AI Trends 2026: Market Concentration, DeepSeek, and European Digital Sovereignty

Executive Summary

The FAZ Digitalwirtschaft Podcast analyzes the most important technology trends for 2026. The focus is on valuation bubbles in the AI sector despite impressive investments, surprising competition from the Chinese model DeepSeek, and Europe's efforts toward digital sovereignty. While Nvidia leads with a market capitalization of 4.5 trillion dollars, skepticism toward unrealistic profit expectations is simultaneously growing. The discussion shows: AI will be transformative, but the path to get there could be more volatile than previously priced in.

People

Topics

Detailed Summary

AI Market Trends and Valuation Questions

The two hosts discuss a significant shift in market perception around artificial intelligence. While the year began with major announcements such as Trump's Stargate initiative (500 billion dollars for AI infrastructure), increasing nervousness set in over the course of the year. Continuous billion-dollar investments and circular deals between AI providers—such as Nvidia's stakes in OpenAI and Anthropic—ultimately led to a shift in market sentiment.

A central topic is Nvidia's decoupled valuation. With a market capitalization of approximately 4.5 trillion dollars, the question arises whether business growth can still justify this valuation. While the company's technical superiority in AI high-performance chips is undisputed, stock price growth could now be outpacing real business developments.

The hosts compare the current situation to the Gartner Hype Cycle: technologies are overestimated in the short term and underestimated in the long term. While they believe AI will achieve great things in the long term, there could be significant market corrections on the path to long-term gains. Particularly noteworthy is the extreme market concentration: by year-end, the ten leading US stock market companies accounted for over 40 percent of the S&P 500 value—significantly higher than in previous years or even during the dot-com bubble.

The DeepSeek Surprise and Chinese Competition

A turning point was the release of the Chinese model DeepSeek R1 in early January. The model achieved comparable performance to top models from OpenAI and other American providers with a fraction of the training resources. This triggered massive market reactions: in a single day, up to one billion euros disappeared from market capitalizations as investors questioned why gigantic investments in AI data centers are necessary if this can be done more cheaply.

Although initial panic has subsided, DeepSeek remains relevant. The model is open source, and European companies like the technology consultancy TNG from Munich have already developed variants that reduce Chinese censorship measures. DeepSeek has taken over the role of Meta's Llama models as the best open-source solution in many places.

At the same time, other Chinese startups and large companies like Kimi are developing promisingly. An arms race between China and the USA in the AI sector is in full swing.

Europe's Role and Digital Sovereignty

The discussion shows a realistic assessment of Europe's positions: while Europe is falling behind American and Chinese competition in foundation models, interesting AI application companies are emerging. Companies like Lovable from Sweden (one of the fastest-growing companies overall), as well as German startups like N8N, Valor, and Langdog show that innovation is happening at the application layer.

A promising player is the Schwarz Group with its digital division Schwarz Digits. The company positions itself as a European player with ambitious goals: the cloud solution Stackit is to function as a European alternative, and an 11-billion-euro data center in Lübbenau (Brandenburg) is under construction. However, even these investments cannot compete with American hyperscalers.

Regarding digital sovereignty, a pragmatic approach is developing: instead of complete independence, major European corporations are pursuing multi-cloud strategies. They work with two to three cloud providers to reduce dependency risks. Particularly sensitive data could be stored with European providers, while less critical data continues to flow through American providers.

Maximilian Sachs argues that Europe's own foundation models would be desirable to prevent value creation from being completely lost to the USA. However, he himself lacks a "recipe for success" for their development. The central challenge is not just developing a model, but operating it sustainably—with appropriate hardware, data centers, and financial staying power.

AI Competition Among Tech Giants

Contrary to earlier predictions, no clear winner has emerged. Google was skeptical when OpenAI introduced ChatGPT, but with Gemini 3, the company showed how it can respond to the market with full force. Other providers like Microsoft and Anthropic are also establishing themselves. This is to be valued positively, as no AI monopolization is taking place.

Application Scenarios and User Behavior

In personal use as a productivity assistant, AI experienced enormous growth in 2025. Millions of people adopted these technologies in record time—a success that even surprised OpenAI.

