Niobia, a San Francisco startup, is making significant strides in an often neglected area of battery and chip manufacturing: assisting engineers in diagnosing failures. Their innovative AI software detects production flaws and enables engineers to inquire about the causes via a chatbot, providing rapid insights in minutes instead of days.
The process works as follows: Clients install AI agents along their manufacturing lines to identify defects while involving human engineers to investigate potential failure sources. These agents operate on self-sufficient GPUs attached directly to the manufacturing machinery, functioning similarly to sensors. They are capable of detecting minute imperfections, including tiny scratches. Additionally, Niobia’s AI consolidates all relevant data, allowing engineers to ask a chatbot about any issues and explore deeper insights into their origins.
Swift resolution is vital because chips and batteries are among the most significant global products, currently facing supply constraints. As manufacturers innovate, challenges inevitably arise during the research and development phases. Engineers can spend days or even longer troubleshooting these issues before resuming production. Each defect can incur substantial costs, sometimes reaching millions, and lead to delays in manufacturing schedules.
The co-founders of Niobia, Gaurav Jha and Harveen Kaur, both possess PhDs in chemistry and have prior experience working with chip and battery producers, equipping them with valuable industry insights. In a recent discussion before participating in TechCrunch Disrupt 2026, where Niobia is a finalist, Jha noted that their specialization addresses a multibillion-dollar market, catering to major players in the battery and semiconductor industries.
Jha commented, “It’s a very specific niche; people may not be aware of it.” He continued, “However, behind the scenes, this is essential to everything we use daily… even if it’s not immediately visible.”
“Virtually every item around you, such as glasses, footwear, wallets, and headphones, involves these manufacturing processes,” Jha elaborated.
Niobia’s AI agents utilize open-source models which the company further trains using data from battery and chip research. This training allows them to identify tiny defects effectively. Jha mentioned that Niobia opted for open-source models because their clients are concerned about keeping sensitive company data out of a public AI cloud, where it could potentially be exposed.
“Our clients prefer not to risk their proprietary information in the cloud, even with assurances of its safety,” Jha stated. The startup also emphasizes its commitment to cybersecurity, especially given the competitive landscape in battery and chip innovation. For instance, Niobia allows clients to retain their own encryption keys, ensuring the company cannot decrypt client data.
Jha pointed out that the company aims to concentrate on three main areas: film coating, electroplating (the application of metal to surfaces using electric currents), and fluidized bed reaction (where solid particles are suspended in a gas or liquid).
He also expressed a vision for Niobia’s technology to evolve towards greater autonomy, striving towards the eventual establishment of fully automated factories for their customers.
Those interested in learning more about Niobia can watch their presentation at TechCrunch Disrupt this afternoon in San Francisco. The company is competing against 19 other startups in the prestigious Startup Battlefield competition, taking place from today through Thursday, October 15.
