Astrabeam's Sub-THz Radar Aims to See Through the Smoke for Autonomous Robots

For an autonomous robot navigating a warehouse fire or a drone inspecting a smoldering industrial site, the most critical sensor is the one that doesn't quit when the air turns opaque. This is the clinical problem Astrabeam is trying to solve, not with another iteration on cameras or lasers, but by pushing radar into a frequency band where it can finally see in detail. The Tarrytown-based startup is developing millimeter-wave and sub-terahertz imaging radar systems, a hardware-and-software stack designed to give machines a reliable sense of their surroundings in smoke, fog, dust, and darkness [Astrabeam site].

It is a bet on a fundamental shift in robotic perception. The standard toolkit for autonomy,cameras, LiDAR, and conventional radar,each has a failure mode in degraded visual environments. Astrabeam's proposition is that by moving to higher frequencies in the sub-terahertz range, its radar can achieve the resolution needed for tasks like real-time object detection and SLAM (simultaneous localization and mapping), all while remaining immune to the conditions that blind optical sensors [F6S]. The company calls it a full-stack perception platform, combining custom-designed integrated circuits with AI-driven sensor fusion software [Astrabeam site].

The hardware wedge in a software-heavy field

Astrabeam's differentiation is rooted in its status as a fabless semiconductor designer. While many autonomy startups focus on layering AI atop commodity sensors, Astrabeam is building the sensor itself. Its core technology is a sub-terahertz imaging radar built with mass-producible silicon, which the company claims offers ultra-low size, weight, power, and cost (SWaP-C) compared to existing high-end LiDAR and imaging radar systems [F6S]. This is not merely an academic exercise. The target is integration into the constrained physical and economic budgets of commercial robots and drones, where every gram and watt matters.

The technical leadership points to this focus. Founder and CEO Kevin Gu has a background in millimeter-wave and sub-THz integrated circuit design, with prior research work at IBM on frontend technologies for 5G radios [Astrabeam About]. Co-founder Devin Reimer brings a different kind of product-build experience from his time as CEO and CTO of Owlchemy Labs, the VR studio behind Job Simulator that was acquired by Google [LinkedIn, 2026]. It is an unusual pairing,a deep-tech RF engineer and a creative software studio head,that suggests a company built to bridge the gap between advanced hardware and deployable product.

Navigating a capital-intensive path

Astrabeam's journey reflects the patient, often non-dilutive funding path common to hardware-focused deep tech. The company has raised an estimated $3 million to date, with backing from firms like Anorak Ventures, Boost VC, and The Venture Reality Fund [PitchBook]. Perhaps more telling is its participation in multiple accelerator programs, including NYC Innovation Hot Spot, Plug and Play Tech Center, and Westchester County's Element 46 Tech Accelerator, which provided non-dilutive support and R&D grants [NYC Innovation Hot Spot, May 2023] [Element 46, 2026]. This staged, grant-augmented approach can conserve equity but also signals the long development cycles and significant capital required to move from chip design to volume production.

The company's early-stage position is clear in its public footprint. There are no named customer deployments or detailed case studies yet. The beachhead market, however, is explicitly defined: manufacturers of autonomous ground robots and drones, followed by potential applications in space vehicles and industrial safety monitoring [A3/Automate.org]. The competitive landscape includes other players in advanced radar, such as Lunewave and Teradar, though Astrabeam stakes its claim on the specific combination of sub-THz frequency, silicon-based design, and integrated AI processing.

What could slow the signal

The risks here are inherent to the category. Building a new class of sensor is a multi-year endeavor with formidable technical and commercial hurdles.

• The integration challenge. Success depends on robot OEMs choosing to design in a novel sensor. This requires not just superior performance, but smooth software integration and demonstrable reliability that meets industrial standards.
• The scaling cliff. Moving from prototype chips to volume manufacturing at a competitive cost is a classic valley of death for fabless semiconductor startups. Astrabeam's mass-producible silicon thesis must be proven on a production line.
• The algorithmic burden. The promise of "AI-driven sensor fusion" means the company must also develop and maintain sophisticated perception software,a significant ongoing R&D commitment beyond the hardware itself.

The company's reliance on accelerator funding and government grants, while prudent, underscores that the venture-scale capital needed for a full commercial ramp may still be ahead. The next twelve months will be critical for transitioning from promising prototypes to design wins with autonomous system integrators.

For the engineers building robots meant to operate in the world's most challenging environments,search and rescue, mining, firefighting, and planetary exploration,the standard of care today is a fragile compromise. Teams often layer multiple sensor types, hoping one will work when another fails, or accept severe operational limitations. Astrabeam's bet is that a single, robust sense of sight, built on radio waves that pierce through obscurants, could change that calculus. If the technology performs as specified, it wouldn't just be a new component; it would redefine the environmental envelope for autonomous operations, turning missions once deemed too risky into routine procedures. The patient population, in this case, isn't defined by a disease code but by a job description: machines and the people who deploy them where vision is a luxury and failure is not an option.

Sources

1. [Astrabeam] About | https://www.astrabeam.com/about
2. [F6S] Company Profile | https://www.f6s.com/company/astrabeam-inc
3. [LinkedIn, 2026] Devin Reimer Profile
4. [PitchBook] Company Profile | https://pitchbook.com/profiles/company/522857-71
5. [NYC Innovation Hot Spot, May 2023] Speeding Up with Astrabeam | https://www.nycinnovationhotspot.org/post/speeding-up-with-astrabeam
6. [Element 46, 2026] Eight Cohort - Fall 2024 | https://www.element46.org/eight-cohort-fall-2024
7. [A3/Automate.org] Member Profile | https://www.automate.org/companies/astrabeam-llc