AT&T launched OTel 2.0 telco AI models as usage hit 1T tokens

AT&T launched OTel 2.0, an open-source set of telecom AI models, as its internal AI usage passed 1 trillion tokens per month, according to Fierce Network.

TL;DR

• AT&T released OTel 2.0 as an open-source telecom AI model set.
• AT&T said the models were trained on AMD silicon.
• AT&T said its AI usage now exceeds 1 trillion tokens per month.
• The company framed the launch around practical AI deployment rather than rapid rollout without clear use cases.

AT&T said OTel 2.0 is a new version of its open-source telecom AI models. The company said it trained the models on AMD hardware. Fierce Network reported the launch as part of AT&T's broader AI work inside its network and operations.

AT&T said its AI usage has passed 1 trillion tokens per month. A token is a unit of text that AI models process. The reported usage figure indicates the scale of AT&T's internal AI activity, but the source excerpt did not specify a breakdown by application, business unit, or customer-facing service.

AT&T's message on AI deployment was to avoid rushing implementation without a defined purpose. Fierce Network summarized that view with the phrase about stopping the habit of “racing from stoplight to stoplight.” The source excerpt did not provide further technical detail on model sizes, licensing terms, or deployment timelines for OTel 2.0.

Related questions

• What is OTel 2.0 and how does AT&T describe its purpose?
• What does AT&T mean by 1 trillion AI tokens per month?
• What role did AMD silicon play in training AT&T's OTel 2.0 models?
• How is AT&T positioning practical AI deployment in telecom operations?