Deutsche Telekom targets €2.5B savings with AI outside US

Deutsche Telekom used its AI Day, held on Monday (5) in Bonn, to quantify the impact of artificial intelligence on its cost structure. The operator expects to save approximately €2.5 billion in indirect costs by 2030 compared to 2023, outside the United States. For 2027, the target for gross savings from AI and automation has increased to around €1.1 billion, up from the €800 million pledged at the 2024 Capital Markets Day.

On the revenue side, the company has raised its projected AI-related revenue for 2026 to approximately €250 million, from €200 million previously, and forecasts €800 million by 2030, also excluding the American operation. The guidance for the year remains unchanged, as part of the additional savings from 2027 will be reinvested in digitalization and fiber-optic expansion in Germany.

Where the cuts occur

Telekom has identified four areas: network operations, customer service, software development, and administrative processes. These are the areas with high volumes of repetitive work, and the numbers shared illustrate the scale. The chatbot Frag Magenta handled approximately 2.6 million calls in the first half of 2026. In the network, the RAN Guardian agent, developed with Google Cloud, reduced response times to events from several hours to about one minute, according to the company.

More than 100,000 employees have already been trained to work with AI. The company provides tools such as the internal assistant AskT, ChatGPT Enterprise, and Microsoft Copilot, with the rule that the employee defines the objective, reviews the result, and is responsible for the decision.

Telekom did not disclose how many jobs automation will eliminate. This is the most significant gap in the presentation: the €2.5 billion target aims specifically at customer service and administration, the functions with the most personnel, and the company preferred to discuss reinvestment rather than workforce reduction.

The perspective across countries

The "outside the US" stipulation is important. The target excludes T-Mobile US and covers Germany, its European subsidiaries, such as operations in Greece, Hungary, and Poland, as well as T-Systems. It is within this European perimeter, characterized by a strong union presence, that the operator will need to show how AI reduces costs without announcing a formal job-cut program.

In Germany, the bargaining chip is fiber. The company links savings to a higher rate of connections installed, and part of the additional gains from 2027 will fund a faster installation pace. Savings from AI become, in Telekom’s narrative, an investment in national infrastructure.

In India, the effect comes through another channel. T-Systems ICT India maintains centers in Pune, Gurugram, and Bengaluru that develop software for the group, and software development is one of the four areas of savings. For Indian IT service providers with a strong portfolio in telecommunications, the message is the same as what other European clients have been signaling: fewer contracted hours per delivery and increased pressure for results-based contracts.

The counterpoint

There are good reasons for skepticism. "Gross" savings targets measured against a 2023 baseline include automation gains that were already underway before generative AI, and the distinction between what counts as AI and what is conventional efficiency relies on the company's own accounting. AI revenue, projected at €800 million by 2030, is also small relative to the size of the group, and the company limits it to operations outside the US.

The other side of the argument is that Telekom increased its 2027 target by €300 million in two years, suggesting that use cases have delivered more than planned. The company also maintained its guidance, indicating that it treats its target as a commitment, not just an aspiration.

For CIOs of other operators and sectors with mass customer service, such as banks and insurers, the most useful data isn’t the total for 2030. It’s the decision to publish AI targets as a commitment to investors, with a deadline and a comparison base. From now on, the question boards will ask those still treating AI as a pilot is why Deutsche Telekom can put a number on it while they cannot.