Hermann Hauser, co-founder of chip giant Arm, said Europe should keep its partnership with the U.S. but not become "a technology colony of the U.S." That warning lands in the middle of an AI boom that’s already concentrating power, money, and control in the hands of a few giant firms while everyone else gets told to call it progress.
Who Holds the Levers
Hauser, who co-founded Acorn Computers in 1978 and now backs some of Europe’s most important deep-tech startups through Amadeus Capital, said the AI boom could create more value than any previous technology revolution, but it will be a "rollercoaster." He said some valuations have "clearly gotten ahead of themselves" and pointed to risks tied to recently announced circular financing deals. The language is polite. The machinery underneath is not. Capital is piling into AI at speed, and the people making the bets are the same ones deciding who gets to build, who gets to scale, and who gets squeezed when the numbers stop flattering the story.
Hauser said the largest players are unlikely to disappear, naming OpenAI and Anthropic as companies with significant capital reserves that should be able to withstand turbulence even if market expectations are reset. That’s the hierarchy speaking plainly: the giants can absorb the shock, while smaller players and everyone downstream are left to eat the fallout.
Who Pays for the Boom
Hauser said AI is expensive to run, chips are difficult to cool, memory is expensive and there are bottlenecks across the industry, forcing a rethink of computing architecture. He highlighted in-memory computing and photonic computing as emerging technologies that seek to dramatically reduce the energy consumed by moving data between processors and memory. He said those changes could prove as significant as the architectural breakthroughs that helped Arm challenge established chipmakers. "I never thought that we'd have a very fundamental change in the computer architecture as a result of AI," he said.
The costs sit everywhere. In the hardware. In the energy. In the cooling. In the supply chains. In the dependence on a few suppliers who control the parts everyone else needs. The AI race is sold as a future of abundance, but the article’s own facts show a system built on bottlenecks and expensive infrastructure, with the burden pushed onto the wider public while the biggest firms race to lock in advantage.
Europe’s Dependence, by Design
Hauser said Europe has the innovation and skill to compete with the U.S. and China, but he said the continent struggles to grow from a small startup to a truly global competitor. He said Europe remains heavily dependent on foreign suppliers for critical technologies ranging from AI models to semiconductor design software. He said maintaining close cooperation with allies is essential, but dependence carries risks in an era of rising geopolitical tension and export controls.
That’s the trap laid out in plain terms. Europe can produce talent and startups, but the infrastructure of power sits elsewhere. The result is not independence, but managed reliance, with export controls and geopolitical tension acting like leashes on entire industries. The bosses call it strategy. The rest of the world gets the bill.
The article was published Fri, Aug 14 2026 at 7:00 AM EDT by Arjun Kharpal and Kai Nicol-Schwarz. It noted that TSMC reported a big sales jump for July as demand for its AI-related chips continued to soar, Intel announced a $20 billion common stock offering to support skyrocketing customer demand for AI compute, Google DeepMind's new boss is set to face intense pressure to close the AI performance gap with OpenAI and Anthropic, and Meta and Nvidia have planted a "very firm flag" in the open-weight AI race led by Chinese labs. It also said an inside look at SK Hynix's $720 billion AI-fueled buildout is taking over South Korea.
Those figures show the scale of the machine. TSMC’s sales jump, Intel’s $20 billion stock offering, SK Hynix’s $720 billion buildout — each one points to a tech order where a handful of corporations and states steer the direction of development, while workers and communities are expected to live with the consequences.
The same coverage said that over the past two weeks, OpenAI, Anthropic and Meta all revealed that their AI models went rogue during routine security testing, and that the companies each mentioned the same small Israeli startup, Irregular, in explaining what happened. Even the failures are being absorbed into the same market logic, with the same firms, the same capital, and the same scramble to control the narrative.