Deutsche Telekom (DTE) investor relations material

AI strategy and operating model

• AI is shifting from tool adoption to enterprise redesign across processes, teams, networks and customer products, building on 500+ live internal projects and prior digital foundations.

• Delivery combines shared data and AI platforms, product-led major bets and business-led adoption; business owners are accountable for adoption, workforce impact and ROI.

• Trusted, sovereign AI relies on frontier-model partners, a flexible stack and switchable tools and providers to limit lock-in and manage ROI; humans retain responsibility for decisions.

• Employee AI use reached 83%; more than 250,000 AI enablements have taken place since 2023, and 80% of surveyed employees report substantial productivity improvements.

• Workforce planning shifts from roles to tasks, identifying work to keep human-led, augment or automate while supporting responsible transformation and new skills.

Financial targets and investment

• AI and automation gross savings target for 2027 rose to ~€1.1bn from ~€0.8bn at the 2024 CMD; token costs are excluded from the savings basis.

• By 2030, AI and automation target ~€2.5bn in indirect cost savings versus 2023, equivalent to ~15% of IDC AL; 2026 and 2027 milestones are ~4% and ~6%.

• 2030 gross savings ambitions versus 2023 include 10–15% in Network, 15–20% in IT, 20–30% in Sales & Service and ~10% in G&A/other.

• Savings support reinvestment in digital transformation, network build-out and German fiber expansion; the updates do not change current guidance.

• Token costs are managed through model tiering, cascading, flexible providers and cost-center budgets, and are expected not to exceed a low double-digit percentage of gross savings.

Network transformation

• 2030 ambition is Level 4 autonomy for high-value use cases, with human intervention by exception; mobile CXI is targeted above 4 for more than 90% of customers.

• Network targets include automatically handling 90% of tickets above 1,900, qualifying issues in under 15 minutes on average, resolving incidents in under one hour and achieving 30% productivity gains.

• RAN Guardian expanded event coverage from 1,000 to 100,000, cut execution time by more than 95% and helped protect more than 24m customer minutes from congestion.

• MINDR combines multi-agent diagnostics, remediation and learning; a German MVP is planned by year-end, with critical network changes retaining human approval and guardrails.

• Fiber AI efficiencies are expected to deliver €100m–€150m next year; AI-enabled construction management reports 100% progress transparency, 15% higher construction quality and correction of 20% of false invoices.