PostHeaderSection

Symbol

Global NavBar

PostContentSection

Global Footer

Blog Detail Page

Beirit Harvey

Today GSMA announced the launch of Open Telco AI, a global initiative designed to accelerate the development of telecom-grade AI through open collaboration across operators, vendors, researchers, and infrastructure providers.

TensorWave Partners with AT&T and AMD to Build the Foundation for Telco-Grade Models

Today, GSMA announced the launch of Open Telco AI, a global initiative designed to accelerate the development of telecom-grade AI through open collaboration across operators, vendors, researchers, and infrastructure providers.

As a launch partner, TensorWave is proud to support this effort alongside organizations including AT&T, AMD, and a broad ecosystem of telecom operators and academic partners. This is not another AI research program, it is infrastructure for an industry transition, powered by cloud infrastructure expertise.

And it reflects something telecom leaders already understand: General-purpose AI does not yet meet the operational reality of networks.


AI Still Doesn’t Speak Telco

Frontier models have advanced fast, but they routinely underperform on telecom-specific work:

 * interpreting network logs and telemetry
 * understanding standards documentation
 * supporting reliable troubleshooting and network ops automation

The gap shows up in adoption: only a small share of GenAI deployments in telecom have made it into network operations. Telco-grade performance requires more than scale. It requires industry-specific models, objective benchmarks, and production-grade infrastructure.


What Open Telco AI Actually Builds

The initiative introduces open-telco.ai, a shared platform for developing and evaluating telecom-focused AI systems.


1. Telco-First Models

Open models designed specifically for telecom tasks, including AT&T’s Bell model family, built to operate across hardware and cloud environments.


2. Industry-specific data

Curated datasets including network logs, telemetry, standards material, and structured knowledge graphs from research institutions and operators.


3. Production Compute

Infrastructure for training, fine-tuning, inference, and evaluation: powered by AMD GPUs and supported by the TensorWave platform and expert engineers.


4. Objective Benchmarks

The Telco Capability Index and telecom-specific evaluation metrics that measure real performance on operational tasks.


5. A Working Community

Competitions, challenges, and collaborative research programs that solve real operational problems, not hypothetical ones.


Why Open Infrastructure Matters

Telecom networks are long-lived systems, so the AI that supports them can’t be locked into rigid choices.

Open Telco AI is built around portability: open models, open toolchains, shared benchmarks, and hardware/cloud-agnostic development. It’s an operational requirement for regulated, long-lived systems.


Infrastructure Alone Isn’t Enough. Expertise Is Critical.

Moving telecom AI from experimentation into live networks is not simply a compute problem. It is an operational one.

Telco operators are deploying AI into environments where downtime carries real-world consequences. Models must behave predictably across long training cycles, evolving datasets, and distributed infrastructure environments. Success depends on partners who understand how AI systems perform in production, not just how they benchmark in isolation.

This is where infrastructure providers, like TensorWave, become technical collaborators and expert partners.

“Our team is proud to support the Open Telco Project by delivering high-performance AMD-powered compute infrastructure and working closely with ecosystem stakeholders. By enabling scalable experimentation, transparent benchmarking, and shared innovation, we’re helping accelerate the development of efficient, production-ready telco solutions.”
Brad Bokal

TensorWave works directly with operators, model developers, and enterprise AI teams to help translate research progress into deployable systems. From architecture design to workload optimization and evaluation environments, organizations rely on TensorWave teams to reduce the operational friction that often slows adoption.


TensorWave’s Role

TensorWave supports Open Telco AI by providing infrastructure for model training, fine-tuning, inference, and evaluation on AMD GPU hardware.

Our focus is straightforward: Deliver compute environments where telecom AI workloads can run predictably, reproducibly, and at scale.

That means:

 * High-memory GPU capacity for large models
 * Stable runtime environments for long training cycles
 * Reliable, consistent performance for accurate testing and evaluation.
 * Infrastructure designed for sustained throughput
 * Toolchains aligned with open standards and portable deployment
 * White glove expert support to ensure everything runs smoothly

Telecom AI development does not happen in short experimental runs. It requires sustained compute, consistent performance, and transparent system behavior. That is the environment TensorWave builds and supports.


Connecting the Ecosystem Behind Production AI

Open Telco AI reflects a broader shift happening across the industry: innovation now happens through ecosystems, not isolated vendors.

TensorWave plays an active role in bringing those ecosystems together.

In October 2025, TensorWave hosted a joint industry panel alongside AT&T and ScalarLM focused on deploying large-scale AI systems in regulated and operational environments. The discussion brought together infrastructure architects, telecom leaders, and AI researchers to address a shared challenge: how to move domain-trained models from experimentation into reliable production systems.

These conversations matter because production AI is interdisciplinary. Progress requires alignment between hardware providers, software platforms, operators, and researchers working against real constraints.

Through initiatives like Open Telco AI and industry collaborations with partners such as AT&T, AMD, and ScalarLM, TensorWave helps create the environments where those conversations translate into deployable and sustainable infrastructure.

The goal is simple: accelerate adoption by reducing fragmentation across the AI stack.


TensorWave Cloud: Engineered for AI Velocity

TensorWave is the industry’s premier AMD-exclusive cloud provider, delivering memory optimized AI infrastructure for training, fine-tuning and inference. 

Our platform combines the raw computational power of AMD's latest GPU architectures with innovative liquid cooling technology and ultra-high-speed networking, creating an optimized environment for the most demanding AI workloads. From large language model training to real-time inference, TensorWave provides the performance, scalability, and cost-efficiency that modern AI initiatives demand.

Ready to get started? Connect with a sales engineer. 

TensorWave Partners with AT&T and AMD to Build the Foundation for Telco-Grade Models

Product

Solutions

Resources

Company

Product

Resources

Solutions

Company