0

Deutsche Telekom is putting financial targets behind its AI strategy, but the more consequential question is whether it can capture incremental revenue or simply reposition existing businesses around enterprise AI spending

Telecommunications operators have spent years describing how artificial intelligence (AI) will transform their networks, operations and business models. Deutsche Telekom is taking a more concrete approach, attaching financial targets to its ambition to become an AI-native telco.

At its October 5 AI Investor Day, CEO Tim Höttges outlined a goal of more than €1 billion in annual AI-related revenue outside the United States by 2030, alongside €2.5 billion in gross operating savings. “To be honest, this is a conservative number,” Höttges said of the revenue opportunity, suggesting internal planning scenarios are more optimistic.

The strategy raises a fundamental question about telecom’s role in the AI economy. Is DT establishing a genuinely new revenue pool, or leveraging existing infrastructure, customer relationships and enterprise technology capabilities to capture spending migrating toward AI? The answer is likely some combination of the two. More importantly, the distinction will determine how much incremental economic value the strategy ultimately creates.

DT sees enterprise AI as an €800 million opportunity

DT’s most substantial and clearly defined revenue opportunity resides in its business-to-business operations. The company expects approximately €250 million in AI-related enterprise revenue in 2026, growing to €800 million by 2030, roughly 34% compound annual growth.

The proposition combines sovereign AI infrastructure, industry-specific solutions and packaged AI applications for smaller businesses. Central to the infrastructure strategy is the Industrial AI Cloud in Munich, developed with NVIDIA and other partners. With approximately 10,000 GPUs, the facility supports workloads including industrial simulation, digital twins, robotics and AI-enabled manufacturing processes.

DT is evaluating additional capacity, including a potentially much larger European AI gigafactory, but management emphasized that expansion depends on validated demand, utilization and appropriate financing structures.

Beyond infrastructure, T-Systems is positioning itself to integrate AI into existing enterprise technology environments. The IT services and consulting subsidiary is also developing an agent platform targeting Germany’s Mittelstand, packaging capabilities such as customer-service automation, sales support and logistics optimization for businesses without extensive internal AI resources.

The strategy draws directly on DT’s existing commercial footprint. “We can upgrade these customers,” Höttges said, referring to a B2B business generating approximately €15 billion in annual revenue.

That statement captures both the opportunity and its principal analytical complication. DT already supplies connectivity, cloud infrastructure, security, integration and managed services to enterprise customers. AI provides opportunities to expand those relationships, but some resulting revenue may represent substitution for traditional ICT products and services rather than genuinely incremental spending.

Sovereignty as differentiation, not a new business model

DT is explicit about where it believes it can compete in the AI value chain. “Our place is where trust matters,” Höttges said, emphasizing data sovereignty, security and established customer relationships rather than direct competition with frontier-model developers or hyperscale cloud platforms.

That positioning is particularly relevant in Europe, where regulated industries and public-sector organizations increasingly prioritize data residency, operational control and compliance. It is also consistent with DT’s established enterprise strategy. Sovereign cloud, secure connectivity, systems integration and managed infrastructure were meaningful businesses before generative AI accelerated investment in computing infrastructure.

AI substantially expands the potential market for these capabilities, but does not fundamentally change their underlying competitive dynamics. DT must still compete against hyperscalers, European cloud providers, systems integrators and enterprise software vendors. Sovereignty may provide differentiation in certain customer segments, but sustainable returns will depend on utilization, service margins and the ability to deliver integrated solutions more effectively than competitors.

The investment case goes beyond the trend that enterprise AI spending will grow; it is that DT can capture a profitable share of that spending by combining infrastructure and capabilities that already exist within the business.

The cannibalization question

This distinction surfaced explicitly during the Investor Day’s question-and-answer session. Bernstein analyst Ulrich Rathe asked how investors should evaluate the €800 million AI revenue ambition against potential cannibalization of DT’s legacy IT services.

