Amarjot Singh, the founder of Skylark Labs, likes to talk about the physics of the problem. His company builds AI for machines that move through the real world, where connectivity is spotty, data is scarce, and a mistake can mean a crashed drone or a foreign object sucked into a jet engine. The solution, he argues, isn't a bigger cloud model. It's a brain that lives on the device, learns from a single experience, and doesn't need to phone home. It's a technical philosophy that recently found its most demanding test bed: the flight deck of an Indian Navy aircraft carrier [PR Newswire, March 2026].
For a startup selling "edge-native, self-learning AI," there is no better validation than a successful demo in one of naval aviation's most chaotic environments. Skylark's system used fixed cameras to feed live imagery to a ground staff dashboard, where its software automatically detected, classified, and prioritized debris for removal [Aviation International News, March 2026]. It's a classic Skylark play: take a high-stakes defense validation and immediately point to commercial applications at airports [FODNews, March 2026]. The company, based in Palo Alto and founded in 2018, is threading a needle between the slow, lucrative sales cycles of government contracts and the adaptive needs of modern physical infrastructure.
The Wedge Is in the Wiring
Skylark's portfolio reads like a checklist for securing a modern nation-state: counter-drone systems, autonomous aerial and ground robots, airfield safety monitoring, and infrastructure surveillance [Perplexity Sonar Pro Brief]. The through line isn't a single product, but an architectural bet. The company emphasizes on-device processing to reduce latency, protect sensitive data, and operate without reliable connectivity [Perplexity Sonar Pro Brief]. In practice, this means an AI model deployed on a drone or a fixed camera that can adapt to new threats or environmental changes without a lengthy retraining cycle in the cloud.
Its core technical product is called CFAM, described as an AI brain for humanoids, drones, and ground vehicles that keeps learning after deployment [skylarklabs.ai]. The company says it has tested CFAM across five different physical systems, including a quadruped, a humanoid, and an off-road vehicle [Robotics and Automation News, September 2026]. This flexibility is the commercial pitch: a single, adaptable intelligence stack that can be sold into defense, then repurposed for monitoring a Coca-Cola manufacturing campus or an Adani Group infrastructure site [PR Newswire, March 2026] [YouTube].
Traction Through Strategic Partnerships
Skylark's growth appears less about a viral SaaS product and more about landing strategic design wins and partnering with established defense integrators. Its customer and partner roster is a mix of blue-chip names and government entities, suggesting a business built on credibility and referrals.
| Entity | Type | Engagement | Source |
|---|---|---|---|
| U.S. Space Force | Government | Entered $500M MATTER SBIR Phase III IDIQ contract vehicle | [PR Newswire, June 2024] |
| Hexadyne Corporation | Defense Integrator | Strategic partnership; $4M order for 20 ARIES counter-UAS systems under a GCC program | [Business Insider Markets, August 2026] |
| Indian Navy | Government | Successful FOD detection demo on an aircraft carrier | [Yahoo Finance, March 2026] |
| Mercedes-Benz R&D India | Commercial | Completed Phase I of an AI-driven Accident Hotspot Detection System | [Yahoo Finance, December 2024] |
| Adani Group | Commercial | Contracts for AI-enabled safety and security across drone operations | [PR Newswire, March 2026] |
The $4 million order from August 2026 is particularly telling. It wasn't a direct sale, but part of a larger, Hexadyne-led program for Gulf Cooperation Council defense that could be worth up to $32 million [The AI Journal, August 2026]. This is the classic path for a defense tech startup: prove the tech, then ride the distribution of a prime contractor.
Building a Defense-Grade Team
Founder Amarjot Singh's background is almost a template for this category. He holds a doctorate in AI from the University of Cambridge, did a postdoc at Stanford, and served as a scientist with DARPA on an advanced drone program [Perplexity Sonar Pro Brief] [LinkedIn - Mesh]. In 2025, he formed a defense advisory board, adding figures like John Donovan, the former CEO of AT&T Communications and a Lockheed Martin board member [skylarklabs.ai, September 2025].
The most significant operational hire came in August 2026, when Skylark tapped Vince Nakayama, former CTO of GoPro, as its Chief Product Officer [Business Insider Markets, August 2026]. Nakayama's resume includes hardware and product roles at Apple, Sony, and Lucid Motors [The AI Journal]. For a company whose "AI brain" must eventually live inside shipped hardware, bringing in an executive who has repeatedly navigated the journey from prototype to mass production is a clear signal of intent.
The Cash and the Competition
Financially, Skylark operates with a level of opacity common in dual-use tech. While one source reported the company secured about 70% of commitments for a $150 million round targeting a $1 billion valuation [Whalesbook], this is not independently verified. What is clear is that revenue is starting to flow through contract vehicles and direct orders. The capital intensity of this space is high, requiring not just software R&D but also integration work, security certifications, and likely custom hardware deployments.
