Artificial intelligence may generate more value than any previous technology wave. That does not mean every company, valuation or investment built around it will succeed.

This is the central warning from Hermann Hauser, co-founder of Acorn Computers, a key figure in the creation of Arm and one of Europe’s most experienced deep-tech investors. Speaking on CNBC’s The Tech Download podcast, Hauser described AI as a genuine technological revolution, but one that is likely to produce significant market turbulence along the way.

Transformative technology can still be overvalued

Technology revolutions and financial bubbles are not mutually exclusive. A technology may fundamentally reshape the economy while individual companies become overvalued, business models fail and investment expectations exceed what the market can realistically deliver.

Hauser believes parts of the AI ecosystem have already moved ahead of their underlying fundamentals. He pointed in particular to high valuations and circular financing arrangements, where companies invest in, purchase from or financially support other businesses within the same ecosystem. Such structures can create momentum, but they may also make it harder to assess genuine demand and sustainable revenue.

A market correction would not necessarily undermine the long-term potential of AI. Well-capitalised organisations such as OpenAI and Anthropic may be able to navigate periods of uncertainty, even if investor expectations are reset. Smaller companies without clear differentiation, reliable revenue or access to capital could face greater pressure.

For technology leaders, the lesson is straightforward: belief in AI’s potential should not replace commercial discipline.

AI is forcing the computing industry to rethink its foundations

The AI boom has created enormous demand for semiconductors, data centres, memory and specialised infrastructure. Chipmakers have benefited as companies invested heavily in the technologies required to train and operate increasingly powerful models.

However, the current approach comes with important limitations. AI workloads remain expensive to run, advanced chips generate significant heat, memory costs are high and bottlenecks continue to affect the wider supply chain.

These constraints are encouraging researchers and semiconductor companies to explore new computing architectures.

In-memory computing, for example, aims to perform more operations where data is stored instead of repeatedly transferring it between memory and processors. Photonic computing uses light to process or move information, potentially reducing energy consumption while increasing speed.

Hauser believes AI could therefore trigger a fundamental change in computer architecture. The transformation may eventually become comparable to the innovations that allowed Arm’s energy-efficient designs to challenge established semiconductor models.

This matters far beyond the chip industry. Architecture-level changes could influence infrastructure costs, cloud strategies, software performance and the types of AI applications businesses can deploy at scale.

Europe has the expertise, but scaling remains difficult

Hauser remains confident that Europe has the researchers, engineers and entrepreneurs required to compete in AI, semiconductors, quantum computing and other deep-tech fields.

The continent’s persistent challenge is scale.

Europe regularly produces promising startups and valuable intellectual property, but many companies struggle to secure the capital, commercial opportunities and unified market access necessary to become global competitors. As a result, technologies developed in Europe may eventually be acquired, funded or commercialised elsewhere.

The issue is becoming increasingly important as AI shifts from an isolated software category to a strategic layer of the global economy.

Technological sovereignty is becoming a business priority

Europe’s dependence on foreign technology extends across AI models, cloud infrastructure, semiconductor manufacturing and chip-design software. In stable conditions, international cooperation supports innovation and economic growth. In a period shaped by geopolitical tension, trade restrictions and export controls, the same dependencies can become strategic risks.

Hauser argues that Europe should maintain its close relationship with the United States while developing enough independent capability to protect its economic and technological interests. Cooperation should remain strong, but it should not lead to permanent dependency.

For European companies, technological sovereignty does not necessarily mean replacing every global provider. It means understanding critical dependencies and ensuring the organisation retains meaningful control over its systems, data and future options.

That may involve:

avoiding unnecessary dependence on a single AI or cloud provider;
designing portable, interoperable technology architectures;
maintaining clear ownership and governance of business data;
evaluating the operational and regulatory risks of external AI models;
building internal expertise around strategically important systems;
creating realistic migration and continuity plans.

What this means for technology leaders

The AI revolution is real, but successful adoption will depend on much more than access to the latest model.

CTOs and business leaders need to distinguish durable capabilities from short-term market enthusiasm. Every AI initiative should be evaluated against clear operational needs, measurable business outcomes and the organisation’s long-term technology strategy.

Companies should also consider whether today’s decisions will create tomorrow’s constraints. A solution that is fast to deploy may become expensive to operate, difficult to integrate or impossible to migrate once it is embedded across critical workflows.

The strongest AI strategies will balance innovation with architectural flexibility, cost control, security and governance.

Building for value beyond the hype cycle

AI will likely create new products, markets and operating models. It may also bring failed experiments, corrected valuations and major changes in the infrastructure supporting modern computing.

Businesses do not need to predict every stage of that evolution. They do need systems capable of adapting to it.

At Control F5 Software, we help organisations clarify complex systems and AI initiatives, modernise critical platforms and build operational software around genuine business requirements. The objective is not simply to adopt AI quickly, but to create secure, scalable and maintainable technology that continues to deliver value after the initial excitement fades.

The coming years may indeed feel like a rollercoaster. Companies with a clear strategy and a resilient technological foundation will be better positioned to benefit from the revolution without becoming dependent on the bubble.

Watch Hermann Hauser’s full conversation on CNBC’s The Tech Download.