Why Telcos Are Becoming AI Cloud Providers: Lessons From NVIDIA GTC 2026
Nine telecom operators across Europe, the Middle East and Africa are building or expanding NVIDIA-powered AI infrastructure. The harder question is what an AI cloud business demands beyond GPUs.

Telecommunications operators have spent decades building the infrastructure that connects businesses, devices, and people. Now, many are looking at how those same assets can support a different business: AI cloud services.
At NVIDIA GTC Berlin 2026, held October 20–22, the company outlined how telecom operators are expanding accelerated computing across data centers and distributed network locations. The announcements included developments involving Deutsche Telekom, Orange, Elisa, Nokia, and other operators investing in AI infrastructure.
The direction is becoming clearer. Telcos are exploring how to move beyond connectivity into AI hosting, inference, enterprise cloud services, and distributed computing.
But building an AI cloud business requires more than installing GPUs. It means delivering a complete infrastructure platform that enterprise customers can use, manage, and trust.
NVIDIA GTC 2026: Telcos Are Becoming AI Infrastructure Providers
On October 21, NVIDIA announced that nine telecom operators across Europe, the Middle East, and Africa were building or expanding NVIDIA-powered infrastructure across centralized and distributed locations.
Several announcements illustrate how the industry is approaching the opportunity:
Deutsche Telekom plans to expand its Industrial AI Cloud, which hosts open models and supports customers developing industrial AI applications. The company is also testing AI-powered radio access network (AI-RAN) technologies with Nokia and NVIDIA.
Orange is hosting open AI models as managed services through its Live Intelligence platform, while exploring distributed AI applications such as real-time voice translation.
Elisa has deployed NVIDIA's Aerial RAN Computer at a distributed network location, supporting enterprise AI applications and testing AI capabilities within its network.
T-Mobile completed an AI-RAN call on a commercial network using NVIDIA and Nokia technologies, demonstrating how accelerated computing can support telecommunications workloads.
NVIDIA's broader vision involves using telecom infrastructure to deliver AI services closer to the locations where data is generated and applications operate.
The topic also received dedicated attention during GTC Berlin's October 22 panel, Telcos as AI Providers: Infrastructure Strategies and Monetization Models, featuring executives from Nokia, Orange, Elisa, and Deutsche Telekom.
These developments reflect two related opportunities. Operators can use AI to improve their existing networks, and they can use their infrastructure to provide AI services to external customers.
The second opportunity could become particularly important for telcos seeking new enterprise revenue streams.
Why Telcos Are Moving Beyond Connectivity
Telecom operators already possess several assets that are valuable in the AI infrastructure market: data centers, fiber connectivity, regional network facilities, enterprise customers, and established operational teams.
These assets offer a potential foundation for delivering cloud services.
1. Enterprise customers need local AI infrastructure
As enterprises deploy AI into production, infrastructure requirements become more specific.
Organizations operating in financial services, healthcare, government, and other regulated sectors may need stronger assurances about where their data is stored, processed, and accessed.
Latency also matters. Applications such as industrial automation, computer vision, and real-time analytics can benefit from compute resources located closer to users and devices.
Telcos with regional infrastructure may be positioned to address these requirements through locally hosted services.
However, proximity alone does not guarantee compliance, security, or performance. Operators still need appropriate access controls, operational safeguards, and infrastructure management.
2. AI services create new monetization opportunities
Traditional telecom businesses rely heavily on connectivity services, including mobile, broadband, and enterprise networking.
AI infrastructure introduces additional commercial possibilities.
An operator could offer dedicated GPU capacity to enterprise customers, managed inference endpoints for application developers, private compute environments for regulated workloads, or storage services integrated with its network.
These services may create new revenue opportunities beyond connectivity contracts.
The economics will depend on customer demand, capacity utilization, hardware costs, power availability, and the operator's ability to deliver reliable services.
3. Existing infrastructure can support regional AI deployment
Telcos operate physical and network infrastructure across geographically distributed locations.
Some facilities may be suitable for additional computing capacity, particularly where power, cooling, connectivity, and space are available.
Regional deployments can support data residency requirements and reduce the distance between applications and compute resources.
For operators, the opportunity is to build on existing infrastructure advantages while extending their service portfolios.
The challenge is turning those assets into a cloud business customers can actually consume.
What Does It Take to Become an AI Cloud Provider?
A GPU cluster is an important component of AI infrastructure, but it is only one part of a commercial cloud service.
Enterprise customers need ways to provision resources, deploy workloads, access data, manage permissions, and understand costs.
An AI cloud offering typically requires several interconnected layers.
Infrastructure layer | Why it matters |
|---|---|
GPU and compute | Runs training, inference, and general-purpose workloads |
Storage | Holds datasets, model weights, checkpoints, and application data |
Networking | Connects workloads, infrastructure, and customer environments |
Inference services | Makes AI models available through managed endpoints and APIs |
Identity and security | Controls access, isolation, credentials, and permissions |
Management and billing | Enables provisioning, usage tracking, pricing, and customer operations |
