Trump is experienced in running things. If wind power would be worth something Trump wouldn't put it into trash.
“Artificial intelligence is increasingly becoming a key source of hope for Infineon. CEO Jochen Hanebeck explains where he sees Europe’s opportunities in AI, how his company can help curb energy consumption in data centers—and why not all robots need legs.
Mr. Hanebeck, Infineon is currently benefiting from the excitement surrounding artificial intelligence. You manufacture power semiconductors designed to enable efficient power supply in data centers. Should we imagine it this way: that there is hardly an AI data center out there without Infineon chips?
It would likely be difficult to find one where we aren't present. Against this backdrop, the timing is also perfect for our five-billion-euro investment in our new chip factory in Dresden. The first equipment is currently being installed, and we will officially begin production this summer. Here, we will manufacture analog and power semiconductors for AI applications, and later on, chips for the automotive industry as well.
The business involving these chips for AI data centers is growing rapidly, while your other segments are currently lagging. Does this mean that, without this 'AI effect,' you would be facing a real problem right now?
No. We always have several irons in the fire. Five years ago, no one could foresee the rapid development of AI applications. However, it was already clear back then that the power requirements of future processors would rise. Consequently, we began building up our expertise early on. When it subsequently became clear just how powerfully and quickly the AI trend—and high-performance AI chips—were emerging, it wasn't particularly difficult for us to translate those capabilities into products. Today, this is our fastest-growing business segment—growing at a pace I haven't witnessed in my 30 years at Infineon.
There is currently a great deal of discussion regarding the risk of a potential 'AI bubble.' How high do you estimate that risk to be?
We are dealing here with a general-purpose technology that possesses broad application potential—and is backed by a very real business.” I need only look at my own company to see this.
AI is delivering tangible productivity gains for us—for instance, in customer support, software development, and the creation of chip layouts.
Can you put a figure on those time or cost savings?
In chip design, it can mean accelerating the work process by weeks. Another example is "Ask Infineon"—an AI agent we use internally to handle employee inquiries, which has been trained using relevant information from within the company. Our HR department currently receives 30 percent fewer inquiries than before, as employees are able to find the information they need themselves via "Ask Infineon."
Nevertheless, there are many voices sounding the alarm and pointing to warning signs—for instance, transactions between companies like OpenAI and Nvidia that create the impression that money is simply being shuffled back and forth.
Naturally, financing is always a hot topic in sectors experiencing a boom. However, most of the key players in this market are able to fund their investments using their own cash flow. Furthermore, from my perspective, there is a completely different aspect to the AI debate that has largely gone overlooked until now: In many regions of the world, we are seeing a growing demand for what is known as "sovereign AI." Governments are coming to the conclusion that AI infrastructure is just as integral to a nation as any other form of infrastructure. This includes China and the EU, among others, as well as countries in the Gulf region. Their motivations differ significantly from those of companies that merely operate large-scale data centers.
US President Donald Trump recently announced the "Genesis Mission"—a national AI initiative. Does this serve as a model for Europe?
It is a comprehensive AI research program. It demonstrates that policymakers recognize the added value of artificial intelligence and are taking appropriate action. Artificial intelligence is a foundational technology that benefits a wide range of economic sectors. In the coming years, it will play a significant role in our lives—taking many different forms and impacting a vast array of fields.
Can Europe keep pace? After all, the most important companies are located in the USA.
That is certainly true. On the other hand, we also possess capabilities and expertise here in Europe. We need to approach this differently, and we should not succumb to the temptation of trying to copy the USA one-to-one.
What does that mean in concrete terms?
Let’s start at the level of AI models. I see opportunities here within the EU to develop industry-specific models—that is, beyond the realm of large language models. Europe also has a great deal to offer at the chip level. Our power semiconductors are one example of this, as are microcontrollers—a product category in which European providers hold a 50 percent share of the global market. Microcontrollers play a crucial role, for instance, in Edge AI—that is, when data is processed locally rather than in the cloud. This is often more efficient, faster, and more secure. Another flagship for Europe is the lithography technology developed by the chip-machine manufacturer ASML—without which no AI chip in the world could be produced. Europe must forge its own path, build upon its existing strengths, and act decisively to avoid becoming unilaterally dependent in these high-tech fields.
So, you believe Europe already possesses significant assets it can leverage in this regard?
Yes, absolutely. It is far better to build upon our own strengths and, on that foundation, collaborate with partners who possess different strengths. This includes manufacturers of AI processors in the United States.
The energy aspect of AI is playing an increasingly prominent role in public discourse, as data centers consume enormous amounts of electricity. In the U.S., reports suggest that in regions with a high concentration of data centers, electricity bills for consumers are rising; in some places, resistance to the expansion of AI infrastructure is beginning to emerge. Could this put the brakes on the industry?
Precisely for this reason, it is crucial to tap into additional energy sources as quickly as possible. In my view, in the near future, this cannot be achieved solely through the construction of additional gas-fired power plants—or even nuclear facilities. The simplest and most obvious solution lies in renewable energies. Wind farms and solar installations—particularly when paired with battery storage systems—can be scaled up quite rapidly.
However, that does not appear to be the U.S. government’s preferred path at the moment. Donald Trump is an avowed opponent of wind power and has even expressed a desire to halt such projects.
