AI is going to change the world like no technology before. Why am I so confident? AI is bringing a potent mixture of traits on to the table: it mimics the cognitive functions of humans and does so for incredibly low marginal costs. As I have outlined in the last post, this is the pattern of every big technology shift in history. At the beginning only, jobs are in danger, thereafter, we humans start flooding the world with waves of innovations just creating more jobs than ever.   

My favorite example of this mechanism? The aqueducts of ancient Rome. Of course, at the beginning, some water carriers lost their jobs. But once, the marginal costs of fresh water were actually zero, the Romans built an incredible sanitary infrastructure with drainage channels like Cloaca Maxima, shared toilets, and up to 1000 baths made for leisure.  d up to 1000 baths just for fun.  

The new technology did not only free human labor from carrying water, but it also improved the sanitary situation in the megapolis. Then, new powerful applications of water popped up: Street cleaning, domestic water supply, powering machinery, new manufacturing processes, better irrigation in agriculture, and even simulated naval battles for entertainment. 

Source: https://www.wissenschaft.de/geschichte-archaeologie/verbleites-trinkwasser-im-alten-rom/

We need to grow our whole AI ecosystem  

The effect of improving sanitation, wealth and life quality in ancient Rome by building aqueducts is what the ECO Verband called  “Spillover Effects” in its latest study. Of course, it does not specifically talk about ancient Rome´s sewage systems, but about AI infrastructure being the backbone of future economic growth in Germany. (“Spillover-Effekte von Rechenzentren: Rückgrat der KI-Revolution in Deutschland.”)  

Applying our lessons from history as well as from ECO Verband’s scientific research, we need to build AI data centers in Germany. What sounds simple as an idea, appears challenging when looking at the details. To build data centers and to create measurable spillover effects, Germany needs to invest into its overall AI ecosystem. Such a system isn´t just a bunch of specialized engineers building some halls and making GPUs work in them. Germany requires more effort along the whole value chain from physical resources (like soil, energy and water) over technology (like GPUs, network, and cooling) to regulation, capital and, eventually, humans and culture. 

From AI infrastructure to spillover effects for everyone

Together with the expert group “AI infrastructure” from the KI-Park, we have created an overview of all components belonging to a successful AI ecosystem:   

• Humans & Culture: Openness to new technologies, optimistic risk vs. opportunity assessment, ability to deal with ambiguity, experimental culture, adoption speed.
   
• Regulation: Sufficient clarity of regulation, room and incentives for innovation, acceptable risk for investors, coherence of federal levels, coherence of regulatory goals. 
  
• Technology: Access to relevant data, AI-using applications, access to foundational models, availability of chips, availability of data centers, reliable networks. 
  
• Physical Resources: Place for data centers, access to sufficient energy, access to cooling resources, access to data network, access to heat grid.  
  
• Capital: Venture capital for seed and growth phases of startups, public and private investments in AI infrastructure, incentives for research and development.  

1: Humans still remain in the centre of AI 

Even though we talk about Artificial intelligence, humans and the way they interact with each other stay in the center of the AI ecosystem. Specifically, we need the following kinds of experts: 

• AI Experts: Researchers, Data Scientists, Data Engineers, Software Engineers, Experts for Chips, hardware, network, security, and more. 
  
• Ecosystem Experts: Qualified experts supporting the core AI experts in legal matters, energy infrastructure, facility management, education, and more. 
  
• Leaders: Management Leaders, investors, and policy makers supporting the development of the AI ecosystem. 
  
• Consumers: End consumers and businesses demanding AI services, using them, creating competition, giving feedback and direction.

But the outcome of the AI ecosystem also depends on how these humans interact with each other. When a new technology says, “Hello World”, different cultures react differently. Some organizations and countries immediately see the risks, others focus more on opportunities. Some sit down and think, others touch and experiment. Some focus on themselves and start running, others also think of their peers, and what running would mean to them.  

A successful ecosystem with overall positive contributions would have the following cultural characteristics: 

• Openness & Optimism: Openness towards new technologies and the accompanying changes coming with them. Believe that society will profit from new technologies while risks will be managed over time.
  
