New York Startup Midcentury Secured $15 Million Funding
The robotics firm launched a massive egocentric dataset and cloud-based simulation platform to address training bottlenecks.
Updated on Sept. 26, 2026 in Robotics
Live Poll
Should companies be allowed to restrict access to the datasets used to train physical AI robots?
New York-based Midcentury emerged from stealth on September 23, 2026, announcing $15 million in seed funding. The company also released a proprietary egocentric dataset and a cloud simulation environment designed to accelerate robotics development.
Why it matters
The company aims to solve the data bottleneck currently limiting robotics training by providing high-scale human behavioral data. This effort addresses the scaling challenges inherent in physical automation research.
The new egocentric dataset contains over two million hours of data across 50 environments and 20,000 tasks, dwarfing the 3,600 hours provided by the research-grade Ego4D v2. This collection includes 50,000 hours of gameplay and 69,000 hours of conversational voice data.
The players
Midcentury
A New York-based startup focusing on robotics training infrastructure through proprietary datasets and simulation.
Chetan Kulhari
The CEO and leader of Midcentury.
The details
Midcentury uses a cloud simulation platform called Matrix to build digital twins—virtual replicas of physical spaces—using real-world data to test robotics performance. Engineering teams can run parallel tests on GPU clusters, allowing them to automatically convert system test failures into structured training examples for robot learning. The dataset supports this with high-fidelity inputs including 3D hand pose tracking, depth maps, and point tracks.
Timeline
January 2026: The company sold $8.9 million in equity as revealed by an SEC filing.
Q2 2026: The median seed round for firms in the big data sector reached $4.5 million.
September 23, 2026: Midcentury officially launched its platform and dataset.
The Tech Race
Midcentury's entry marks a shift from publicly accessible research benchmarks like Ego4D v2 toward proprietary, commercial-grade infrastructure. The company is positioning itself to lead the data-at-scale race required to move robotics beyond laboratory settings.
For robotics engineers in New York and beyond, the Matrix platform and dataset offer new tools to iterate on autonomous tasks without relying solely on physical hardware testing. The availability of these tools for third-party developers is currently unannounced.
The takeaway
The company's capacity to scale robotics training depends on whether these two million hours of behavioral data translate into reduced development timelines for automation systems. Watch for future benchmarks demonstrating if this dataset successfully lowers error rates in complex real-world tasks.
Further reading
For broader context on the industry's approach to automation, visit Robotics.
Source note: This article includes information reported by Ventureburn.
Live Poll
Should companies be allowed to restrict access to the datasets used to train physical AI robots?
