Night vision has remained largely unchanged for decades, relying on expensive analog hardware that costs upwards of $30,000 per unit and requires frequent replacement. The technology that helps our military, law enforcement, and first responders see in the dark is not only cost-prohibitive but also delivers suboptimal performance that hasn’t kept pace with modern digital imaging advances.

Lucas Young, cofounder & CEO, and Thomas Li, cofounder.

When we first met Lucas Young and Thomas Li, we were immediately struck by their radically different approach to solving this problem. Rather than trying to incrementally improve existing analog night vision technology, Lucas and his team at Deepnight are completely reimagining it from the ground up by combining AI and commodity digital sensors. Their breakthrough allows standard CMOS sensors – the same type found in smartphone cameras – to see clearly in near-total darkness when combined with their proprietary AI image processing.

MODERN TECH THAT ALSO REDUCES COSTS BY 90%

The night vision devices market has been dominated by a handful of defense contractors using decades-old legacy technology. But the fundamental architecture is shifting from analog to digital, creating an opening for innovative approaches. What’s particularly exciting about Deepnight’s solution is that it achieves superior performance to traditional night vision while reducing costs by over 90% through the use of mass-manufactured components.

AN EXCEPTIONAL TEAM

The technical credentials of this founding team are exceptional. Lucas previously worked on computational photography efforts at Google Pixel. Their Chief Scientist, Dr. Anurag Ranjan, authored seminal work on optimizing AI models for mobile devices while at Apple. The team has attracted other renowned advisors and investors Vladlen Koltun, whose paper “Learning to See in the Dark” helped establish the foundations of AI-based low-light imaging. 

What truly sets Deepnight apart is their proven ability to execute. In less than a year, they’ve secured millions of dollars in contracts with the Air Force, Army, and leading defense technology companies like Anduril along with SRI, which contributed to the invention of the CMOS sensor in the 1960s. Their system has demonstrated superior imaging quality compared to $13,000 military-grade night vision goggles at a fraction of the cost.

WHY WE INVEST IN COMPUTER VISION

The implications extend far beyond defense applications. There are over 40 billion CMOS cameras deployed globally today – in everything from security systems to autonomous vehicles to smartphones – and none of them work well in darkness. One of our longstanding theses at Initialized is investing in cameras that can see and understand the world and through our prior history with companies like Flock Safety and Pano, we know that night vision is essential to both solving crime at night and catching wildfires when they often start in darkness. Deepnight’s technology has the potential to fundamentally change how we operate at night, improving safety and capabilities across sectors like emergency response, infrastructure security, and autonomous systems.

DEEPNIGHT IS HIRING

We’re thrilled to partner with Lucas, Thomas, and the Deepnight team as they build the foundational technology layer that will help the world see in the dark. If you’re passionate about pushing the boundaries of computer vision and AI to solve crucial real-world problems, they’re hiring for key roles in computational imaging research.

The timing for this breakthrough couldn’t be better. As the world becomes increasingly instrumented with cameras and sensors, the ability to operate effectively in low-light conditions becomes critical for applications ranging from autonomous vehicles to disaster response. Deepnight is poised to become the technology layer that finally allows our digital eyes to see clearly, no matter how dark it gets.