T-Systems CEO Ferri Abolhassan acknowledged the substitution risk. “The customer will not look for the old linear programs anymore,” he said, explaining that AI-based automation can displace traditional enterprise software and implementation projects.

But Abolhassan argued that T-Systems’ integrated AI stack creates a competitive advantage, characterizing the transition as a new unique selling proposition rather than current cannibalization. 

If AI allows DT to win new customers, expand its share of enterprise IT budgets or deliver higher-value managed services, it represents incremental growth. If AI primarily replaces existing integration, automation and outsourcing revenue, the benefit depends on whether those replacement services generate better margins or improve retention.

The €800 million target is meaningful, but without greater disclosure around revenue composition, substitution and profitability, it cannot by itself establish the net financial contribution of AI.

Consumer AI needs distribution but is the monetization guaranteed? 

Beyond enterprise services, DT expects consumer AI to contribute to its broader ambition of more than €1 billion in annual AI-related revenue and financial benefits. That contribution could combine directly monetized offerings, such as the Magenta AI Assistant, with indirect benefits from lower churn, more effective marketing and improved pricing power. Unlike the €800 million enterprise target, management has not provided a separately quantified consumer AI revenue forecast.

The opportunity is to create additional service tiers, strengthen customer relationships and potentially capture value as AI becomes a more prominent interface between people and digital services. However, the commercial model remains less developed than the enterprise proposition.

There is also a structural challenge. Device manufacturers and software platforms increasingly integrate AI functionality into their existing offerings, potentially limiting customers’ willingness to pay their connectivity provider separately. Improved retention, lower churn and targeted upselling may ultimately contribute more value than standalone AI subscription revenue.

Efficiency is perhaps the more immediate economic return

While new revenue attracts attention, DT’s savings ambitions may be more consequential in the medium term. The company expects €1.1 billion in annual gross savings from AI and automation by 2027, increasing to €2.5 billion by 2030, measured against a 2023 baseline. Applications span customer service, software development, network operations and administrative workflows.

DT’s RAN Guardian Agent provides a concrete example. The system anticipates congestion around large events and coordinates corrective actions across network resources. According to DT, the AI-enabled workflow has reduced response times from several hours to approximately one minute, while enabling the network team to analyze more than 100,000 events annually.

This reinforces the conclusion from the recent Intelligent RAN Forum coverage. AI’s most measurable network returns currently reside in improved efficiency and operational performance rather than entirely new services.

Importantly, DT’s €2.5 billion target represents gross indirect cost savings, including capitalized labor and excluding AI inference costs, rather than an equivalent increase in earnings or cash flow. Some savings will also be reinvested. CFO Christian Illek outlined mechanisms for managing token expenses, including model selection, usage budgets and internal cost accountability. DT aims to limit token costs to a low-double-digit percentage of gross savings, underscoring the importance of evaluating AI’s net economic contribution.

Capital discipline also extends to anticipated network demand. As RCR Wireless News recently reported, DT does not expect AI-driven mobile uplink growth to require incremental RAN capital expenditure in the medium term. That contrasts with its longer-term vision of distributed inference and physical AI applications creating new requirements for connectivity, latency and network orchestration. As board member Rodrigo Diehl explained, some emerging workloads will require intelligence closer to users and devices.

But the financial case for that infrastructure remains prospective. Notably, DT’s €800 million enterprise AI forecast excludes potential revenue from physical AI and a future gigafactory.

Capturing value from the AI transition

DT’s strategy is neither a wholesale reinvention of the telecommunications business nor simply the relabeling of existing services. It is an effort to use established infrastructure, customer relationships and operational expertise to capture value as enterprise spending shifts toward AI, while developing longer-term opportunities in consumer services and distributed computing.

The near-term investment thesis is strongest where DT can demonstrate measurable returns from existing capabilities around enterprise ICT services, sovereign infrastructure and operational automation. The more transformative opportunities remain less certain, particularly where monetization depends on emerging applications or changes in how customers consume connectivity.