Newsletter
AI

Deutsche Telekom Targets €2.5 Billion in AI Savings by 2030

Deutsche Telekom expects AI to trim indirect costs by about €2.5 billion by 2030. The numbers, the use cases behind them and what is still unproven.

Telecom tower with antennas against a magenta and blue evening sky above a city skyline
AI-generated Image

Deutsche Telekom has put numbers on what it expects from artificial intelligence, according to the company’s own site. At an AI Investor Day in Bonn on October 5, the telecom group said AI and automation should cut its indirect costs by about €2.5 billion by 2030, compared with 2023. It also set a nearer goal of roughly €1.1 billion in gross savings outside the US by 2027. These are company targets, not results, so the interesting question is how much of the story is already visible in daily operations.

Two targets, two deadlines

The 2030 figure covers indirect costs, which means the overhead of running the business rather than the equipment itself. The 2027 figure is narrower in geography: it counts only operations outside the US, where the group operates through T-Mobile US, and it is stated as gross, before reinvestment. Deutsche Telekom says the savings are expected where large volumes of repetitive work are done: network operations, customer service, software development and administration.

Part of the money is earmarked. The company expects €100 million to €150 million of AI and automation savings between 2023 and 2027 to pay for a faster rollout of fibre connections in Germany, with more going into digital transformation. That makes fibre the visible reward for customers if the plan works.

The call center is already the proving ground

The strongest evidence sits in customer service. Deutsche Telekom says its Frag Magenta chatbot handled about 2.6 million customer calls in the first half of 2026. In the US, the company says AI agents now deal with 40% of customer contacts and that the number of customer service calls there has fallen by 55%.

AI also works behind the human agent. Systems can supply information during a call and complete the paperwork afterward, and other tools are designed to spot problems while a new connection is being set up. Early use cases in connection setup have, according to the company, reduced complaints by 30%, because the problem is caught before it turns into a complaint.

A network assistant that answers in a minute

On the technical side, Deutsche Telekom describes a RAN Guardian Agent for its mobile network. RAN is the radio access network, the part that connects phones to towers. The agent is designed to detect pressure building on the network, such as a traffic surge from a large event, and to help the operations team respond. The company says response times for such events dropped from several hours to about one minute.

AI is also used in software development and in replacing older IT systems, though the company gave no numbers for those areas.

Training thousands, keeping a human in charge

More than 100,000 employees have been trained under the company’s “AI for All” program. The tools available to them include AskT, ChatGPT Enterprise and Microsoft Copilot. The rule the company stresses is that employees set the goals, review what the AI produces and remain responsible for decisions.

The announcement does not say how the savings relate to staffing, so the figures should not be read as a headcount plan.

From cost cutting to selling AI

Savings are only half of the pitch. Deutsche Telekom expects about €250 million in AI-related business revenue outside the US in 2026 and aims for around €800 million by 2030. The products behind that include an Industrial AI Cloud in Munich, built for industrial simulations and AI workloads while letting customers keep control of their data, and an AI platform for small and medium-sized businesses that automates recurring tasks such as customer service and logistics.

For consumers, the company sees its Magenta AI Assistant as a base for new services and rate plans. It can translate mobile conversations in real time, answer questions and summarize content without a separate app or device.

How to read these figures

Everything above comes from the company, and the targets rest on 2023 as a baseline. A drop in call volume can mean problems are solved faster, or that customers gave up on the phone line, and the announcement does not separate the two. The sturdier test will be whether the group reports savings in its regular financial results and whether the fibre rollout speeds up as promised. Chief executive Tim Höttges summed up the plan by saying the company wants to roll out across the whole group what already works in individual applications.

More from Webitlio