Deep Knight co-founders Lucas Young and Thomas Lee have been friends since childhood. Both worked as software engineers at Google. Digital Night Vision Tech when Young decides to crack the code on an issue that has been wreaking havoc on the US military for decades.
Most night vision technologies are still analog. The goggles use photo lenses and chemical processes to convert nightly scarce light into images, Young told TechCrunch. It also costs between $13,000 and $30,000 from military contractors such as L3Harris and Elbit America.
For years, the US Army has tried to digitalize high-tech, but focuses primarily on hardware. A case in point: A $22 billion budget for the integrated visual augmentation system (IVAS) project, which Anduril took over from Microsoft and its Hololens Tech.
Young, who holds a Calpoly calculation photography degree, worked for five years in smartphone camera software. He wrote code to offset the small opening limit used on smartphones, a cheap $50 digital camera. And Li's background is AI technology, especially computer vision.
One day, Young read a scientific paper from 2018 on The Dark In The Dark, co-authored by the well-known scientist Vladlen Koltun, who is currently at Apple. We discussed using AI for low-light imaging, but at the time it wasn't fast enough to support the 90 frames (FPS) needed for real-time viewing.
In 2024, Young realized that AI Accelerators (SOCs) running on the system on the chip have made enough progress to support 90 FPS. He told his friend Li to quit their job and set up a startup called Deepnight. And they quickly entered the winter cohort of Y-combinators.
The photo of the founder of Deepnight is taken at night with a regular camera on the left and the AI model generates on the right: Image credit: Deepnightimage credit: Deepnight
Their smartphone app surprises the army
The military was their first obvious clientele, but they couldn't just roll over to the Pentagon and book a meeting. So Young found an industry event attended by people from the US Army's Night Vision Institute.
He wrote a white paper outlining his ideas as a software issue: Night Vision. He handed out copies at the event. This included an Army Colonel who agreed to read the paper. “It was just a hallway conversation. I didn't even have business outfits, Young recalls.
The Colonel liked what he read enough to get in touch with people from the lab, officially known as the US Army C5ISR Center.
The founders are eager to show their people that their concept works. I put my smartphone in a VR set that holds my smartphone.
It was a rudimentary prototype and was impressive enough to lead to their first sale.
“The Army awarded Y Combinator a $100,000 contract for a month in February 2024 based on smartphone demonstrations and white papers and presentations,” Young said.
Young and Lee had to present progress in a more formal demonstration. The pair flew to Washington, DC, where ten people showed a room packed with software mechanics and cutting-edge goggle mechanics. (This is a YouTube video where they demonstrate their tech.)
The meeting led to more contracts. A year after its launch, the startup booked around $4.6 million contracts from the federal government, including the US Army and Air Force, as well as companies such as Sionyx and SRI International.
Deepnight also immediately attracted investors. By the end of YC, we had collected a $5.5 million round led by initialized capital. They have an angel like Brian Singh, a former IN-Q-TEL partner, and Matthew Bellamy, the lead singer of the band muse. The Y Combinator was also tipped for standard trading.
Perhaps more than anything, Kortun, the scientist who wrote the papers that inspired the company, has also become an angel investor.
Deepnight offers hardware manufacturers, software and partners, including Goggle manufacturers, military helmets and other products.
“Now we can see everything in the world in the dark, because it's just a software program. It's maritime, like cars, security, drones, boats, electronics, navigation cameras,” explains Young. And because it all relies on ready-made $50 smartphone cameras, their technology doesn't require expensive, bespoke hardware.