Deepnight Inc.

Deepnight Inc., a minority-owned, small disadvantaged business headquartered in San Francisco, California, develops night vision machine learning algorithms, autonomous unmanned aerial systems navigation software, and advanced sensor technologies for defense-related R&D as a prime contractor under full-and-open and total small business set-aside designations. All seven recent federal awards are prime contracts under SBIR and STTR programs, with 100% small business set-aside positioning across the portfolio.

Deepnight's customer base is tightly concentrated on two defense agencies. The Department of the Army accounts for 4 awards and $4.5 million in combined potential value, while the Air Force Research Laboratory accounts for 3 awards and $3.1 million — covering all recent activity. Dollar and volume rankings converge completely, with no tension between them. The company specializes in low-light operational environments and C5ISR (command, control, communications, computers, cyber, intelligence, surveillance, and reconnaissance) applications, positioning itself as a pure defense innovator rather than a diversified contractor.

Recent representative awards include a $2.0 million Army SBIR Phase II contract (August 2025) for a deep night vision machine learning algorithm supporting autonomous drone swarm navigation and sensing, with completion scheduled for August 2027; a $1.25 million Air Force SBIR Phase II contract (August 2024) for neural network-driven digital night vision software for goggle sensor platforms, completing May 2026; a $1.775 million Air Force purchase order (August 2025) for a Night Vision Reconnaissance Augmentation System combining autonomous unmanned aerial systems navigation, digital imaging, and neural networks; and a $1.999 million Army SBIR Phase II direct-to-Phase-II award (April 2026) on topic A254-020, completing April 2028. Deepnight also holds two Army SBIR Phase I contracts at $249.8 thousand each (February and April 2025) for dual bolometric/CMOS passive range-finding neural networks and machine learning algorithms in low-light environments. All awards carry firm fixed-price structures.

The company does not sit on GWACs, BPAs, or agency-wide IDIQs; all work flows directly from SBIR and STTR solicitations. Activity has grown from one award in 2024 to four in 2025 and two in 2026 (to date), with four awards combining to $6.0 million in potential value reaching ultimate completion within the next twenty-four months and therefore facing recompete risk. The portfolio is entirely prime — no subcontract footprint appears in the source. Deepnight's narrow focus on night vision, neural networks, and autonomous systems for defense remains unstandardized at the federal level; the company has not yet secured a seat on a governmentwide vehicle, making each solicitation a discrete competitive event and creating both barrier to entry and vulnerability to competitive displacement on renewal.