
SoftBank Corp. Pursues Telco AI Cloud to Become AI Infrastructure Provider
SoftBank has launched a program called Telco AI Cloud that combines a centralized GPU cloud for large‑scale model training with an AI‑RAN driven MEC layer for low‑latency inference, fronted by a software stack named Infrinia AI Cloud OS. The company described orchestration via an AITRAS Orchestrator and a Dynamic Scoring Framework that schedules workloads across distributed sites based on latency targets, resource availability and power constraints. SoftBank is positioning the stack to support operator‑hosted edge inference, industrial on‑prem deployments and an open AI‑RAN ecosystem by opening key orchestrator components and prioritizing O‑RAN interworking. Demonstrations and partner agreements revealed in late February and early March 2026 indicate the program is targeting near‑term pilots rather than purely long‑term research.
Concurrently, a separate industry track has surfaced: the GSMA’s Open Telco AI initiative emphasizes a model‑and‑dataset‑centric approach, publishing validated telecom model libraries, fine‑tuning datasets, and a public leaderboard of seven telecom‑specific benchmarks to make model evaluation reproducible across operators. Founding contributors named at launch include AT&T and AMD, and compute routing support was noted for partner TensorWave; GSMA’s portal is designed to seed reproducible baselines and community activity such as competitions and synthetic data pipelines. That effort is explicitly complementary to, but different from, an NVIDIA‑anchored consortium that publicly focuses on embedding accelerators, telemetry pipelines and low‑latency inference primitives into radio and edge stacks.
The coexistence of these approaches creates a practical tension: benchmark‑and‑model centricity (GSMA) aims to standardize objective metrics and datasets, while the NVIDIA consortium pushes reference implementations that tightly integrate NPUs/DPUs and telemetry into operator stacks. SoftBank’s Telco AI Cloud sits between these poles — it signals openness at the software and interface level while simultaneously depending on deterministic hardware behavior and telecom‑grade accelerators to meet latency and safety commitments. This means real‑world operator pilots will likely act as demand signals that steer semiconductor and orchestration roadmaps toward the accelerators and form factors that satisfy both reproducible model performance and hard latency/safety requirements.
Operationally, SoftBank’s architecture separates centralized training pools from edge inference and introduces a scoring/orchestration layer designed to factor energy and site heterogeneity into placement decisions; but heterogeneous hardware, site power limits and GPU scheduling remain top practical risks. From a regulatory and safety perspective, embedding learned models into RAN control loops raises questions about spectrum safety, deterministic latency guarantees and model auditability — constraints that will push participants toward staged field trials, curated datasets and joint lab validations before broad commercial rollout. If operators can coordinate standards, benchmarks and hardware references, value may shift to carriers offering managed, low‑latency inference; if not, the market risks bifurcating between hyperscaler managed edge services and narrowly focused industrial on‑prem deployments.
Read Our Expert Analysis
Create an account or login for free to unlock our expert analysis and key takeaways for this development.
By continuing, you agree to receive marketing communications and our weekly newsletter. You can opt-out at any time.
Recommended for you
GSMA Launches Open Telco AI to Build Telco-Grade Models and Tooling
GSMA unveiled Open Telco AI, a shared portal for telco-specific models, datasets, compute and benchmarks backed by AT&T and AMD to accelerate operator-grade network automation. The move arrives alongside a separate, NVIDIA-anchored industry push focused on embedding low-latency inference and orchestration primitives into radio and edge architectures, creating two complementary — and potentially competing — tracks for telco AI adoption.
Neoclouds Challenge Hyperscalers with Purpose-Built AI Infrastructure
A new class of specialized cloud providers—neoclouds—are tailoring hardware, networking, and pricing specifically for AI workloads, undercutting hyperscalers on cost and operational fit. This shift emphasizes inferencing performance, predictable latency, and flexible billing models, reshaping where companies run model training, tuning, and production inference.

Bell partners with Hypertec to deepen Canada’s sovereign AI infrastructure
Bell and Hypertec announced a Canada-hosted sovereign AI offering that pairs domestic GPU systems with a nationwide data-centre platform. The deal aims to keep sensitive workloads and regulated data inside Canada, shifting procurement and risk models for public sector and enterprise AI deployments.

NVIDIA-led consortium targets AI-native 6G architecture
A consortium led by NVIDIA and several carriers aims to bake intelligence into 6G network design, shifting radio control toward software and specialized accelerators. This move accelerates demand for telco-grade AI silicon, cloud-edge orchestration, and standards influence that could reshape procurement and vendor leverage.

G42 and Cerebras to deliver 8 exaflops of AI compute infrastructure in India
Abu Dhabi’s G42 and U.S. chipmaker Cerebras will install an on‑shore supercomputing system in India providing roughly 8 exaflops of AI processing capacity under Indian hosting and data‑sovereignty rules. The announcement, made at a high‑profile Delhi AI summit that also lifted related infrastructure stocks (an estimated ~$4 billion combined market‑cap gain for listed suppliers), signals strong political and commercial momentum — but delivery hinges on signed supply, land and power agreements, permitting and constrained accelerator allocations.

AI’s financialisation accelerates as tech giants commit $700bn to compute infrastructure
Five major US technology firms are planning roughly $700bn of capital expenditure this year, catalysing a market that treats compute capacity as collateral and spawning a wider set of financing vehicles — from bonds and CMBS to bespoke structured credit — while concentrated demand, permitting snarls and underutilisation risk sharpen credit and regulatory attention.

AMD deepens India push with TCS to deploy Helios rack-scale AI infrastructure
AMD and Tata Consultancy Services will roll out AMD’s Helios rack reference design across India in a partnership that packages AMD’s hardware stack with TCS’s local systems‑integration skills, targeting up to 200 megawatts of aggregated AI compute capacity. The program shortens procurement-to-live timelines but faces the same execution risks seen in other large-scale AI builds — municipal permitting, transmission and substation upgrades, chip and packaging supply limits, and the potential for idle capacity if build‑out outpaces verified demand — which could stretch deliveries into a 24–36 month window.
Australian AI infrastructure firm wins $10B financing to accelerate data‑center buildout
Firmus Technologies closed a $10 billion private‑credit facility led by Blackstone‑backed vehicles and Coatue to underwrite a rapid roll‑out of AI‑optimized campuses in Australia. The debt package targets deployment of Nvidia accelerators and up to 1.6 gigawatts of aggregate IT power by 2028, embedding the project in a wider global wave of specialized, high‑power data‑center financing.