Nvidia invests in Verkada to scale physical AI across 2.4M devices
Verkada has secured an investment from Nvidia to integrate its Cosmos foundation model and Physical AI Data Factory toolkit into Verkada's ecosystem of 2.4 million connected security devices. This technical partnership aims to improve computer vision features, including video search and scene reasoning capabilities, for Verkada's enterprise clients.
Key Takeaways
- Verkada reported a 68% improvement in mean average precision for spatial-temporal search queries since beginning the Nvidia collaboration.
- Integrated tools include the Physical AI Data Factory, which generates synthetic footage to supplement real-world training data for rare edge cases.
- The investment follows a December 2025 funding round led by Alphabet’s CapitalG that valued Verkada at $5.8 billion.
- Verkada currently serves 30,000 organizations globally, including more than 100 members of the Fortune 500.
Why It Matters
This deal marks a shift from general-purpose AI to 'physical AI,' where foundation models interact directly with real-world sensor data at the edge. By embedding the Cosmos model stack into Verkada’s 2.4 million devices, Nvidia moves beyond theoretical demos into a massive, live infrastructure for computer vision. For the streaming and surveillance ecosystem, this validates synthetic data generation as a primary method for training high-stakes security models. Watch for whether this partnership improves automated threat detection in retail and manufacturing, two sectors where Verkada is currently testing multi-model reasoning agents to replace manual monitoring.
Additional Context
The strategic tie-up coincides with a surge in the AI-powered video surveillance market, which is projected to reach approximately $8.16 billion in 2026 at a 20.7% CAGR, according to Research and Markets (December 2025). This growth is increasingly driven by 'Video Surveillance as a Service' (VSaaS) and the integration of edge-AI chipsets that allow for local processing of complex analytics. Nvidia’s Cosmos 3, unveiled at Computex in June 2026, represents the core technology behind this expansion; it is a mixture-of-transformers omnimodel that natively processes vision, sound, and spatial reasoning to reduce training cycles from months to days. Verkada’s move to deepen its AI stack also addresses historical vulnerabilities. Per SiliconANGLE, the company suffered a high-profile breach in 2021 where hackers accessed live feeds from 150,000 cameras. By shifting toward autonomous reasoning and synthetic training pipelines, the company aims to move beyond simple video streaming toward 'actionable intelligence,' where systems identify safety incidents without human oversight. This strategy aligns with broader industry movements, including Cisco’s 2025 launch of an AI-driven video suite for Webex and Genetec’s rollout of intelligent investigation tools, as reported by SNS Insider in June 2026. Furthermore, the use of Nvidia’s Physical AI Data Factory addresses a chronic industry bottleneck: the lack of diverse data for rare events like factory floor accidents. According to Nvidia (March 2026), this toolkit is already being used by firms like Uber and Volvo to simulate long-tail scenarios. For Verkada, which recently surpassed $1 billion in annualized bookings, the partnership serves as both a technical upgrade and a defensive moat against competitors like Milestone Systems and Hikvision in the race to dominate the smart-building and industrial security sectors.
Read full article at siliconangle.com
Get this in your inbox → Subscribe
Enjoy our coverage?
Add StreamingMeme as a preferred source on Google to see more of our streaming news at the top of your Search results.
Add as preferred source