Deepnight

What do they actually do

Deepnight sells software that turns very low‑light video from common CMOS sensors into clear, near‑daylight‑looking footage in real time. Their stack runs on phones/edge‑class chips, positioning it as a digital alternative to bulky analog image‑intensifier hardware used in many night‑vision systems today Deepnight site, TechCrunch.

Today they license and demo this software to government research labs, military programs, and hardware partners rather than selling consumer devices. A typical setup uses an off‑the‑shelf CMOS camera feeding Deepnight’s on‑device model, with processed video output to goggles, helmets, camera feeds, or drones. Public demos and integrations include base‑defense trials with Picogrid and validation work with SRI; they’ve also highlighted a smartphone demo that preceded an initial Army contract TechCrunch, Picogrid, SRI.

The company is early‑stage (YC W24) with seed funding and DoD pilots rather than fielded mass‑market products. The near‑term focus is making the software robust for real‑world, on‑device use and moving from pilots to deployed integrations with OEMs and defense programs YC, TechCrunch.

Who are their target customer(s)

• Military units and DoD program managers: Need lighter, cheaper, and more reliable night‑vision than analog goggles so troops can operate longer with better situational awareness; Deepnight is already running Army/Air Force pilots that test this premise TechCrunch.
• Defense OEMs and helmet/goggle manufacturers: Current designs depend on costly intensifier tubes and legacy optics; they want a software approach that uses commodity sensors and accelerates new product cycles TechCrunch, Picogrid.
• Government research labs and test programs: Labs need reproducible ways to evaluate digital night‑vision on off‑the‑shelf sensors without building custom optics each time; Deepnight’s stack has been used in validation and prototypes (e.g., with SRI) SRI, TechCrunch.
• Base and critical‑infrastructure security operators: Nighttime surveillance often yields noisy video that drives false alarms and staffing; they want clearer night feeds that work with existing cameras and edge hardware Picogrid.
• Drone and small‑vehicle system integrators: Small platforms are weight and power constrained and can’t carry heavy night‑vision hardware; they need real‑time low‑light imaging on embedded processors TechCrunch.

How would they acquire their first 10, 50, and 100 customers

• First 10: Run high‑touch demos and quick proofs‑of‑concept with DoD program offices and labs, leveraging existing Army/Air Force contracts and partner intros to land initial pilots TechCrunch.
• First 50: Convert pilots into integrations with defense OEMs and additional programs by offering ready‑to‑install builds/SDKs and running joint field demos that partners can publicize to speed procurement Picogrid, TechCrunch.
• First 100: Expand via OEM and systems‑integrator channels into base security, drones, and maritime; license the stack, provide pre‑qualified builds/certification support, and bundle trials with channel partners to reduce integration cost Picogrid, SRI, TechCrunch.

What is the rough total addressable market

Top-down context:

Global night‑vision devices were about $11.4B in 2024, with roughly 65.8% from military customers (~$7.5B) IMARC. Image‑intensifier tubes—the component Deepnight aims to replace—were about $0.9B in 2024 ReAnIn.

Bottom-up calculation:

As an initial wedge, treating the image‑intensifier spend as the replaceable component yields an immediately addressable defense OEM opportunity of roughly $0.9B annually ReAnIn. The broader defense device spend (~$7.5B) frames the longer‑term ceiling if software‑first digital night vision substitutes across device categories IMARC.

Assumptions:

• OEMs are willing to substitute software + CMOS sensors for intensifier tubes in a meaningful share of new designs.
• Military share and procurement patterns remain broadly similar near‑term to 2024 mix reported by IMARC.
• Adjacent surveillance/thermal markets are excluded from core TAM to avoid double‑counting; they are expansion options requiring separate channels Grand View Research.

Who are some of their notable competitors

• Visionary.ai: AI “software ISP” for low‑light and denoising on edge devices; runs on common SoCs and targets OEM image quality, overlapping with Deepnight’s software approach site, Telecompaper.
• SiOnyx: Supplier of digital low‑light sensors and finished camera modules for helmet, marine, and surveillance; offers sensor+camera hardware bundles that some OEMs may choose instead of software‑only integration site.
• Elbit Systems (incl. L3Harris as similar prime): Incumbent defense suppliers of night‑vision goggles and image‑intensifier tubes; default providers for many military programs that Deepnight must displace or partner with overview, product.
• Teledyne FLIR: Leader in thermal imaging systems used for seeing in total darkness; represents an alternative technology buyers may pick over visible‑light digital approaches FLIR explainer.
• Ubicept: Startup combining novel ultra‑fast/SPAD‑style sensors with software for extreme low‑light, motion‑robust imaging; competes for similar drone/military/OEM pilots from a different hardware+algorithm path coverage.