Bengaluru: Robotics has not yet had its “GPT moment”, but the ecosystem is beginning to take shape around it, with startups increasingly building the underlying tools needed to train and deploy autonomous systems, according to Y Combinator general partner Jon Xu.“Fundamental autonomous systems that can truly make decisions and act – we have not seen that moment come yet. It is still up for grabs,” Xu told TOI on the sidelines of Y Combinator’s first Startup School event in India, on Saturday. Instead, early momentum is emerging in the infrastructure layer around robotics. Xu pointed to a growing number of startups building tools that help robotics companies train systems and create autonomous value.Y Combinator is one of the world’s most influential startup accelerators known for backing companies such as Airbnb, Stripe and Dropbox. “These are still early-stage companies, and a lot of their customers today are other startups. But that is similar to how AWS started, where the first set of customers were startups that later became large companies,” he said, adding that such companies could become foundational to the sector’s growth.The comments come at a time when global attention on artificial intelligence has been concentrated on software and large language models, even as hardware-led innovation remains uneven across geographies. Xu acknowledged that China has made significant strides in deploying robotics, particularly due to its manufacturing depth, but said the core breakthroughs in autonomy are still unresolved.Within India, Xu said there are early signs of technical depth emerging beyond software, particularly in hardware and frontier technologies, though much of it remains under the radar.“I have had conversations with founders building drone devices, and others working on hardware to detect drone threats,” he said. “Young, curious founders are building in these spaces.”While categories such as drones may appear crowded, Xu argued that technically strong teams can still differentiate themselves. “In any space that is perceived to be crowded, people with the right insights and deep technical understanding can stand out,” he said.Beyond robotics, Xu highlighted areas such as model training, reinforcement learning, and efforts to narrow the gap between open-source and proprietary AI models as emerging focus areas. He also pointed to scientific problem-solving as a potential frontier for AI systems.On founder selection, Xu said Y Combinator continues to prioritize individuals over ideas, but cautioned against treating the accelerator as a validation milestone. “I would not look at YC as a signal for whether you should work on something,” he said.Instead, he said the firm looks for depth of insight into users and clarity of thinking. “The intentionality with which you design for a specific customer, those insights don’t lie,” Xu said.The event on Saturday saw over 2,000 early-stage builders and students in technology participate, selected from about 25,000 applicants. Those chosen also received $25,000 in AI and cloud credits.
