AI – growth driver or glorified pocket calculator?

The expectations for it are high. But despite billions in capital expenditure the economic payoff remains unclear. Until AI delivers tangible return on investment, shareholder confidence will waver. So, where exactly do we stand today – and what needs to happen for AI to realize its true economic value?

In Silicon Valley, artificial intelligence has long been seen as a miraculous cure-all, comparable with the invention of the steam engine or even the discovery of fire. But all this euphoria begs the question: How and when will the effect of AI be reflected in economic progress? 

Microsoft CEO Satya Nadella cuts straight to the core in a recent podcast: the ultimate measure of artificial intelligence isn’t beating humans at Go or solving intricate equations – it’s whether it can supercharge the global economy. As he puts it, “The real benchmark is the world growing at 10%.”

To put that into perspective, global GDP – the sum of every product made and service delivered, from haircuts and building homes to lines of code – has historically grown at a crawl. Before the Industrial Revolution, growth was virtually nonexistent. Even during that seismic shift, the UK managed just 0.5% to 1.5% annually. Today’s developed nations average 2% to 3%. Yes, countries like China once hit 10% during rapid development, but Nadella is pointing to something far bolder: a transformation so profound it rewrites the limits of global productivity itself.

So far, though, the promised productivity boom remains elusive. Despite billions poured into ever-larger AI models, clear economic returns are still hard to pinpoint. The biggest beneficiaries to date aren’t the AI builders themselves, but the infrastructure players – the makers of GPUs like NVIDIA, and cloud giants such as Microsoft, Google, and Amazon, who supply the compute muscle behind the scenes. For now, it’s the suppliers of picks and shovels who are cashing in, not the gold miners. Monetization remains an open question, especially as users grow accustomed to free tools. What sets the current leaders apart is not that they’re profiting directly from AI just yet – but that they can afford to keep playing.

Despite all this, skepticism may be premature. After all, great technological revolutions often need time to take full effect. The steam engine changed the world – but not straight away: Productivity only increased measurably decades after its invention. The same also applied for electricity: Its economic impact only came once entire factories, cities and infrastructures had been connected to the grid. 

Perhaps AI is at exactly this point: The technology exists, the potential is huge, but it first needs to have a widespread impact. Analysts at Goldman Sachs and J. P. Morgan already expect a turning point – and believe that AI could realistically drive global GDP growth of up to 9% in the next ten years. 

Where AI is already unleashing revolutions in efficiency 

Yet even without noticeable economic growth, AI is already radically increasing efficiency. It analyzes enormous quantities of data in a fraction of the time a human would need. In the field of medicine, AI identifies illnesses, tests potential cancer medicines and helps develop new antibiotics. It solves complex equations a thousand times quicker than any human and can even simulate the creation of galaxies in simplified models – in minutes rather than years. 

At our company, we already feel the effects of this efficiency revolution every day and at every level. Our in-house knowledge management application myGPT suite alone saves employees 3,600 working hours every week. Our employees are trained specifically to integrate AI optimally into their everyday working lives – from support during routine tasks, all the way up to research and development.  

To put it simply, the Electronics business sector of Merck KGaA, Darmstadt, Germany uses AI for applications including the development of new, highly specialized materials that make semiconductors faster, more efficient and more heat-resistant. In addition to simulating material properties, it also conducts experiments virtually. For instance, this enables us to test how a material behaves at 500 °C, whether it reacts with other substances or to what degree of purity it can be manufactured – and all before it is even mixed in a laboratory. Or, as our researcher once explained to me: “In the past, looking for new materials was like fishing – you never knew exactly what you would catch. Today it is like a deep-sea expedition in a high-tech submarine.” 

Why efficiency does not necessarily lead to more growth 

One obstacle has long stood in the way of overall growth, however: AI solutions are usually aimed at large corporations. Small companies that would gain the greatest benefit from AI often possess neither the expertise nor the resources to use the technology productively. An AI solution that generates PowerPoint slides will have little effect on GDP if nursing staff still have to fill out paper forms. 

Nobel prize winner and economist Daron Acemoğlu therefore believes that only around 5% of work processes in the United States can be profitably taken over by AI in the near term. He concludes that AI will increase global GDP by just 1% to 1.5% in the next decade – a far cry from the optimism of the major banks. 

Will AI really usher in a new era of economic growth? Or are we witnessing vast amounts of money being invested in a technological plaything that will never pay off? The fact is, AI is changing our world, but economic growth requires more than just neural networks – it needs time, structural change and clever implementation. 

Europe’s opportunity: tailored solutions and ethical standards 

There’s no sugarcoating it – when it comes to the development of generic large language models such as GPT-40 or DeepSeek R1, Europe is lagging behind the United States and China. So far, Europe has also played a minimal role in key technologies such as GPU design or cloud infrastructure. The direct economic benefit is small as a result of this. 

However, the area of specialized AI harbors enormous potential for Europe. After all, this is not about building the largest or cleverest AI solution, but the most suitable and most integrated one. 

Whether in manufacturing, health, the energy sector, or the public sector, European companies and institutions possess in-depth specialist expertise, still offer a very strong industrial sector and have complex and often strictly regulated fields of application. This is precisely where bespoke AI solutions can offer genuine added value. The success factor for a successful application is not just the one and only right model. Only by combining it with specialist knowledge and the company's proprietary high-quality data, supported by AI, it becomes a competitive advantage. Successful and sustainable solutions are created at this interface between people, data and AI. 

Another special opportunity lies in the automation of back-office processes – in fields including law, finance, strategy, HR, communication, and IT. Here in particular, Europe can increase its productivity, counteract the shortage of specialist staff and remain competitive in the long term. “AI made in Europe” could become a byword for well thought-out, efficient and responsible applications. 

The emphasis here lies on “responsible”, as in Europe in particular there is a great deal of interest in how the decisions of algorithms can affect each one of us. What are the social consequences of automation? And how do we reconcile technical progress with data protection and the environment? All this can make Europe look slow and ponderous at first glance, but those who set standards in these areas not only shape the market, but also the values that will underpin the digital future. 

It is likely that future generations will not judge our actions not only by whether we succeeded in growing global GDP with AI, but also by how responsibly we used this powerful new technology.