• Agility & Speed: Hands on attitude across decision makers and engineers to experiment, to learn from failures, and to move forward. 
  
• Ethics & Inclusion: Ability to include all relevant stakeholders and their diverse views in the overall process, to make the right ethical decisions and anchoring them in solid institutions.  

What would be the biggest human and cultural lever to help Germany’s AI industry to flourish? We need a more opportunity-oriented mindset. In history, we have always profited when we adopted new technologies. Of course, AI technology will bring negative externalities, as we have seen when coal darkened the skies over the Ruhrgebiet during the Wirtschaftswunder. Today, with smart regulation, new technologies and the following structural change, life quality in Westfalen is better than ever. There shouldn´t be a reason to believe we will be able to manage the dark sides of artificial intelligence better just because we indulge in negativity.  

2: Making the potpourri of regulation manageable 

Even before the EU introduced its AI Act, there was a set of regulations in place affecting Germany´s AI ecosystem. There is GDPR regulating which data can be used for which purpose. There is industry-specific regulation applicable to where and how to operate applications and infrastructure in finance or utilities. There are export controls applicable to hard- and software, and there are regulations aiming for optimization of energy usage and CO2 emissions. Also building laws in general and construction rules coming from municipalities have major effects on the data center landscape. Eventually, even cybersecurity regulations apply to a certain extent, for example on data networks. 

Regulations usually follow good intentions. AI models shall behave responsibly, AI infrastructures shall be safe for everyone. Emissions shall be low, heat shall be reused, and data center buildings shall not disturb anyone. But what happens when one of the laws states, heat shall be reused for heating nearby homes, but local residents don´t want an industrial data center close by? Even worse: what happens, when energy is only available on the Northern Germany countryside (because of the wind), but there aren’t many homes to heat?   

The biggest regulatory lever to help Germany´s AI ecosystem flourish, would be a smart set of rules that is clear and coherent, and is easy to manage for investors and provides a clear incentive.  

A positive example? The BMWi once decided to optimize their set of rules for solar panels on balconies (“Balkonkraftwerke”) igniting an unprecedented growth of this sector since 2023.   

3: In terms of technology, mostly chips and model capabilities are missing 

Technology is a broad term. Talking about a healthy AI ecosystem, it specifically refers to: 

• Data: High-quality, diverse, and representative datasets, data governance and compliance frameworks. 
  
• Applications: Industry-specific AI solutions, general-purpose AI tools and platforms, model deployment and monitoring systems, continuous integration and delivery for AI. 
  
• AI Models:  Pre-trained large language models and other foundation models, APIs and interfaces for model access and fine-tuning, machine learning frameworks and libraries, AutoML and model optimization tools. 
  
• Chips: Specialized AI hardware (e.g., GPUs, TPUs, NPUs), edge computing devices for on-device AI.
    
• Data Centers: Cloud computing infrastructures, high-performance computing (HPC) facilities, specialized AI infrastructures. 
  
• Data Networks: High-speed internet connectivity, 5G and future network technologies (for edge) 
  
• Ecosystem Players: Academic-industry partnerships, open-source AI communities and initiatives, ecosystems for sharing and monetizing AI models and applications, collaborative platforms for AI development. 
  
• Cybersecurity: Robust security measures to protect AI systems and data, secure enclaves for sensitive AI workloads. 

Germany possesses deep expertise in many areas of technology. However, in two areas most relevant for the AI ecosystem, we cannot compete with globally leading companies:

• Software Scale-Ups: There is no German company able to keep up with American providers of Large Language Models such as OpenAI, Meta, Google, and X. We also lack a significant number of large-scale software companies integrating these models in their standard services and wrapping them with business models (like Microsoft365 or GitHub Copilot).  
  
• Chips Value Chain: A few German companies are particularly strong in semiconductors for automotive and industrial applications (like Bosch). Still, our contribution to the high-end versions of chips is rather small (i.e. Zeiss producing mirrors for lithography). 

Pulling both levers would have significant positive effects on the German AI ecosystem. More companies developing and using the latest AI models would significantly boost the demand for AI infrastructure, whereas more contribution to the global semiconductor value chain would help Europe to have better access to chips.    

