YC Grad Deepnight Raises $5.5M for AI Night Vision

Deepnight: Revolutionizing Night Vision with AI

Lucas Young and Thomas Li, the co-founders of Deepnight, share a long-standing friendship originating in their childhood. Both individuals previously held positions as software engineers at Google before Young initiated efforts to resolve a long-standing challenge faced by the U.S. military: the development of advanced digital night vision technology.

The Limitations of Current Technology

The majority of existing night vision systems still rely on analog technology. These goggles utilize optical lenses and a chemical process to amplify available light into visible images, as Young explained to TechCrunch. Consequently, these systems carry a substantial cost, ranging from $13,000 to $30,000 per unit, when procured from defense contractors such as L3Harris and Elbit America.

For many years, the U.S. Army has been actively pursuing the digitization of this technology, with a primary focus on hardware improvements. The $22 billion allocated to the Integrated Visual Augmentation System (IVAS) project serves as a prime example, recently transitioning from Microsoft’s HoloLens technology to oversight by Anduril.

Leveraging Expertise in Computational Photography and AI

Young’s academic background includes a degree in computational photography from Cal Poly. He dedicated five years to refining smartphone camera software, developing code to overcome the inherent limitations of the small apertures and inexpensive $50 digital cameras commonly found in mobile devices.

Li, on the other hand, possesses extensive expertise in AI technology, specifically within the field of computer vision. This complementary skillset proved crucial for the formation of Deepnight.

The Inspiration from "Learning to See in the Dark"

A pivotal moment occurred when Young encountered a 2018 scientific publication titled Learning to See in the Dark. Co-authored by Vladlen Koltun, a prominent scientist now at Apple, the paper explored the application of AI to low-light imaging. However, at that time, the processing power of on-device AI chips was insufficient to achieve the required 90 frames per second (fps) for real-time visualization.

The Rise of Deepnight

By 2024, Young determined that advancements in AI accelerators integrated into system on chips (SoCs) had reached a level capable of supporting the necessary 90 fps performance. He successfully convinced Li to leave their positions at Google and co-found Deepnight. The startup was quickly accepted into the Y Combinator winter program.

Deepnight aims to disrupt the multi-billion dollar night vision industry by offering a software-based solution. Their approach leverages recent advancements in AI and hardware to provide a more affordable and effective alternative to traditional analog systems.

The company’s success in securing funding and acceptance into Y Combinator highlights the potential of their innovative technology.

Innovative App Captures Military Attention

Initially targeting the military sector, accessing key decision-makers proved challenging. Direct engagement with the Pentagon wasn't immediately feasible, prompting a strategic approach by the founders.

Young identified a relevant industry gathering where personnel from the U.S. Army’s night vision laboratory would be present. He leveraged this opportunity to disseminate his core concept.

A detailed white paper was prepared, framing night vision enhancement as a solvable software challenge. This document was distributed at the event, ultimately reaching an army colonel who agreed to review it.

The initial interaction was informal – a simple conversation in a hallway. Young emphasizes he wasn’t formally dressed, wearing just a T-shirt at the time.

The colonel’s positive assessment of the white paper facilitated an introduction to researchers at the US Army C5ISR Center, the laboratory itself.

Driven to validate their concept, the founders rapidly developed a night vision application for smartphones. This was integrated with a virtual reality headset to simulate the user experience.

This rudimentary, yet functional, prototype proved compelling enough to secure their first contract.

“In February 2024, just one month after joining Y Combinator, we were awarded a $100,000 contract by the army, based on the smartphone demonstration, our written materials, and our presentation,” Young stated.

Further demonstration of their progress was required, necessitating a formal presentation. Young and Li traveled to Washington, D.C., to showcase their software’s capabilities alongside existing, state-of-the-art goggles.

This presentation resulted in additional contracts. Within a year of its launch, the startup had secured approximately $4.6 million in contracts from various federal entities, including the U.S. Army and Air Force, as well as collaborations with companies like Sionyx and SRI International.

Deepnight’s potential also attracted significant investor interest. By the conclusion of the Y Combinator program, they successfully raised a $5.5 million funding round, led by Initialized Capital, and included angel investments from Kulveer Taggar, Brian Shin, and Matthew Bellamy.

Notably, the scientist whose original research inspired the company, Koltun, also joined as an angel investor.

Deepnight’s business model centers on providing software solutions and partnering with hardware manufacturers, such as goggle producers and military helmet developers.

“Our technology allows virtually any device to gain night vision capabilities, as it’s fundamentally a software application,” Young explained. “This extends to automotive, security, drones, maritime applications, and various electronic devices.”

A key advantage of their approach is its cost-effectiveness, relying on readily available, $50 smartphone cameras rather than requiring specialized, expensive hardware.