This transition is occurring as the telecommunications giant manages a massive volume of data, currently processing 45 billion tokens—the basic units of text or code that AI uses to process information—every day.
This transition is primarily driven by a desire to control costs and avoid being locked into a single technology provider.
By utilizing open models instead of relying solely on expensive, proprietary systems from major vendors, AT&T can run AI applications on its own cloud infrastructure or at the "edge" of its network, where data is closer to the user.
This flexibility allows the company to match specific tasks, such as summarizing customer service interactions or generating software code, to the most cost-effective and appropriately sized model available.
To manage this complex environment, the company relies on an internal platform called "Ask AT&T," which acts as a gateway to route employee requests to the most efficient model for the job.
This system helps the company maintain security while scaling AI capabilities across its workforce.
As AT&T nears its 80% goal, the move reflects a broader effort within the enterprise sector to gain more transparency and autonomy over the specialized software powering their digital infrastructure.