San Francisco-based startup Niobia is focusing on a key yet often overlooked aspect of battery and chip manufacturing: helping engineers understand how and why problems occur. The company claims that its AI software can identify production defects and allows engineers to delve into the causes through chatbots, with the goal of reducing the time it takes to get answers from days to just minutes.
Its general operation mode is as follows: Customers deploy AI agents on the production line to detect defects, and then human engineers determine the possible reasons for product failures. These agents run on independent GPU devices that are directly installed on manufacturing equipment, similar to sensors. They can identify flaws as small as pinholes, such as scratches. The AI of Niobia also aggregates all the data in one place, making it convenient for engineers to ask chatbots about the issues and further investigate the causes.
Speed is of paramount importance because chips and batteries are among the most crucial products in the world, and they are also becoming increasingly scarce. As manufacturers continue to develop and improve technology, it is almost inevitable that problems will arise during the research and development process. Engineers often have to spend days, or even longer, to find solutions before production can resume. Every defect or mistake can be costly—sometimes costing millions of dollars—and can slow down the manufacturing process.
Niobia Co-founders Gaurav Jha and Harveen Kaur both hold Ph.D. degrees in chemistry and have worked at chip manufacturers and battery manufacturers, which gives them first-hand experience of the industry's challenges. In a call before the launch of TechCrunch Disrupt in 2026, Jha explained that although what Niobia does is highly specialized, it serves a market worth billions of dollars, which includes some of the largest battery and semiconductor manufacturers today.
“This is very niche; perhaps people will never know about it,” Jha said when talking about the battery and chip manufacturing markets that the company serves. “But it’s like what’s behind the scenes, providing power for everything… we just can’t see it in the real world.”
Jha added, "Every product around you, such as glasses, shoes, slippers, wallets, and headphones – all these processes involve every single part of them."
The agents of Niobia run on open-source models, and the company uses battery and chip research, as well as other data, to further train these models. It is this training that enables them to identify defects as small as pinholes. Jha states that the main reason for choosing open-source models is that customers do not want their highly sensitive corporate information to enter public AI clouds, nor do they wish for this data to be used to train models.
“No customer wants their proprietary data to go to the cloud, even though we tell them it’s safe,” says Jha. Considering the high-risk situation of competitive innovation among rivals in the battery and chip industries, this startup has also explained its cybersecurity practices. For example, Niobia allows customers to hold their own encryption keys, so that the company itself cannot decrypt the customers’ data.
Jha indicates that the company's current focus is on three areas of expertise, namely thin film coating, electroplating (which involves depositing metal on a surface using electric current), and fluidized bed reaction (a chemical process in which solid particles are suspended in a gas or liquid).
He said that the company also hopes to make technology more autonomous, with the ultimate goal of creating a fully autonomous factory for customers.
If you want to learn more, Niobia will be making a presentation this afternoon on the stage at TechCrunch Disrupt in San Francisco. Everyone is welcome to come and watch not only this company but also the other 19 startups that are competing alongside them as part of the Startup Battlefield competition, which takes place from today, Tuesday, through Thursday, October 15th.
