In the podcast "Diagnose: Zukunft", Gottfried Ludewig speaks with Maximilian Greschke, CEO of Recare, about digital sovereignty in healthcare—and why it is not the models but the available computing power that is currently the real bottleneck.
The healthcare sector is facing profound change. Artificial intelligence (AI) can accelerate diagnostics and medical history taking, simplify documentation, and relieve the burden on medical staff. Even more decisive, however, is its potential to analyze large volumes of data from sensors, devices, and treatments. This allows us to recognize earlier when health transitions into illness—and to rethink prevention.
This will change the entire care system in the long term: the role of physicians, the organization of hospitals, and the question of when inpatient treatment is actually necessary. If medications are developed more quickly and risks are identified earlier, many hospital stays can be avoided in the future. This is not a distant vision, but a development we must prepare for today.
At the same time, aspiration and reality are still far apart. In many clinics, modern AI applications encounter software that is 15 or 20 years old, missing interfaces, and processes that still partly rely on paper. Digitalization has often brought only marginal improvements in recent years. AI now possesses, for the first time, the potential to solve precisely those structural problems that have caused so many projects to struggle.
The biggest hurdle is not technology alone. People initially follow their established patterns. That is why we need to create practical use cases that prove their value in everyday life. Those who experience firsthand how AI makes their own work easier develop curiosity and a willingness to change. This applies to hospitals just as much as to companies.
The second prerequisite is a high-performance technical foundation. AI applications require computing capacity, secure data processing, and a resilient cloud and integration architecture. A single isolated solution is not sufficient. What matters are open, scalable platforms that integrate existing systems while simultaneously opening the path to a new digital way of working.
Moreover, we must leave an outdated paradigm behind. For over 20 years, the healthcare sector has attempted to create a fully structured data foundation. This has been a failed experiment. Not because structured data is unimportant, but because the notion that people must first capture and standardize all information perfectly has not worked in practice.
Modern AI models are increasingly able to handle unstructured data—language, findings, documents, and information from existing systems. This reverses the order: we no longer need to wait for perfectly structured data. AI can extract insights from the available information and then take on the necessary structuring itself. This opens up a more realistic and significantly faster path to digitalization.
The actual bottleneck is increasingly not the models, but the available computing capacity. At Deutsche Telekom, we built the Industrial AI Cloud with 10,000 GPUs in Munich within just a few months at the beginning of the year. This increased AI computing capacity in Germany by around 50 percent in one fell swoop. And yet this capacity is already almost sold out. This shows: AI computing capacity is not an abstract question for the future. It is needed today—for research, diagnostics, simulation, robotics, and industrial applications.
Digital sovereignty therefore means very concretely: Europe must have sufficient computing capacity, secure cloud platforms—such as our T Cloud—and powerful networks. It is not about avoiding technological dependencies entirely. It is about remaining capable of action, having choices, and being able to control data, processes, and critical applications even under changing political and economic conditions. This also includes competitive electricity prices and suitable framework conditions for operating data centers. This is also crucial for the healthcare sector: clinics are increasingly dependent on a reliable digital infrastructure—and at the same time, face further major challenges.
AI will not simply digitalize the healthcare sector. It will reorganize its structures, processes, and roles. Our task is to actively shape this transformation—so that technology not only increases efficiency, but above all improves the care of people.
Because clinics are currently under pressure from three directions. Demographic change is leading to more patients and fewer nursing staff and doctors. At the same time, financial pressure is increasing due to necessary savings and structural changes in the healthcare sector. Added to this is a new threat landscape in the area of cybersecurity. Where previously weeks were sometimes available for a security patch, the time window has now shrunk to just a few hours.
These challenges cannot be overcome with individual measures. AI must therefore not be a task that is delegated exclusively to the IT director. It must be a matter for senior management. The leadership of a hospital must understand which processes are changing, which data is required, which risks arise, and which technical prerequisites need to be established.
Clinics need a bridge between the old and the new world. Day-to-day hospital operations must function today—even when systems are outdated, interfaces are incomplete, and data is still partially available on paper. At the same time, new applications must be integrated.
We also need to rethink our approach to investments. When we talk about the Hospital Transformation Fund, it must not be solely about beds, buildings, and concrete. That would be thinking from the 1990s. Of course, locations and a functioning structural infrastructure are still needed. But investments should equally enable new forms of care: telemedicine, the consolidation of data, a 360-degree view of patients, and the targeted use of AI.
AI will operate not only in central data centers, but also in time-critical settings, such as emergency rooms, where on-device AI can play an important role. Processing data on-site can reduce latency and provide greater control over sensitive information. Regional mini data centers will also make applications available closer to where they are needed.
The future cannot be fully designed on the drawing board. We should start boldly, create concrete use cases, and learn from practice. It will now be decisive to create the conditions for change: with sufficient computing power, sovereign cloud and data platforms, secure networks, and a shared will to shape the future.
AI will not simply digitalize healthcare. It will reorganize its structures, processes, and roles. Our task is to actively shape this transformation—so that technology not only makes things more efficient, but above all improves the care of people.
I spoke about all these topics in the podcast "Diagnose: Zukunft" (in German) with Maximilian Greschke, CEO, Recare:
You can display all external content on the website at this point.
I agree that personal data may be transmitted to third-party platforms. Read more about this in our privacy policy.
