. Scale AI and Micro1 are expanding beyond LLMs into physical data. The market recognizes that robotics scaling hinges on data infrastructure, not just model architecture.
The “Mecha” Concept
The name deliberately evokes giant robots controlled by humans—a metaphor for Mecka’s actual product. The company is building the nervous system for teleoperated and autonomous systems by systematizing how humans transfer motor knowledge at scale.
Investment Implications: Sequoia’s Robotics Bet
Sequoia’s lead signals confidence that data infrastructure plays command higher multiples in robotics than in AI. While foundation model valuations have plateaued, training data providers are compressing timelines between rounds and justifying unicorn-track metrics. Mecka went from $60M to $500M on a single metric: annual run rate.
The compressed timeline also suggests round terms are still fluid—deal documentation may shift before close. However, the valuation trajectory implies Sequoia sees Mecka as foundational to the robotics supply chain, similar to how scale AI became critical infrastructure for LLM training.
What’s Missing From the Story
- Round size unknown—Mecka hasn’t disclosed capital being raised
- Customer list undisclosed, limiting validation of demand signals
- Unit economics opaque; unclear if $100M ARR is profitable or cash-consumptive
- No competitive moat analysis; other data providers can copy the crowdsourced model
The Broader Context: Data as the New Chip
Mecka’s valuation mirrors a sector-wide trend: specialized training data is becoming more valuable than generalist model code. Robotics companies need embodied, task-specific datasets. Mecka is betting it can become the exclusive or primary supplier before competitors scale. At $500M, Sequoia is essentially pre-acquiring that infrastructure position.
The real test arrives at Series C. If Mecka can maintain $100M+ ARR growth and expand into multi-modal data (vision, touch, force), it could justify unicorn status. If competitors commoditize egocentric capture, valuation momentum evaporates fast.
