What Happened

XDOF, a startup that builds real-world teleoperation data pipelines for training general-purpose robots, is in late-stage talks to raise a Series B at a valuation of approximately $1.2 billion, according to TechCrunch. The round is reportedly being led by 8VC.

This comes less than three months after XDOF emerged from stealth in June 2026 with a $70 million Series A backed by Thrive Capital, Andreessen Horowitz, Lux, and Spark Capital. The company was not actively planning to raise again so soon, but VCs approached XDOF after seeing rapid growth — annualized revenue is reportedly approaching $50 million.

The total capital being raised and whether the $1.2 billion valuation is pre- or post-money are not yet known. The terms are not final and could still change. Neither XDOF nor 8VC responded to requests for comment from TechCrunch.

XDOF was co-founded in 2024 by UC Berkeley researchers Philipp Wu (CEO) and Fred Shentu (CTO). Their work originated from a project called GELLO, a low-cost teleoperation system for generating robot training data, which produced an influential robotics paper. The company is now partnering with UC Berkeley's AI Research lab to release what it calls ABC — claimed to be the largest collection of high-quality robot training data ever assembled.

Why It Matters

The robotics industry faces a fundamental data problem that LLMs never had: there is no internet-scale corpus of physical interaction data. Robots can't train on scraped text — they need structured, real-world demonstrations of tasks like folding clothes, flattening boxes, and manipulating objects.

XDOF's approach combines remote robot teleoperation (human operators steering robotic arms remotely) with egocentric data collection (humans wearing body sensors to capture everyday movements). The company plans to hire and train data collector teams worldwide, creating what investors describe as the 'Scale AI or Mercor for physical robotics.'

The fact that a company three months out of stealth is generating ~$50M in annualized revenue from 20 customers — including several frontier AI labs — tells you something important: the robotics data layer is not a future problem. It's a present-day bottleneck that well-funded labs are paying premium rates to solve right now.

This round also fits a broader pattern. In the past two months, Lyte raised $165M at a $1.6B valuation for robot perception silicon, and General Intuition raised at a $6B valuation while pushing into robotics. The capital flowing into robotics infrastructure is accelerating across the stack — data, perception, and models.

Who Is Affected

Robotics startups and frontier AI labs are the direct customers here. XDOF's 20-customer base reportedly includes several frontier AI labs, which means the companies building foundation models for robotics are already outsourcing data collection rather than building it in-house.

AI infrastructure investors should note the velocity: $70M Series A to $1.2B Series B talks in under three months. The robotics data category is compressing the typical funding timeline dramatically.

Competitors in this space include Mecka AI, as well as human-data platforms expanding beyond LLMs — notably Scale AI and Micro1. The competitive landscape is still forming, but XDOF's early traction and Berkeley pedigree give it a meaningful head start.

Strategic Implications

For AI startup founders: If you're building robotics or embodied AI, the data layer is rapidly becoming a buy-not-build decision. XDOF's $50M ARR from 20 customers in under a year suggests labs are paying premium for structured real-world data. Factor this into your build-vs-buy analysis — building proprietary data pipelines may be a distraction from your core model work.

For developers/operators building with AI APIs: Robotics data infrastructure is emerging as a distinct vendor category. If your roadmap includes physical AI, start evaluating data partners now. The field is consolidating fast, and early customers may lock in preferential access to the highest-quality data pipelines.

For non-technical business owners evaluating AI tools: The robotics data supply chain is getting funded at billion-dollar valuations, which means physical AI products will likely reach commercial deployment faster than expected. Watch for XDOF's data partnerships to translate into measurable improvements in robot capabilities from vendors you may be evaluating.

What to Watch Next

Monitor whether XDOF's Series B closes at the reported $1.2B valuation and whether 8VC officially leads. Also watch for the release of the ABC dataset with UC Berkeley — if it's open or partially open, it could shift the competitive dynamics for robotics data startups. Any announcement of new customers or expansion of data collector teams would further validate the outsourced data-supply-chain thesis.

Frequently Asked Questions

Q: What does XDOF do?

A: XDOF collects real-world teleoperation data for training general-purpose robots. The company combines remote robot teleoperation (humans steering robotic arms) with egocentric data collection (humans wearing body sensors to capture everyday tasks) to build training datasets that frontier AI labs and robotics companies use to train their models.

Q: Why is XDOF raising again so soon after its Series A?

A: According to TechCrunch, XDOF wasn't planning to raise again so soon, but VCs approached the company after seeing rapid growth — annualized revenue reportedly approaching $50 million. The company raised a $70M Series A in June 2026 and is now in talks for a Series B at a ~$1.2B valuation just three months later.

Q: Who are XDOF's competitors?

A: Competitors include Mecka AI, as well as human-data platforms expanding beyond LLMs such as Scale AI and Micro1. The robotics data category is still forming, with XDOF positioning itself as the 'Scale AI for physical robotics.'