These layers must work together.
For example, an enterprise deploying an AI application might require GPU instances, private networking, object storage, and an inference endpoint. If those components are managed through separate systems, operating the application becomes more complicated.
For telcos planning to serve multiple customers, the requirements extend further. Each customer may need its own tenant, credentials, resource limits, and billing arrangements.
The operator must also manage the underlying infrastructure, including monitoring, updates, capacity planning, and support.
This is where an infrastructure platform becomes commercially important. It provides the operational foundation needed to turn physical computing resources into services that customers can purchase and manage.
Why Sovereign and Regional AI Clouds Matter
One of the advantages telcos can offer is the ability to deliver infrastructure within specific geographic markets.
Many organizations want greater visibility into where their data resides, which entities operate their infrastructure, and how sensitive workloads are accessed.
Sovereign AI infrastructure addresses these concerns through deployment, operational, and governance controls designed around jurisdictional requirements.
Data residency is part of this equation, but it is not the entire requirement.
Customers may also need defined access policies, encryption, network isolation, auditability, and assurances about who can administer their environments.
Regional operators can potentially offer these controls while maintaining closer commercial relationships with local enterprise customers.
However, sovereignty requires more than a local data center address. It depends on the complete operating model, including legal, technical, and administrative controls.
For telcos entering AI cloud services, building those capabilities into the platform from the beginning can be an important differentiator.
How Aurora Helps Telcos Build Their Own AI Cloud
The expansion of telcos into AI infrastructure creates an opportunity for platforms that simplify the deployment and operation of cloud services.
Aurora provides private AI infrastructure for telcos, regional cloud providers, and managed service providers looking to deliver cloud services under their own brands.
The Aurora AI Cloud brings compute, GPU infrastructure, S3-compatible storage, networking, and inference into a unified platform.
Rather than developing a separate management system for every infrastructure component, operators can offer these services through a common environment.
A white-label cloud for telecom operators
Through Aurora's partner program, telcos can offer cloud infrastructure under their own logo, domain, and pricing.
Customers access the operator's branded portal, while Aurora manages the underlying infrastructure.
This allows operators to maintain ownership of their customer relationships and commercial offerings.
The white-label model also includes Aurora Backoffice, which provides tools for managing the business side of cloud services.
Operators can:
Create and manage individual customer tenants.
Configure customer-facing plans and pricing.
Track resource usage and billing.
Manage cloud services through a centralized platform.
These capabilities matter because operating an AI cloud involves both infrastructure engineering and commercial service delivery.
Infrastructure that supports different deployment requirements
Aurora also supports private and regional deployment models, including managed infrastructure, dedicated compute environments, and deployments designed around specific customer requirements.
For telcos with existing hardware or data center resources, Aurora's platform can provide the cloud management and operational layer.
Operators seeking new capacity can explore managed infrastructure or dedicated deployments.
The objective is to give operators a way to expand their cloud offering without independently building every element of the infrastructure and management stack.
This approach can help reduce operational complexity while allowing telcos to focus on enterprise customers, market development, and service differentiation.
Importantly, each deployment still requires an assessment of infrastructure readiness, workload requirements, security controls, and commercial viability.
The Next Phase of Telecom Infrastructure
NVIDIA GTC Berlin 2026 offered a clear indication of where parts of the telecommunications industry are heading.
Operators are exploring how their networks, facilities, and computing capacity can support AI applications alongside traditional connectivity services.
For some, that means using AI to improve network operations. For others, it means building new enterprise offerings around inference, private compute, storage, and managed AI infrastructure.
Not every operator will pursue the same strategy. The economics of AI cloud services depend on infrastructure investment, operational capability, and sustainable demand.
But the opportunity is becoming increasingly relevant as enterprises look for alternatives that provide greater geographic control, integrated services, and infrastructure suited to production AI workloads.
The next challenge for telcos is commercial execution: packaging infrastructure into services customers can provision, operate, and pay for.
Aurora's white-label AI cloud is designed to support that transition, giving operators a managed infrastructure platform they can bring to market under their own brand.
Explore how Aurora can help you build an AI cloud business.
Related posts

Why AI Inference Is Becoming a Storage Problem
AI inference is no longer just a GPU question. As context windows grow and agents make repeated model calls, KV caches and the memory that holds them are shaping the economics of running a model.

AI Agents Can Write Software. Who Runs It?
AI agents can now build and debug whole applications, but generating code is only half the job. The harder problem is giving that code a secure place to run, store data and reach other systems at scale.

What Makes a Neocloud Sustainable?
Neoclouds are measured by GPU count, but the harder question is whether they can earn a return on that hardware. A look at utilization, financing costs and why busy GPUs are only part of the equation.