In reporting on artificial intelligence, headlines are often dominated by American and Chinese companies. While OpenAI and other US companies together are investing around 1 trillion dollars in AI, DeepSeek has developed a competitive AI model with only 6 million dollars in development costs and 2,000 NVIDIA chips (instead of the usual 16,000). This increase in efficiency shows that AI innovation is possible even with limited resources.

However, high-quality semiconductors are required to develop and operate these models, and Europe’s real strength lies in its control of the key technology to produce these AI chips. Without the highly specialized EUV lithography machines from Europe, it is not possible to manufacture the most advanced chips or develop the most powerful AI systems.

The critical triad of AI chip production

Global competition in the field of artificial intelligence is characterized by a unique interregional dependency: Europe (ASML) supplies the critical production technology with its EUV machines at a unit price of 250 million dollars, Asia (TSMC) manufactures the most advanced chips with a market share of over 90%, and the USA (NVIDIA) develops the leading AI accelerators with its 30 years of GPU expertise.

The complex interdependence between Europe, Asia, and the US creates a challenging ecosystem of technological dependencies. Even the most efficient AI developments such as DeepSeek, which manage with significantly fewer chips, are ultimately based on this infrastructure.

ASML – Europe’s key to global chip production

ASML is the only manufacturer of EUV lithography machines that holds a key position in the global semiconductor industry. The Dutch company dominates the advanced chip production technology market with a market value of over 300 billion euros and a gross profit margin of 51.1% (Q1-Q3 2024).

Production occurs in seven cleanroom factories in the EU, the US, and Asia, with the main site in Veldhoven (Netherlands).

TSMC – The global chip manufacturer

The global market leader in contract manufacturing of semiconductors, with a market share of 64.9% (Q3 2024), followed by Samsung (South Korea) with 9.3% and SMIC (China) with 6%.

The company is the main producer of advanced AI chips and has a dominant market share in advanced manufacturing processes. TSMC is also dependent on ASML’s EUV technology, which is crucial for the production of high-end products.

NVIDIA – The AI chip pioneer

NVIDIA is leading in the AI chip market and holds a 95% market share in data center GPUs.

The new Blackwell chips are already fully booked for the next 12 months. Production will take place at TSMC using ASML’s EUV machines.

Global strategies in the race for chip sovereignty

Current market situation and targets for 2030

EU (EU Chips Act / 2023)

• Current: 10% market share
• Target 2030: 20% market share

USA (CHIPS for America Act / 2022)

• Current: 12% production capacity
• Target 2030: 30% market share

Asia (Taiwan/South Korea)

• Current: Dominance with ~80% total share
• Target 2030: decline to ~54% expected

 

European Chips Act: The Strategic Framework

The European Union plans to invest 43 billion euros in the Chips Act. One core project is four technology test facilities (so-called pilot lines).

These pilot lines are designed as a bridge between laboratory development and industrial mass production and represent state-of-the-art research and development facilities. One example is APECS (Advanced Packaging and Heterogeneous Integration for Electronic Components and Systems), where new packaging technologies for chips are developed and tested. These facilities are available to large companies, SMEs, and start-ups to test innovations before they go into mass production.

Germany as a pioneer in European chip production

With investments of over 20 billion euros, Germany plays a leading role in the European semiconductor industry. These are supported by substantial state subsidies and include four major projects:

European Semiconductor Manufacturing Company (ESMC), Dresden

• Investment volume: 10 billion euros
• Partners: TSMC (Taiwan), Bosch (Germany), Infineon (Germany), NXP (Netherlands)
• Start of construction: Ground-breaking ceremony (August 2024)
• Start of production: End of 2027
• Special feature: First TSMC plant in Europe

Infineon Fab, Dresden

• Investment volume: 5 billion euros
• Start of construction: Ground-breaking ceremony May 2023
• Final building permit for the last construction phase (May 2024)
• Start of production: 2026
• Significance: Strengthening of the “Silicon Saxony”

Bosch expansions, Dresden & Reutlingen

• Investment volume: 3 billion euros
• Time frame: Expansion by 2026
• Strategy: Expansion of existing competence centers

Wolfspeed factory, Ensdorf/Saarland

• Investment volume: 7 billion euros
• Partners: Wolfspeed (USA) and ZF Friedrichshafen (Germany)
• Start of construction: 2025 (open)
• Start of production: 2027

 

CHIPS for America Act: Offensive for Technological Dominance

The USA has earmarked a budget of USD 52.7 billion for the CHIPS and Science Act, with several major projects already underway. The program is part of a comprehensive strategy to regain technological leadership in the semiconductor industry.

