Robotics data startup XDOF is in late-stage negotiations to secure a Series B funding round at a valuation of approximately $1.2 billion, led by venture capital firm 8VC. The capital raise follows a rapid commercial acceleration for the firm, which emerged from stealth mode less than three months ago and now reports an annualized revenue run rate approaching $50 million.
The fast-moving fundraising cycle highlights the intensifying market demand for high-quality, real-world physical training datasets required to develop general-purpose robots. While XDOF has not finalized terms or disclosed the exact quantum of capital targeted, the ongoing discussions cement its status as a foundational infrastructure player in the emerging physical artificial intelligence sector.
Rapid Asccent Following Stealth Exit
XDOF was established in 2024 by UC Berkeley researchers Philipp Wu and Fred Shentu, who sought to address a severe shortage of large-scale physical training data in robotics research. The co-founders developed GELLO, a low-cost teleoperation system enabling human operators to control robotic arms remotely and generate structured movement datasets, publishing academic findings that laid the groundwork for the startup.
The venture previously closed a $70 million Series A financing round in June with backing from institutional firms including Thrive Capital, Andreessen Horowitz, Lux, and Spark Capital. Although the company did not initially plan to approach markets for capital so soon, surging commercial interest prompted investors to initiate conversations for the current unicorn-level round.
Bridging the Robotics Data Bottleneck
Unlike large language models that harvested vast textual repositories from the internet, physical robotics lacks an equivalent public data reservoir to train autonomous systems. XDOF operates as an outsourced data supply chain, building collection tools, annotation pipelines, and telemetry systems to feed frontier AI laboratories and robotics developers.
The firm utilizes a hybrid approach combining remote robotic teleoperation with human data collectors wearing body sensors to record routine daily actions such as folding apparel and flattening shipping containers. Through a partnership with UC Berkeley's AI Research lab, the company is preparing to release the ABC database, which it describes as the largest collection of robot training data assembled to date.
Competitive Landscape and Client Base
The startup has previously confirmed active business relationships with roughly 20 customers, encompassing several leading frontier AI laboratories. Market participants and venture investors increasingly compare XDOF's strategic positioning to textual data-labeling giants like Scale AI, viewing physical sensor pipelines as the primary constraint in embodied intelligence.
Competitors operating within the same emerging real-world data collection ecosystem include Mecka AI, alongside established human-data platforms expanding their operational scope beyond text-based models. The intense competition underscores how venture capital is aggressively pricing foundational robotics infrastructure to capture early market share.
Outlook and Unresolved Terms
As negotiations remain fluid, specific deal parameters such as the exact cash injection amount and whether the $1.2 billion valuation incorporates incoming capital remain undisclosed by the parties. Neither XDOF nor lead investor 8VC responded to media requests for official comment regarding the late-stage term sheets.
Whether the startup can sustain its rapid revenue expansion and validate its billion-dollar valuation through scaled commercial deployments will depend on execution over the coming fiscal quarters. Industry observers are closely monitoring the finalization of the Series B round as a key indicator for deep-tech venture pricing.