The competitive field is crowded with well-funded players pursuing different slices of the autonomous systems and defense AI market. Skylark must contend with:
- True Anomaly and Saronic: Companies focused squarely on space and maritime defense domains.
- LandingAI: An industrial computer vision platform founded by Andrew Ng, targeting manufacturing and inspection.
- Scenera: A competitor in edge AI for video analytics and security.
The risk for Skylark is one of focus. The ambition to be a general "AI brain" for any physical system is vast, spanning humanoids, drones, and off-road vehicles. Meanwhile, the sales motion is inherently slow, navigating the procurement labyrinths of multiple governments and large conglomerates. The company's answer appears to be its partnership model, letting integrators like Hexadyne handle the final mile to the end customer while Skylark supplies the core intelligence.
The Unit Economics of a Saved Engine
The financial case for Skylark's technology is best made not in software margins, but in avoided disasters. Take the foreign object debris (FOD) system. A single FOD-related incident on a commercial flight can cost an airline millions in repairs, delays, and cancellations. On a military aircraft carrier, the cost is measured in national security readiness.
A back-of-the-envelope calculation puts the value in perspective. If Skylark's airport system, priced similarly to its defense order, costs around $200,000 per installation (estimated), it would need to prevent just one serious incident every few years at a major hub to pay for itself. The math gets even clearer in defense, where the cost of a lost drone or a compromised mission can run into the tens of millions. The company isn't selling features; it's selling a reduction in existential risk.
For Skylark Labs to graduate from a promising contractor to a defining platform, it must prove it can do more than a handful of impressive demos. It must scale its CFAM "brain" across enough systems and customers to create a true ecosystem advantage, moving from a vendor of point solutions to the default operating system for autonomous, edge-native machines. Its most telling competitor, then, isn't another startup. It's the internal R&D teams at giants like Lockheed Martin or Northrop Grumman, and the entrenched, bespoke way the defense industry has always built intelligence into hardware. Skylark's bet is that a startup's agility and a unified software approach can outpace them.
Sources
- [PR Newswire, March 2026] Skylark Labs Successfully Completes Fixed Foreign Object Debris Detection Demonstration Aboard an Indian Navy Aircraft Carrier | https://www.prnewswire.com/news-releases/skylark-labs-successfully-completes-fixed-foreign-object-debris-detection-demonstration-aboard-an-indian-navy-aircraft-carrier-302713291.html
- [Aviation International News, March 2026] Skylark Labs Deploys Fixed FOD Detection at Commercial Airports | https://www.ainonline.com/aviation-news/defense/2026-03-20/skylark-labs-deploys-fixed-fod-detection-commercial-airports
- [Perplexity Sonar Pro Brief] Company and product description for Skylark Labs
- [Robotics and Automation News, September 2026] Skylark Labs Tests CFAM AI Brain Across Five Physical Systems | https://roboticsandautomationnews.com/2026/09/15/skylark-labs-tests-cfam-ai-brain-across-five-physical-systems/124567/
- [Business Insider Markets, August 2026] Skylark Labs Bags $4 Million ARIES Order Under Hexadyne-Led GCC Counter-UAS Program | https://markets.businessinsider.com/news/stocks/skylark-labs-bags-4-million-aries-order-under-hexadyne-led-gcc-counter-uas-program-worth-up-to-32-million-1036476641
- [PR Newswire, June 2024] Skylark Labs Announces Entrance into Monkton’s $500M MATTER SBIR Phase III IDIQ | https://www.prnewswire.com/news-releases/skylark-labs-announces-entrance-into-monktons-500m-matter-sbir-phase-iii-idiq-302168610.html
- [Yahoo Finance, December 2024] Skylark Labs and Mercedes-Benz R&D India Complete Phase I of AI-Driven Accident Hotspot Detection System | https://finance.yahoo.com/news/skylark-labs-mercedes-benz-r-124500478.html
- [Business Insider Markets, August 2026] Skylark Labs Taps GoPro CTO Vince Nakayama as Chief Product Officer | https://markets.businessinsider.com/news/stocks/skylark-labs-taps-gopro-cto-vince-nakayama-as-chief-product-officer-to-scale-its-defense-robotics-stack-1036508140
- [skylarklabs.ai, September 2025] John Donovan Joins Skylark Labs Defense Advisory Board | https://skylarklabs.ai/about-us.html
- [The AI Journal, August 2026] Skylark Labs Expands Self-Learning Defense Robotics Across US, GCC | https://www.theaijournal.com/article/skylark-labs-expands-self-learning-defense-robotics-across-us-gcc-leveraging-hexadynes-three
- [Whalesbook] Report on Skylark Labs funding commitments | https://app.whalesbook.com/company/skylark-labs