The core elements are 31 strategic technology centers that serve as innovation hubs for the next generation of semiconductor technology.

However, it should be noted that the status and timelines of these projects may be subject to change due to the current political uncertainties surrounding the change of government in the US. The following overview provides a summary of the major projects:

Samsung

• Investment volume: Planned investment of up to USD 17 billion
• Start of construction: 2022
• Start of production: 2025, originally planned for 2024
• Details: New plant in Taylor, Texas

TSMC

• Investment volume:Finalized funding of 6.6 billion USD
• Start of construction: 2023
• Start of production: 2025
• Details: Construction of three state-of-the-art plants in Arizona

Intel

• Investment volume: Funding of USD 7.9 billion
• Start of construction: Partly already started
• Start of production: Ohio, before 2030, originally planned for 2025
• Details: Projects in Arizona, New Mexico, Ohio and Oregon

GlobalFoundries

• Investment volume: Finalized funding of USD 1.5 billion
• Start of construction:
  New York, expansion of existing plant by 2025.
  Vermont: Modernization already underway
• Start of production: Not specified
• Details: Expansion of production in New York and Vermont

Micron Technology

• Investment volume: Planned investment of up to USD 20 billion by 2030
• Start of construction: 2024
• Start of production: Planned for 2030
• Details: New production capacity in Clay, New York

With this coordinated measure, the USA is pursuing the goal of tripling its semiconductor production capacity by 2032. The focus is on advanced chips below 10nm, where the US share is set to rise to 28% of global production.

 

China’s “Made in China 2025”: New Paths to Technology Leadership

With the “Made in China 2025” plan and the “Big Fund III”, China is pursuing an ambitious strategy for semiconductor autonomy. A state-supported investment fund of initially USD 21 billion demonstrates the goal of achieving technological independence in the chip sector. DeepSeek’s recent successes show an alternative approach: Instead of focusing only on hardware dominance, China is increasingly focusing on optimizing software and using resources more efficiently.

Challenges and innovative solutions:

• No access to ASML’s EUV technology
• Development of alternative lithography processes
• Focus on mature process nodes (older technology)
• Optimization of chip usage (DeepSeek: 2,000 instead of 16,000 chips)
• Innovative software architectures for greater efficiency
• Development of closed supply chains
• Strengthening domestic suppliers

SMIC, a leading foundry company, already has quarterly sales of around USD 2 billion.

The development of DeepSeek shows an even more threatening dimension of the Chinese strategy: with development costs of only 6 million dollars – compared to the billions invested by Western companies – an AI model has been created that shatters the basic assumptions of the market. This extreme cost efficiency could not only jeopardize the business models of established AI companies, but also raise the question of whether their massive investments are actually justified.

However, Chinese AI systems also raise serious security concerns: DeepSeek is subject to government censorship, refuses to answer questions on sensitive political issues and could be misused as a tool for disinformation or surveillance.This development has already led to calls for increased security measures and the accelerated development of trustworthy AI systems in Western countries.

Conclusion

The global semiconductor industry is facing the biggest restructuring in its history. Asia’s current dominance, with an 80% market share, is being rebalanced by extensive investment programs in Europe and the USA. The aim is to achieve a more balanced distribution of market shares by 2030 (Asia 50 %, USA 30 %, Europe 20 %).

The unique interdependence of the key players nevertheless remains crucial: Europe’s control of key technology through ASML, TSMC’s manufacturing dominance and NVIDIA’s leadership in the AI chip market continue to form the foundation of global technology development. Germany’s emergence as a chip manufacturing hub with investments of over €20 billion further strengthens Europe’s position.

With the EU AI Act coming into force in February 2025, Europe faces a double challenge: on the one hand, chip production must be expanded and, on the other, it must catch up in AI development – in an environment in which Chinese companies are developing competitive systems with significantly fewer resources. The mandatory employee training and clear regulatory requirements of the EU AI Act offer the opportunity to develop trustworthy and ethical AI systems that stand out from potentially problematic Chinese alternatives.

In future, Europe’s technological sovereignty will not only be measured by its hardware expertise, but also by its ability to develop efficient and trustworthy AI systems.