NVIDIA backs Verkada as AI security tie-up expands

Verkada has announced a technical collaboration with NVIDIA, which has also taken an investment stake in the company.

The deal links Verkada's physical security platform with NVIDIA's artificial intelligence tools as Verkada expands AI-based products across schools, hospitals, retailers and manufacturers.

Verkada says it has more than 2.4 million devices deployed across 170 countries and serves 30,000 organisations. It sells video security cameras, access control, environmental sensors, alarms, workplace tools and intercoms through a cloud-based platform.

NVIDIA's backing follows a strategic investment in Verkada from CapitalG, Alphabet's growth fund, at the end of last year. Financial terms were not disclosed.

Technical work

The companies have already spent months working on systems designed to improve how AI interprets activity in buildings and other real-world settings. The collaboration focuses on video search, semantic search and synthetic data generation for model training.

According to Verkada, the work uses NVIDIA Cosmos world foundation models and NVIDIA Physical AI Data Factory. Those tools have helped the company train and deploy models more quickly across its installed base.

One early result was a 68% improvement in mean average precision, or mAP, in AI-powered search for spatial-temporal understanding. Verkada says this has reduced investigation times by making search more accurate.

The company is also developing what it describes as a multi-model search agent architecture. It is exploring reasoning models for more complex and less structured scenarios, including health and safety incidents on factory floors and retail stock loss.

Built environment

The agreement highlights growing competition to apply AI beyond software and consumer services to physical settings such as offices, shops, schools and industrial sites. In these environments, companies are using computer vision and related models to identify incidents, track patterns and respond more quickly to events.

For Verkada, that means using operational and security data captured by connected devices installed across customer sites. The company argues that this data can be turned into useful information for safety and operations teams.

"Verkada has been building and deploying Physical AI before the term existed. With our footprint of more than 2.4 million devices across 170 countries and 30,000 organizations, we've proven that the built environment is one of the largest beneficiaries of AI," said Filip Kaliszan, Co-Founder and Chief Executive Officer of Verkada.

He added that the company's products are already used in settings where the consequences of missed incidents can be significant, from education to manufacturing.

"Working with NVIDIA supercharges what we've spent nearly a decade building: AI that keeps students safe in schools, protects workers on factory floors, helps retailers prevent theft, and enables organizations to operate more efficiently," Kaliszan said.

Investor signal

NVIDIA has increasingly paired technical partnerships with selective investments in companies building systems around its AI infrastructure. Its investment in Verkada signals interest in a segment where AI models are being applied to cameras, sensors and physical operations, rather than only digital workflows.

The area has drawn more attention as businesses seek practical uses for machine learning in environments that generate large volumes of visual and sensor data. Security and workplace monitoring vendors have been among those moving quickly to add AI features, particularly in search, anomaly detection and incident review.

Verkada says more than 100 Fortune 500 companies are among its customers. Its global installed base gives it access to a large volume of operational footage and related data, which is central to training and refining visual AI systems.

The collaboration is intended to improve multimodal embeddings and vector retrieval, methods used to match natural language prompts with relevant images, events or clips. These techniques are becoming increasingly important as software vendors try to make surveillance and operations systems easier to search without manual review.

Analysts have pointed to synthetic data as another growing area of importance in industrial and security AI. By generating artificial training data, companies can try to cover rare or hard-to-capture scenarios that may not appear often in real footage but still matter in safety and loss prevention.

In Verkada's case, that could include unusual incidents on production lines or theft-related behaviour in retail environments. Its work with NVIDIA is aimed at improving accuracy in those settings.

Verkada says the collaboration reflects a broader push to develop more context-aware AI for the built environment, with the aim of making physical spaces safer and more resilient.