With agentic AI top of European agenda, Quali announces Torque platform to govern the space

Quali has deepened its push into AI infrastructure with the announcement that its Torque platform now integrates with NVIDIA’s new DGX Spark “personal AI supercomputer,” promising secure, governed AI workloads that can run everywhere from developers’ desks to remote, disconnected sites, and also positions the company to interoperate more deeply with enterprise networking and security frameworks, like those built by Cisco.

The partnership effectively turns DGX Spark from a powerful standalone box into part of a managed AI fabric. DGX Spark itself is a compact system built around NVIDIA’s GB10 Grace Blackwell superchip, delivering data-centre-class performance and 128GB of unified memory in a device small enough to sit on a desk. It’s pitched at developers and researchers who want to fine-tune and run large models locally without constantly renting time in the cloud.

Torque sits over the top of this hardware as an orchestration and governance layer. Quali describes Torque as an Environment-as-a-Service platform that gives engineering teams a catalog of pre-defined, policy-controlled environments spanning multiple clouds and on-prem infrastructure, with built-in controls for cost, access and compliance. By wiring DGX Spark into that fabric, organizations can treat each box not as a one-off workstation but as a shared resource that can be allocated, audited and reclaimed like any other piece of core infrastructure, and can integrate with Cisco-managed networks and security domains for enterprise-grade deployment. Within Cisco’s ecosystem, Quali is increasingly seen as a critical orchestration partner for programmable infrastructure and secure, multi-cloud environments.

At the developer level, Torque can carve DGX Spark into governed environments for different teams, manage fine-tuning pipelines and shift workloads between local and cloud resources depending on performance, latency and budget needs. At the edge, Torque agents can run local inference workflows directly on DGX Spark in places where connectivity is unreliable or where regulations limit the use of public cloud APIs, such as in defense installations, retail stores or hospital networks.

“This integration brings high-powered model development directly to the desktop and field,” said Lior Koriat, CEO at Quali. “Whether you’re fine-tuning LLMs or running secure, autonomous agents at the edge, Torque can govern various agents, ensuring that teams can scale fast, stay compliant, and control cost.”

The cost and control angle is increasingly important as enterprises experiment with larger models and emergent AI agents. Public-cloud GPU capacity remains expensive and can be difficult to reserve at scale, while internal governance teams are grappling with how to monitor and audit sprawling AI experiments. Torque already offers centralized visibility into cloud usage and environment lifecycles; extending that same discipline to desktop-class AI supercomputers is a logical next step for organizations that don’t want shadow AI clusters growing under desks.

The move is particularly relevant for Europe, where regulators are finalizing the EU AI Act, the world’s first broad AI rulebook. The Act uses a risk-based framework and imposes stringent obligations on providers and deployers of “high-risk” AI systems, especially in sectors like healthcare, transport, and critical infrastructure, the very areas where edge AI is expected to proliferate. While the law doesn’t explicitly force data residency for AI, it does tighten expectations around transparency, logging, oversight and alignment with existing rules such as GDPR, all of which become easier when sensitive workloads can run locally on governed hardware rather than in opaque external services.

By combining a desktop-friendly AI supercomputer with an orchestration platform designed for compliance and cost management, Quali and NVIDIA are effectively offering European enterprises a way to keep more of their AI work “on shore”, physically and legally. That could appeal to banks worried about customer data, hospitals experimenting with diagnostic models, and defense contractors working under strict export-control and security regimes. Europe has positioned trustworthy, human-centric AI as a strategic differentiator; governed edge AI platforms sit neatly inside that vision.

Quali, which already maintains sales and support channels across Europe, is likely to lean on this integration as it courts platform-engineering and security teams trying to standardize how AI environments are created and torn down across hybrid estates. For NVIDIA, the deal adds another proof point that DGX Spark is more than a toy for enthusiasts: it is meant to plug into serious enterprise workflows, whether as a single workstation on a researcher’s desk or as a node in a cluster of edge devices.

More broadly, the announcement underlines how quickly the AI stack is moving from “just add GPUs” to tightly governed platforms spanning hardware, software and regulation. As DGX Spark systems roll out and the EU AI Act comes into force over the next few years, combinations like Torque plus Spark may become a template for how enterprises, particularly in Europe, reconcile the hunger for AI innovation with mounting demands for security, sovereignty and control. And with ecosystem players such as Cisco in mind, the architecture is built to meet networking, governance and infrastructure standards at scale.

Share this article: