Deutsche Telekom puts euro figures on AI

At a glance: On 5 October 2026, Deutsche Telekom set out how AI is changing its business. It expects AI and automation to deliver gross savings of about €1.1bn outside the US in 2027 and about €2.5bn of indirect-cost savings by 2030, both against 2023. In the US, AI agents handle 40% of customer contacts and service calls have fallen 55%. More than 100,000 employees have been trained in AI. For CEOs, it is a template for reporting AI value in money.

The news

Deutsche Telekom used an AI investor day in Bonn on 5 October 2026 to present its AI applications to the market. It confirmed its forecast and medium-term targets and then attached numbers to AI on both sides of the ledger.

On costs, the company expects AI and automation to generate gross savings of approximately €1.1bn outside the US in 2027, compared with 2023. By 2030 it aims for savings of approximately €2.5bn in indirect costs, also compared with 2023. Part of the 2027 savings will be reinvested in digital transformation and fibre expansion in Germany.

On revenue, it expects approximately €250m of AI-related revenue from businesses outside the US in 2026, rising to approximately €800m by 2030.

Chief executive Tim Höttges set the tone: "AI is fundamentally transforming Deutsche Telekom." He added: "What matters most is that technology delivers measurable value, for our customers and for our company."

The operating evidence

What makes the presentation more useful than most corporate AI announcements is the operating detail behind the targets.

• Customer service in the US. Customer service calls have dropped by 55%, and AI agents now handle 40% of customer contacts.
• Customer service in Germany. The "Frag Magenta" chatbot handled about 2.6 million customer service calls in the first half of 2026.
• Network operations. A "RAN Guardian Agent" detects impending strain on the mobile network and supports countermeasures. Telekom says response time to such events has dropped from several hours to about one minute.
• New connections. Initial AI use cases in setting up new connections show a 30% decrease in complaints.
• People. More than 100,000 employees have been trained in using AI. Under an "AI for All" initiative, staff have access to tools including AskT, ChatGPT Enterprise and Microsoft Copilot.

On the product side, Telekom pointed to its Magenta AI Assistant, which translates phone calls in real time, answers questions and summarises content without a separate app or device; an Industrial AI Cloud in Munich for industrial simulations and AI processes; a platform under development for small and medium-sized businesses to automate recurring tasks; and a "Magenta Moments" rewards programme that uses AI to make offers more relevant.

The WCRC Review: what Telekom gets right

Many large companies talk about AI in terms of pilots, licences bought or employees enrolled in training. Those are inputs. Telekom's presentation is notable, in WCRC's assessment, for leading with outputs that a finance director can test: euros saved against a stated baseline year, a share of customer contacts handled by agents, a percentage fall in calls, a response time measured in minutes.

Four features stand out.

A fixed baseline. Both savings targets are measured against 2023. That stops the goalposts moving and makes later progress reports comparable. Many AI business cases quietly change their reference point as programmes evolve.

Separate cost and revenue lines. Telekom gives savings and AI-related revenue as distinct numbers with distinct horizons. That avoids the common habit of blending efficiency and growth into one headline "value" figure that is hard to audit.

Operational metrics under the financial ones. Calls, contacts, complaints and response times are the mechanisms by which savings arrive. Reporting them lets investors and boards see whether the financial targets are on track before the money shows up.

A stated use for the savings. Saying that part of the 2027 savings will fund digital transformation and fibre build in Germany links AI to the core investment case. It also answers a question every workforce asks: where does the efficiency go?

Höttges made the scaling ambition explicit: "We want to roll out across the entire company what's working today in individual applications."

Where to read with care

A review should also weigh what the numbers do not yet show.

These are targets. The €1.1bn and €2.5bn figures are expectations and aims, not booked results. Delivery will depend on execution over several years.

Scope differs by figure. The 2027 gross savings and the AI revenue figures are stated for businesses outside the US. The most striking service metrics, the 40% agent share and the 55% fall in calls, come from the US. Readers should not combine them into a single group-wide picture.

"Gross" and "indirect" matter. Gross savings come before reinvestment and implementation costs. The 2030 target covers indirect costs specifically. Neither is the same as a net profit uplift.

Customer outcomes need tracking too. Fewer service calls can mean problems solved earlier, or customers giving up. Telekom's 30% fall in complaints in early new-connection use cases is a helpful counterweight; boards adopting a similar model should insist on satisfaction and resolution data alongside volume reductions.

What other CEOs can borrow

The following are WCRC suggestions drawn from Telekom's approach.

• Fix a baseline year for every AI savings target and keep it.
• Report AI cost savings and AI revenue separately, each with its own timeline.
• Pair each financial target with two or three operating measures that move first.
• Say publicly where savings will be reinvested.
• Train widely, but report training as an enabler, not as the result.

What it means for leaders

• Ask your AI programme for euro or dollar targets against a fixed baseline year, not just usage and pilot counts.
• Split efficiency and growth: separate numbers, separate owners, separate deadlines.
• Track leading operational indicators, such as contact share handled by agents or incident response times, to catch slippage early.
• Balance volume metrics with customer measures such as complaints and resolution rates.
• Be clear with staff and investors about what happens to the savings.

Quick answers

Q: How much does Deutsche Telekom expect to save from AI? A: It expects AI and automation to generate gross savings of about €1.1bn outside the US in 2027 and aims for about €2.5bn of indirect-cost savings by 2030, both compared with 2023.

Q: How are AI agents used in Deutsche Telekom's US customer service? A: Deutsche Telekom says AI agents in the US now handle 40% of customer contacts, and the number of customer service calls has fallen by 55%.

Q: How many Deutsche Telekom employees have been trained in AI? A: More than 100,000 employees have been trained in using AI, with tools including AskT, ChatGPT Enterprise and Microsoft Copilot available under the "AI for All" initiative.