Deutsche Telekom targets 2.5 billion euros a year in AI savings by 2030

Deutsche Telekom wants artificial intelligence to show up on the balance sheet, not just in press demos. At its AI investor day in Bonn on Monday, 5 October 2026, the Bonn based DAX group said AI and automation should cut its indirect costs by about 2.5 billion euros a year by 2030, compared with 2023. At the same time it wants AI to become a real revenue line with business customers, and it confirmed its existing forecast and medium term targets.

The plan is unusually concrete for a corporate AI announcement. Telekom named targets for savings and revenue, and it backed them with numbers from systems that already run in its customer service and in its mobile network.

The money: savings first

The headline figure is the 2.5 billion euros. Telekom says that is the expected reduction in indirect costs by 2030, measured against 2023, coming from AI and automation. Indirect costs are the parts of a company that do not directly build networks or sell contracts: administration, customer service, internal IT and similar work.

There is also a nearer milestone. For 2027 the company expects about 1.1 billion euros in gross savings outside the US from AI and automation, again compared with 2023. Part of the extra savings in 2027 is supposed to be reinvested, partly into speeding up the digital transformation and partly into the fibre rollout in Germany.

Note that the two numbers are not measured in exactly the same way. The 2027 figure is gross savings outside the US, while the 2030 figure refers to indirect costs. Both are company targets, not results, and the press release carries the usual warning that forward looking statements may not come true.

The money: AI as a product

The second half of the plan is about selling AI, not only using it. Telekom expects around 250 million euros of AI related revenue with business customers outside the US in 2026. By 2030 it wants that number to reach about 800 million euros.

The bet behind that growth is what Telekom calls sovereign AI. Companies and public bodies in Europe want to use AI for sensitive work while keeping control of their data and of the systems that run it. Telekom wants to sell exactly that, by combining its own networks and European infrastructure with technology from partners. One building block is the Industrial AI Cloud in Munich, where companies can run industrial simulations and AI driven processes. For small and mid sized companies, Telekom is also building a platform to automate recurring tasks, for example in service or logistics.

Consumers are part of the plan too. The Magenta AI Assistant can translate mobile phone calls live, answer questions and summarise content, without an extra app or a new device. Telekom sees room for add on services and new tariffs there.

What already runs today

The most interesting part of the announcement is the set of numbers from systems that are already live:

• The chatbot "Frag Magenta" took over about 2.6 million customer service calls in the first half of 2026.
• In the US, customer service calls fell by 55 percent, and AI agents now handle 40 percent of customer contacts.
• The "RAN Guardian Agent" spots upcoming load spikes in the mobile network, for example around big events. Handling such events used to take several hours and now takes about one minute.
• In first use cases, AI that detects problems while a new connection is being set up cut complaints by 30 percent.
• More than 100,000 employees have been trained to work with AI. Staff get tools such as AskT, ChatGPT Enterprise and Microsoft Copilot under an internal programme called "AI for all".

Telekom also says AI helps developer teams write software and replace older IT systems, which is one of the less glamorous but most expensive problems at any large telecom.

What the CEO says

CEO Tim Höttges framed it as a change of the whole company. According to the press release (translated from German), he said AI is changing Telekom from the ground up: it makes the networks better and smarter, makes service more personal and opens new business models, "from the AI assistant in a phone call to sovereign infrastructure for Europe's industry". The decisive point, he added, is that technology creates measurable value for customers and for the company.

He also described the current state honestly as an early one: what works today in single applications should be used across the whole group. The way Telekom describes the new working model is that employees set the goals, check the results and keep responsibility for decisions, while AI does more of the repetitive work in between.

The market reaction was calm. According to heise online, the Telekom share barely moved in pre market trading on Tradegate compared with Friday's Xetra close.

Why it matters

Most AI news is about models, benchmarks and funding rounds. This is a different kind of story: one of Europe's biggest companies, with a huge customer service operation, putting a euro figure on what AI should save and earn, and tying it to systems it already runs.

It also shows where AI hits first inside large companies. The examples are customer service calls, network operations, software maintenance and administration, all areas with a lot of repetitive work. That is exactly where the savings come from, and it is also where jobs change the most. The 40 percent figure for AI agents in US customer contacts is a concrete signal of how fast that shift can go once the systems work.

For Europe, the sovereign AI angle is just as important. If European companies want AI for sensitive data without sending everything to US clouds, someone has to build and run that infrastructure locally. Telekom wants to be that someone, and the 800 million euro revenue target is the size of its bet.

Dany's take

I like that Telekom put real numbers on the table instead of the usual "AI will transform everything" slide. 2.6 million calls handled by a chatbot and AI agents taking 40 percent of US contacts are not lab results, they are daily operations. The open question for me is how customers feel about it. Fewer calls can mean problems get solved faster, or it can mean people simply give up on reaching a human. The 30 percent drop in complaints is a good sign, but it comes from first use cases, not the whole company. I will watch whether the 2027 savings milestone is actually reached, because that is the first real test of the 2030 promise.

Source: Deutsche Telekom press release: Telekom steigert Wachstum, Effizienz und Qualität durch Einsatz von KI. Also: heise online: Deutsche Telekom will durch KI-Einsatz Milliarden sparen.