OpenAI buys Glass Imaging, adding iPhone camera veterans to its hardware push

Why it matters

OpenAI is buying the imaging layer between a camera sensor and a multimodal model. Glass Imaging gives its hardware group proven on-device vision technology and founders who built core iPhone camera features.

OpenAI acquired Glass Imaging, founded by former Apple camera engineers Ziv Attar and Tom Bishop, in recent months in a deal valuing Glass Imaging above $300 million, The Wall Street Journal reported on September 14th.

The deal gives OpenAI a team that spent years extracting better photographs from the constrained lenses and sensors inside smartphones. That expertise fits directly into OpenAI's expanding hardware operation, where cameras could serve as a primary input for devices expected to understand their surroundings rather than wait for typed instructions.

Glass Imaging was founded in 2019 after Attar and Bishop worked on computational photography at Apple. Attar had already built and sold one camera business to the iPhone maker: he founded LinX Imaging in 2011 and ran it until Apple's acquisition in 2015. LinX developed dual-camera systems for depth effects and low-light image fusion. Attar later led Apple computational-photography projects that included Portrait Mode for the iPhone 7 Plus.

Bishop worked at Apple from 2013 through 2018 on core Portrait Mode technology. His earlier research covered inverse imaging, deblurring, super-resolution and depth estimation, the same class of problems Glass Imaging later attacked with neural networks.

Replacing the conventional camera pipeline

Glass Imaging's product, GlassAI, is a neural image signal processor, or neural ISP. Conventional smartphone camera software passes sensor data through a series of separately tuned stages for functions such as noise reduction, sharpening, demosaicing and multi-frame fusion. Each stage can discard information or introduce artifacts.

GlassAI instead processes bursts of RAW sensor data through a network trained for a particular camera module. Glass Imaging says the system can correct lens aberrations, sensor noise, blur and crosstalk while recovering detail contained in the original captures. Glass Imaging has repeatedly distinguished that approach from generative enhancement systems that invent plausible pixels, a consequential difference for any camera feeding visual information into an AI model.

Glass Imaging demonstrated GlassAI on Qualcomm's Snapdragon 8 Elite platform in October 2024. The demonstration processed RAW image bursts on a Qualcomm reference phone using the chip's neural accelerator and compared the output with an iPhone 16 Pro Max. Those comparisons were produced by Glass Imaging and do not constitute an independent benchmark.

By 2026, GlassAI had moved beyond demonstrations. Glass Imaging said its neural ISP was used for zoom imaging in Honor's 600 series smartphones, giving the technology a commercial device deployment before OpenAI bought the business.

That progression matters. OpenAI acquired working edge-computing software and camera integration experience, rather than an early research project whose path into consumer hardware remained theoretical.

OpenAI is assembling the device stack

OpenAI formally merged the io Products team into its organization in July 2025 after working with former Apple design chief Jony Ive and his design firm LoveFrom. OpenAI's account of the io transaction described a group spanning hardware, software, physics, product development and manufacturing. Ive and LoveFrom remained independent while assuming design responsibilities across OpenAI.

OpenAI has since made its vertical-integration strategy explicit. In an August 25th, 2026 strategy post, Chief Financial Officer Sarah Friar described one system spanning data centers, custom chips, models, software products and AI-native devices. An OpenAI manufacturing request issued in January also sought suppliers for consumer-device components including displays, optics, silicon, electronics and final assembly.

Glass Imaging fills a specific gap in that structure. Models that operate through a physical device need a reliable stream of visual data, and compact cameras produce noisy, distorted inputs before any multimodal model begins interpreting a scene. Improving the camera pipeline at the sensor level can give the model cleaner information while reducing the compute required for later correction.

Attar and Bishop also bring experience co-designing optics and software around the limitations of thin devices. That combination is harder to assemble by hiring general computer-vision researchers. Smartphone photography depends on calibration against particular lenses, sensors and processors, followed by the less glamorous work of making the system run within tight power and latency budgets.

The acquisition therefore points toward an OpenAI device designed to see continuously or frequently, although the WSJ report identifies no particular product using GlassAI. Cameras would give an AI system context that a phone or laptop currently receives only when a user deliberately uploads an image or starts a video session.

A deep-tech exit after the first deployment

Glass Imaging announced a $20 million Series A in 2025 led by Insight Partners, with GV, Future Ventures and Abstract Ventures participating. That followed a $9.3 million seed extension backed by GV, Future Ventures, Abstract Ventures and LDV Capital, as well as an earlier seed led by LDV Capital and GroundUP Ventures.

The acquisition valuation is roughly ten times the approximately $30 million Glass Imaging had publicly identified across its later seed and Series A rounds. Deal valuation cannot be treated as investor proceeds because ownership, preferences and consideration were not detailed. It still represents a significant outcome for a specialized imaging business that had only recently placed its technology in a shipping phone.

OpenAI is effectively paying for the years Attar and Bishop spent learning where smartphone cameras fail, first from inside Apple and then by building an alternative pipeline. Glass Imaging began by trying to narrow the gap between phone cameras and professional systems. Inside OpenAI, that same technology can become part of a larger wager: the next AI interface will need to understand what is in front of it, and the quality of that understanding starts before the model sees a single pixel.