For a robot, a self-driving car, or an industrial inspection system, the most dangerous moment is often the one it cannot see. Standard cameras, the workhorses of modern computer vision, rely on capturing enough light to form a coherent image. In darkness, in blinding glare, or during rapid motion, that image falls apart, leaving the machine blind. Ubicept, a Boston-based startup spun out of MIT and the University of Wisconsin-Madison, is betting its future on a different kind of light: counting photons, one at a time.
Founded in 2021, the company has raised a total of $9 million to commercialize a perception system built around single-photon avalanche diode (SPAD) sensors and a proprietary software stack. The core proposition is not just a better camera, but a different architecture for machine vision, designed from the ground up for the environments where safety-critical decisions happen [TechCrunch, March 2023].
The Physics-Based Wedge
Ubicept’s technology stack, which it calls the Ubicept Toolkit, is its primary product. It is not a finished consumer device but a flexible platform for researchers and product teams in robotics, automotive, and industrial automation [Ubicept, retrieved 2026]. The system pairs with prototype cameras built around compact SPAD sensors from technology partners, though it also supports conventional CMOS sensors [Ubicept, retrieved 2026].
The technical wedge is its approach to light capture. Where a traditional camera integrates light over a fixed exposure time to produce a frame, SPAD sensors detect the arrival time of individual photons. This raw stream of data is then processed in real-time by Ubicept’s algorithms to reconstruct a high-quality image or, more importantly, to feed directly into perception tasks like object detection and navigation [EE News Europe, August 2023]. This method is inherently more robust in low-light conditions and can handle extreme dynamic range and fast motion without the motion blur that plagues frame-based systems.
From Academic Lab to CES Stage
The founding team emerged directly from the academic groups pioneering this field. CEO Sebastian Bauer completed postdoctoral research at the University of Wisconsin-Madison, working on lidar systems, while CTO Tristan Swedish spent eight years as a research assistant and Ph.D. student at the MIT Media Lab [TechCrunch, March 2023]. MIT Media Lab professor Ramesh Raskar, head of the Camera Culture research group, is a founding advisor and co-founder [LinkedIn, April 2024] [MIT Media Lab, retrieved 2026]. The broader technical team includes veterans from Google, Facebook, and robotics companies like Skydio and Optimus Ride [FinTech Futures, January 2025].
Traction has been demonstrated through public showcases, a key signal for a deep-tech hardware company seeking design wins. At CES 2025, Ubicept demonstrated autonomous vehicle navigation in complete darkness and robots operating in low light [The Outpost, January 2025]. It returned to CES in 2026 with live demonstrations of its toolkit [Financial Content, retrieved 2026]. The company has reported a growing customer base including brands in automotive and AR/VR, though specific names are not public [The Outpost, January 2025].
Funding and the Path to Production
Ubicept has secured its runway through a combination of traditional venture capital and strategic awards. A breakdown of its disclosed funding illustrates the progression.
July 2023 Seed | 8 | M USD
April 2024 Award | 1 | M USD
The $8 million seed round in July 2023 was led by Ubiquity Ventures and E14 Fund, with participation from the Wisconsin Alumni Research Foundation and Phoenix Venture Partners [Businesswire, July 2023]. In April 2024, the company won an additional $1 million investment plus $350,000 in Microsoft Azure credits through the TitletownTech Startup Draft [ibmadison.com, April 2024] [LinkedIn, April 2024]. CEO Sebastian Bauer noted in a 2025 interview that the company had evaluation contracts and was in discussions to convert them to production contracts [Pulse 2, 2025].
The Integration Challenge
The most credible risk for Ubicept is not the science, which is peer-reviewed and demonstrated, but the commercial motion of integrating a novel sensing modality into established, cost-sensitive supply chains. The automotive and robotics industries have spent decades optimizing for CMOS-based vision. Adopting SPAD sensors and a new processing stack requires changes to hardware design, software pipelines, and validation processes.
The company’s answer appears to be flexibility and a focus on the software layer. By supporting both SPAD and CMOS sensors, the Ubicept Toolkit allows customers to experiment and integrate its perception algorithms without a full hardware overhaul from day one [Ubicept, retrieved 2026]. Its model seems to be landing initial design wins in high-value, performance-critical applications,like specialty vehicles or industrial robots where a failure in low light is unacceptable,before pushing for broader adoption.
What a New Standard of Sight Enables
For the engineers building the next generation of autonomous systems, the standard of care today is a patchwork. It often involves layering sensors,cameras, radar, lidar,and using complex fusion algorithms to compensate for each one's weaknesses in challenging conditions. This adds cost, complexity, and points of failure. The promise of a single-photon perception system is a more fundamental solution: a vision system that works reliably in the conditions where others falter.
The patient population, so to speak, is every machine that must see to operate safely. This includes autonomous mobile robots in dimly lit warehouses, last-mile delivery vehicles operating at dawn or dusk, and industrial inspection systems monitoring fast-moving assembly lines. For these applications, Ubicept is not selling incremental improvement but foundational capability. The next twelve months will be critical for translating its CES demonstrations and evaluation contracts into announced production partnerships, proving that its photon-counting vision can move from the lab bench to the factory floor and the open road.
Sources
- [TechCrunch, March 2023] Everything you know about computer vision may soon be wrong. | https://techcrunch.com/2023/03/10/ubicept-computer-vision/
- [Businesswire, July 2023] Ubicept Raises $8M to Unlock Computer Vision in All Lighting Conditions by Counting Individual Photons | https://www.businesswire.com/news/home/20230725606397/en/Ubicept-Raises-$8M-to-Unlock-Computer-Vision-in-All-Lighting-Conditions-by-Counting-Individual-Photons
- [EE News Europe, August 2023] Ubicept raises US$8 million for SPAD-based imaging | https://www.eenewseurope.com/en/ubicept-raises-us8-million-for-spad-based-imaging/
- [The Outpost, January 2025] Ubicept Unveils Groundbreaking Computer Vision Technology at CES 2025. | https://www.theoutpost.ai/news-story/ubicept-unveils-groundbreaking-computer-vision-technology-at-ces-2025-10346/
- [FinTech Futures, January 2025] Computer Vision Pioneer Ubicept to Showcase Breakthrough in Machine Perception at CES 2025 | https://www.fintechfutures.com/press-releases/computer-vision-pioneer-ubicept-to-showcase-breakthrough-in-machine-perception-at-ces-2025
- [ibmadison.com, April 2024] A sharper image | https://ibmadison.com/a-sharper-image/
- [LinkedIn, April 2024] Ramesh Raskar congratulates Ubicept on TitletownTech award | https://www.linkedin.com/posts/raskar_titletowntech-makes-27-million-investment-activity-7321953050456190976-6rN_
- [Pulse 2, 2025] Ubicept Profile: Sebastian Bauer Interview | https://pulse2.com/ubicept-profile-sebastian-bauer-interview/amp/
- [Ubicept, retrieved 2026] Unlock Single-Photon Perception | Ubicept | https://www.ubicept.com/
- [Financial Content, retrieved 2026] Ubicept Announces Live Demonstrations at CES 2026 | https://www.businesswire.com/news/home/20230725606397/en/Ubicept-Raises-$8M-to-Unlock-Computer-Vision-in-All-Lighting-Conditions-by-Counting-Individual-Photons
- [MIT Media Lab, retrieved 2026] Ramesh Raskar profile | https://www.media.mit.edu/people/raskar/overview/