Certainly, there is a specific political agenda at play in the United States. However, from an industry perspective, I can state with confidence that renewable energies are poised for a new boom—even here in the Western world. This is driven by the very fact that we urgently need to generate additional electricity-generating capacity—and quickly.
Trump, however, doesn't seem to have any objections to coal.
I would prefer not to comment on American politics. Naturally, we are currently observing regional disparities in the development of renewable energies. China is betting heavily on this—as is Europe—though the political framework conditions in the U.S. are currently different. At the same time, we have often observed in the past—particularly in the U.S.—that if something makes economic sense, opportunities are seized.
Do you see significant potential for innovation on your end to make power supply more efficient and to curb the rise in energy demand?
Yes, there is still plenty of room for improvement. For instance, through the use of materials like silicon carbide or gallium nitride, which enable faster switching speeds than silicon. We can still squeeze out a few extra percentage points there.
However, it is also true that the enormous increase in the performance of AI chips is causing energy demand to rise almost exponentially—and in this context, the power supply must not become the limiting factor.
What is the value share of Infineon chips within data centers?
Current server cabinets are about the size of refrigerators and contain roughly $15,000 worth of power semiconductors. In the upcoming generation, that figure will be significantly higher. By 2030, it will likely exceed $100,000—per cabinet, mind you. A single AI data center can easily house tens of thousands of such cabinets. Of course, we don't control the entire market, but a substantial portion of those chips will be supplied by Infineon.
In the last fiscal year, chips for AI data centers accounted for just five percent of your revenue. How much could that figure grow in the future?
For the current year, we anticipate it reaching approximately ten percent. The total volume of our addressable market is projected to grow to between eight and twelve billion dollars by the end of the decade. Naturally, we have to wait and see how the expansion of AI infrastructure unfolds, but we expect this to remain a very significant growth driver for Infineon over the coming years.
The business with the automotive industry—which remains, by far, your largest division—is currently a less encouraging story. How do you assess the market outlook in this sector? We also see areas of growth here—particularly surrounding the "software-defined vehicle." In addition to the car's entire computing and communication architecture, this encompasses, for instance, automated or autonomous driving systems and safety functions. When it comes to electromobility, the situation is indeed challenging. The U.S. market is currently very weak due to a lack of political support. In China, too, growth is leveling off—owing to the already very high share of electric vehicles—and there is fierce price competition. The outlook in Europe remains comparatively positive, even if progress there is not advancing quite as rapidly as hoped.
Infineon has now had to absorb revenue declines for two consecutive fiscal years; for the current year, you are forecasting moderate growth. Do you think the business involving chips for AI data centers can help you move beyond merely moderate growth in the future?
For this year, we have initially settled on the word "moderate." However, I am very optimistic that we will see very strong growth in the AI sector in the coming years as well. We also expect to see a resurgence in other markets, including the automotive industry. Looking further into the future, I have high hopes for robotics—specifically, mobile AI-assisted robots. And looking even further ahead, we also see potential for ourselves in quantum computing. As I’ve said before: We always have several irons in the fire.
What would Infineon contribute to robots?
Ultimately, from an architectural standpoint, robots aren't all that different from software-defined cars; consequently, we could sell very similar products in that space. For instance, we can help humanoid robots move—including controlling their fingers, which is a technically demanding task. Furthermore, we enable robots to perceive their environment using sensors, and we offer solutions for communication and power management. Assuming a robot contains semiconductors worth more than $1,400, up to $500 of that value could be accounted for by our products. Alongside cars, robots represent the second major category of what is known as "physical AI." And once they are deployed on a broad scale, this could become a key growth area for Infineon.
Elon Musk has made bold predictions regarding Optimus, the humanoid robot his company Tesla is developing. He has stated that it could become the most successful product of all time. Other robot manufacturers appear somewhat more cautious. Where do you stand?
I would say it is quite normal for there to be a certain divergence in predictions when it comes to a new trend. I probably fall somewhere in the middle of that spectrum. However, I see a tremendous opportunity for Infineon, largely because—for us—it doesn't actually represent a particularly massive technological leap. As I said: There are very significant similarities with software-defined vehicles.
So, do you envision a future in which we coexist with many robots?
Absolutely. Although the physical realization may vary greatly. Not every robot will need ten fingers or legs. What matters is what’s inside. There is a computer powered by artificial intelligence that enables the robot to perceive its surroundings and operate even in highly complex environments. But not every robot will look like a human.
Interview conducted by Roland Lindner.
**About the Person**
Infineon looks back on a challenging fiscal year. For the second consecutive year, the semiconductor group’s revenue has declined, totaling just under 14.7 billion euros. Jochen Hanebeck, CEO since 2022, points to "weakness across the majority of our target markets." This includes the automotive industry—by far the most important customer segment—which accounts for roughly half of Infineon’s revenue. However, there is a bright spot, albeit in a business segment that remains significantly smaller for the time being. The company sells semiconductors used in AI data centers, where they are designed to ensure the most efficient power supply possible. Revenue from these chips nearly tripled in the past fiscal year (ending in September), and for the current year, Infineon forecasts a more than twofold increase, reaching 1.5 billion euros. For the group as a whole, "moderate" revenue growth is expected once again. [1]
1. "Europa muss einen eigenen Weg finden". Frankfurter Allgemeine Zeitung; Frankfurt. 23 Jan 2026: 24.
Komentarų nėra:
Rašyti komentarą