XDOF, just three months out of stealth, is in talks for a Series B at a $1.2B valuation

Less than 3 months aft emerging from stealth, XDOF, a startup that collects real-world teleoperation information for grooming general-purpose robots, is successful late-stage talks to rise a Series B astatine a valuation of astir $1.2 cardinal valuation led by 8VC, respective radical with cognition of the woody said.

XDOF was co-founded by UC Berkeley researchers Philipp Wu (CEO) and Fred Shentu (CTO) successful 2024. TechCrunch reported connected the startup’s $70 cardinal Series A successful June, with information from Thrive Capital, Andreessen Horowitz, Lux, and Spark Capital. XDOF wasn’t readying to rise again truthful soon aft that round. But the company’s accelerated maturation — with annualized gross approaching $50 cardinal — prompted VCs to attack it astir a caller round, the radical said.

TechCrunch was incapable to larn the full superior being raised oregon whether the valuation includes the caller funding. The presumption of the woody are not last and could inactive change.

XDOF and 8VC didn’t respond to our petition for comment.

The startup aims to physique the information pipelines, postulation tools, and annotation systems that frontier AI labs and robotics companies can’t easy physique themselves, fundamentally acting arsenic an outsourced data-supply concatenation for the robotics industry.

As a PhD student, Wu was studying however robots larn from ample datasets. One large impediment to his probe was the deficiency of “large-scale information to enactment with,” helium told TechCrunch successful June.

So helium teamed up with Shentu connected a task called GELLO, a low-cost teleoperation strategy that allows a quality relation to power a robotic limb remotely successful bid to make grooming data. Their enactment led to an influential insubstantial successful robotics.

That probe formed the instauration for XDOF, which investors present picture arsenic the Scale AI oregon Mercor for carnal robotics, a notation to the data-labeling giants that helped substance the AI boom. Unlike LLMs, which initially trained connected the entirety of the internet, carnal robots don’t person an equivalent real-world dataset to gully from, making information postulation a captious bottleneck to gathering general-purpose machines.

XDOF is partnering with UC Berkeley’s AI Research laboratory to merchandise what it believes is the largest postulation of high-quality robot grooming information ever assembled, dubbed ABC. 

To seizure this data, XDOF combines distant robot teleoperation with quality collectors who deterioration sensors to grounds mundane tasks similar folding apparel and flattening boxes.

The startup plans to prosecute and bid teams of information collectors worldwide, including teleoperators who steer robots remotely and egocentric operators who deterioration assemblage sensors to seizure question data.  

XDOF antecedently told TechCrunch that it is already moving with 20 customers, including respective frontier AI labs.

Other startups attempting to cod real-world information for robot grooming see Mecka AI, arsenic good arsenic human-data platforms expanding beyond LLMs, specified arsenic Scale AI and Micro1.