4: AI is a lot about Physical Resources  

A lot of components of an AI ecosystem are virtual: Algorithms, models, services, regulation, cybersecurity, to name some. The basis for all of this is still physical though:  

• Location: Large-scale facilities to house computing hardware, distributed data centers for improved latency and redundancy 
  
• Building Management: Know-how to build data centers, resources & knowledge to maintain them. 
  
• Energy Resources: Access to sufficient and reliable energy supply, renewable energy sources for sustainability. 
  
• Cooling Systems: Efficient cooling solutions for data centers, potential for liquid cooling technologies.
  
• Network Access: High-speed, low-latency data networks, robust internet backbone and international connectivity. 
  
• Heat Utilization: Access to heat grid for repurposing waste heat, district heating systems integration.
   
• Security: Physical security of data centers, fire alarm and security technology up to smart building functions such as energy management, room control, bookings, and access management. 

The single biggest positive impact for Germany´s AI ecosystem would be having better access to electrical energy.  

5: Money makes AI go round

Beneath almost all the above-mentioned components, there is one facilitating factor: The supply of capital. Financial investments are necessary in almost every single part of the AI ecosystem: 

• Basic Research: Supporting researchers in semiconductors, algorithms, power generation, energy grids, and more. 
  
• Education: Driving AI readiness across industries, their workforces, and consumers. 
  
• Venture Capital: Providing seed funding to innovate and series A/B/C funding to scale globally. 
  
• Transformation Funds: Supporting critical yet existing industries with funding to transform their business models without too many negative effects on employment. 
  
• Strategic Subsidies: Geopolitically relevant industries are often strongly subsidized by other continents and countries (e.g. solar in China). In these cases, additional financial support is required to keep national players alive. 

Especially capital-hungry yet powerful are the following three levers:  

• Energy Supply: Creating new, green yet reliable sources of energy to power the necessary AI infrastructure. 
  
• Semiconductors: Enhancing Europe´s contribution to the high-end semiconductor value chain. 
  
• Software Scale-Ups: Growing Germany´s software start-ups focussing on globally competitive models and business models integrating them. 

7 Measures to Create Spillover Effects With AI system 

Ancient Romans started building aqueducts leading to an increase in productivity and wealth. How can we create such spillover effects also with AI as our latest technology leap? I see five major levers we should pull: 

1. Ramping-Up Energy Supply: Increase clean energy production to power the growing demands of AI infrastructure and data centers.
    
2. More AI-Software Scale-Ups: Foster an environment that encourages the growth of AI startups and helps them scale rapidly to compete globally. 
   
3. Change to Opportunity-oriented Mindset: Shift focus from potential risks to the transformative opportunities AI presents for economic growth and innovation. 
   
4. Increase Europe’s Contribution to Semiconductors: Invest in domestic semiconductor design and manufacturing capabilities to reduce dependence on foreign suppliers and support AI hardware development. 
   
5. Simplifying Regulation: Streamline AI regulations to along the whole value chain in order to ensure the overall growth of our AI landscape. 

Chart above shows all 5 suggestions, but reveals two more fundamental components: 

6. Capital: Scaling software companies, ramping-up energy supply, and contributing more to semiconductors require enormous access to capital.  
   
7. Dedication & Grit: Ramping-up energy supply, contributing more to semiconductors, changing a national mindset, and simplifying regulations require enormous dedication and grit to get done. 

To conclude, Germany and Europe stand at a pivotal moment in the development of its AI ecosystem. By focusing on key areas such as energy infrastructure, software innovation, mindset shift, semiconductor production, regulatory simplification, capital investment, and overcoming complexity, the continent can position itself as a leader in the global AI landscape.  

Much like the aqueducts of ancient Rome, these efforts will not only drive technological advancement, but create significant spillover effects, leading to increased productivity, economic growth, and improved quality of life for all citizens. Even though the journey ahead may be challenging, we can harness the transformative power of AI to build a flourishing, innovative, and prosperous future with concerted effort and strategic investment.