For 2026, it is expected that pure AI search engines will gain relevance, but classical search engines will not disappear. Google found a good compromise with intelligent summaries alongside traditional search results. Younger users (17–18 years old) are already using YouTube, TikTok, or AI search instead of traditional Google search.

AI agents that perform autonomous transactions are technically possible but have not yet reached the mainstream. Skepticism regarding financial transactions is still too high.

Technological Foundations and Future Directions

One core question remains: how will AI systems be further developed? There are three scientific positions:

1. Scaling thesis: Ever larger models, more data, and more computing power will eventually lead to AGI (Artificial General Intelligence).

2. Hybrid thesis: Pure scaling is insufficient; additional mechanisms such as cause-and-effect understanding and logical thinking are needed. Humans can not only react to sensory impressions but also think about imaginary worlds.

3. Skeptical thesis: Current progress is still rudimentary—however, this position has become significantly weaker.

The discussion suggests that the question will be decided between thesis 1 and 2. The many ambitious predictions (AGI by 2028–2030) are viewed skeptically, as even leading experts cannot realistically assess this.

Social Media, Youth Protection, and Age Verification

A second major topic is youth protection in social media. Several countries are already introducing bans or age restrictions. The hosts prefer a more differentiated approach: effective age verification rather than complete bans.

Technically, functioning verification systems already exist. The main problem so far has been that platform operators have economic incentives against such measures (fewer users = less advertising revenue). However, political pressure is likely to increase.

Scientifically, it is proven that intensive social media use has negative effects on adolescents' brain development—not just functionally, but also in physical brain structure. This justifies stronger regulation.

However, social media also brings benefits: fast communication, spontaneous meetups, peer networking. These should not be completely eliminated.

TikTok Deal and Geopolitical Implications

After several years of uncertainty, the TikTok situation is moving forward. A US joint venture involving Oracle, Silver Lake, and MGX (Abu Dhabi) is taking shape. While Trump originally wanted to block TikTok over spying concerns, he himself benefited from the platform during his re-election. A "friendly deal" resolves this tension.

Interestingly, TikTok is being "protected" from Chinese influence by involving a state fund from Abu Dhabi—a certain contradiction.

For Europe, it is relevant: if two TikTok versions exist (Chinese and American), Trump could potentially pressure European users to use the "better" US version.

Fundamentally, it is about control over communication channels—a phenomenon not limited to China. The USA also prefers control over critical communication tools.

Disinformation Campaigns and Hybrid Threats

A frequently overlooked topic is intensified hybrid threats from Russia. The Office for the Protection of the Constitution and intelligence services report increased intensity of disinformation campaigns. This also manifests itself in drone overflights and other hybrid activities. This topic will remain relevant in 2026.

Key Takeaways

• Market valuations decoupled: Stock gains of AI providers like Nvidia (4.5 trillion dollar market cap) could be decoupled from fundamental business growth. Market corrections are to be expected.

• DeepSeek as a game changer: The Chinese AI model showed that similar performance is possible with significantly fewer resources—an existential shock to the investment narrative.

• Fragmentation instead of monopoly: Contrary to early concerns, no absolute winner has emerged. OpenAI, Google, Microsoft, and others compete successfully—positive for market diversity.

• European opportunities in applications: While foundation models are dominated by the USA and China, promising European startups are emerging in the application area (Lovable, N8N, Valor, Langdog).

• Schwarz Group as European hope: With 11 billion euros in data center investments and the Stackit cloud offering, the company is attempting to enable European digital sovereignty—but with enormous resource requirements.

• Pragmatic multi-cloud strategies: Rather than complete independence, European corporations work with multiple cloud providers and store sensitive data with European providers.

• Age verification before bans: Effective technical protection for adolescents in social media is feasible, but platforms hesitate for economic reasons. Political pressure will increase.

• TikTok solution geopolitically questionable: The Americanization of TikTok via a joint venture with Abu Dhabi participation solves fewer consumer problems than expected.

• Russian hybrid threats increasing: Disinformation campaigns and other hybrid activities from Russia are intensifying and will likely continue to be relevant in 2026.