Nearby Computing Takes Aim at Enterprise AI Sprawl with Launch of AI Mesh

Barcelona, Spain, 30 Sept., 2026: Nearby Computing, the European deeptech company behind the NearbyOne orchestration platform, today announced AI Mesh, a major extension designed to give enterprises a single operational layer for the increasingly distributed infrastructure behind AI.

AI is no longer running in one place. Models are being deployed in AI factories, private data centres, public clouds, edge locations and, increasingly, directly on devices. At the same time, autonomous agents are connecting those models to data, tools and other services across multiple systems.

For large organisations, that creates a new operational and governance challenge: adopting AI as quickly as the technology evolves while retaining control over where workloads and data run, who can access them, how much they cost and which technology providers they depend on.

AI Mesh is designed to make that distributed AI environment operate as one system.

“Our view is that AI is becoming distributed by default, and distributed systems need a unified control plane. Our ambition is for NearbyOne to become that layer: allowing organisations to adopt AI as fast as the technology moves, across their own infrastructure, the edge and any provider, without losing control of where their data goes, who can reach it or what it costs.”

Josep Martí, CEO of Nearby Computing

Enterprises can use AI Mesh to deploy approved models, agents, MCP servers and AI platforms from a governed catalog across private or public infrastructure, reducing the time required to move AI services from experimentation into production. A single OpenAI-compatible API sits in front of self-hosted models and more than 16 public AI providers, allowing applications and agents to move between technologies without rebuilding the surrounding infrastructure or locking the organisation into a single AI vendor.

With the EU AI Act in force since August 2026 and new European rules setting sovereignty levels for public-sector cloud, where AI runs has become a compliance question. Against this backdrop, Nearby Computing is building AI Mesh with reference to NVIDIA’s distributed inference reference architecture, while working with any hardware and any AI provider. The capabilities are currently being tested by Nearby Computing customers. It builds on technology already used to orchestrate highly distributed infrastructure for some of the world’s most demanding telecommunications, enterprise and public-sector environments.

For regulated enterprises and public bodies, the same architecture provides control over where AI workloads execute and where data is processed. An organisation can, for example, keep sensitive inference within a particular country or region while still using public AI services for other workloads. Different teams or departments can operate within isolated environments with their own permissions and data access, while authorised services can still discover and communicate with each other across locations.

Every AI call passes through a managed gateway, creating a common enforcement and audit point for identity, access policy, rate limits, routing and metering. Enterprises can see which models and services were used, where data was sent and who was responsible for a request, whether the model is running internally or through an external provider.

That provides a common governance layer across an AI environment that would otherwise span multiple technology stacks, helping organisations maintain sovereignty over their data and infrastructure while improving audit readiness.

AI Mesh also tackles a less visible problem with enterprise AI: nobody gets a single bill.

Public AI is generally priced by tokens, while private AI consumes GPUs and CPUs whose cost is buried inside infrastructure. NearbyOne measures both. It meters public-provider usage at the provider’s published rates and converts private compute into comparable inference costs, breaking spend down by team, model and provider. It can also identify idle GPU and CPU capacity and project future run rates.

For a large enterprise, that can mean attributing AI costs to individual business units rather than treating AI infrastructure as an undifferentiated technology expense. For a public-sector organisation, it can provide the same visibility by department or service.

That visibility is becoming increasingly important as AI moves deeper into enterprise operations. Just 35% of enterprises have full visibility into their AI spending, while organisations with full visibility into their AI operating costs are five times more likely to report established ROI than those without it, according to KPMG’s Global AI Pulse Q2 2026.

“Agents make this considerably more important. Once software starts acting autonomously across models, data and other services, enterprises need to know what it can reach, what it is allowed to do and what every action costs. That control has to be part of the infrastructure, not something added afterwards.”

Josep Martí, CEO of Nearby Computing

The launch comes as the industry itself moves towards a more distributed model for AI. Alongside the rapid expansion of cloud AI infrastructure, companies including Apple are pushing more AI inference onto local devices. For enterprises, the result is unlikely to be a choice between cloud, private infrastructure, edge or device AI, but a combination of all of them that needs to be operated coherently.

More than half of businesses already identify hybrid cloud as their primary infrastructure strategy, but only 23% of workloads currently run that way, according to HyperFRAME Research Lens.

AI Mesh extends the orchestration and automation principles that Nearby Computing has developed through deployments with some of the most technically demanding infrastructure operators. NearbyOne remains vendor-agnostic by design, working across hardware, cloud and AI providers and plugging into existing environments without forcing organisations to migrate their infrastructure.

About Nearby Computing

Nearby Computing is a European deeptech company that orchestrates and automates distributed cloud-to-edge environments, networks, infrastructure, applications and AI from a single operational platform.

Its NearbyOne platform works regardless of underlying hardware or provider, plugging into what organisations already run rather than requiring them to rip and replace existing infrastructure.

Nearby Computing works with global telecommunications operators, enterprises across multiple industries and public-sector organisations, helping them operate increasingly distributed infrastructure with greater automation, control and visibility.