← The State of Tech RSS Feed

5 juni 2026 · 13:08

AI Catastrophe Risk: Experts Put Odds at 10% by 2031

A survey of 272 AI experts from the University of Queensland and MIT FutureTech puts at least a 10% probability on catastrophic AI outcomes within the next five years, pointing to AI-powered cyberattacks, weapons development, and governance failure as the main drivers. For European policymakers navigating AI Act implementation, the five-year window is shorter than most regulatory timelines. This episode also covers a humanoid robot taking on reception duties at a dental clinic, record semiconductor equipment billings, China's formal accusation against US chip export controls, and the Quantinuum quantum computing IPO.

Humanoid Robots AI Safety Semiconductors Export Controls China Quantinuum

Beluister deze aflevering:

Transcript

Samantha: Welcome to The State of Tech, The European Edition, Friday June five, 2026. I'm Samantha Lawrence.

Bob: And I'm Bob Russell. Today: a humanoid robot starts work at a dental clinic, AI experts put a ten percent chance on catastrophic outcomes within five years, China accuses the US of breaking the global chip supply chain, semiconductor equipment billings hit a record high, and to close out, two things you can actually try this weekend: the quantum computing newcomer Quantinuum hitting the public market, and SpaceX flagging water scarcity as the next big constraint on AI. Let's start with that robot at the front desk.

Faraday Future deploys its humanoid FF Master robot at a Los Angeles dental clinic for reception duties.

Samantha: So this is a small story with a big signal. Faraday Future, the electric vehicle company, has delivered its humanoid robot, called FF Master, to a dental practice in Los Angeles. The Wonderful Life Dental Group. And the robot is not drilling teeth, just to be very clear. It is working the front desk.

Bob: Right, and that distinction matters. The robot handles patient check-in, appointment look-up, reception, and helping people find their way through the clinic. Anything clinical is explicitly off-limits. So this is a humanoid in a customer service role, not a medical role.

Samantha: And that is exactly what makes it newsworthy. We have been hearing for years that humanoid robots are coming to factories and warehouses. This is one of the first concrete cases of a humanoid stepping into a public-facing service job, in a place where ordinary people walk in off the street.

Bob: Faraday Future is framing this as the first real-world healthcare use case for its Embodied AI line. Embodied AI is the buzzword for AI that has a physical body, that can see, move, and interact in real spaces, not just answer questions in a chat window.

Samantha: And the choice of a dental clinic is interesting. Reception work is repetitive, scripted, and exactly the kind of task where a humanoid can demonstrate value without taking on any clinical risk. If it mispronounces your name, that's awkward. If it misreads an X-ray, that's a lawsuit.

Bob: So they are starting in the safest possible corner of healthcare. And the bigger question is whether this scales. Service jobs in Europe, hotels, hospitals, town halls, are under huge staffing pressure. A humanoid that can do the basic greeting, queueing, and wayfinding could land here quickly if the price is right.

Samantha: At the same time, this is one deployment, in one clinic, in Los Angeles. We don't have data yet on how patients actually react. Some people will love the novelty. Others will walk out.

Bob: And there is the labour question. Reception is often a first job, a part-time job, or a role for someone retraining. If humanoids start picking those up at scale, that conversation arrives in Europe fast.

A global survey of AI experts puts a ten percent chance on catastrophic AI outcomes within five years.

Samantha: Onto something heavier. Researchers from the University of Queensland and MIT FutureTech surveyed 272 AI experts around the world. The headline finding: those experts estimate at least a ten percent probability of catastrophic outcomes from AI within the next five years.

Bob: Ten percent is not a small number when we are talking about catastrophic. That is roughly the odds of rolling a one on a ten-sided die, applied to AI causing serious, large-scale harm before 2031.

Samantha: The experts pointed to four areas in particular. AI-powered cyberattacks. AI in weapons development. Competitive pressure pushing labs to release systems too quickly. And governance failure, meaning regulators not keeping up.

Bob: And even setting aside the catastrophic scenarios, they flagged everyday risks that are already reshaping society. False information at scale, overreliance on AI in decisions where a human should be in the loop, and fraud. None of that requires science fiction. It is happening now.

Samantha: The researchers are explicit that they are urging immediate action from the people building these systems and the people governing them. This is not a far-future warning. The five-year window is the point.

Bob: For Europe this lands at a specific moment. The AI Act is in implementation phase, the rules are coming into force in stages, and there is ongoing debate about whether the timelines are too slow given how fast the technology is moving.

Samantha: And the survey adds weight to the argument for international coordination. Cyberattacks and weapons don't stop at borders. A purely European framework is not enough if the labs and the threats sit elsewhere.

Samantha: Quick interruption. If you listen to The State of Tech regularly, hit that like button and subscribe, that way you'll never miss an episode. Okay, moving on.

China formally accuses the United States of destabilising the global semiconductor supply chain through export controls.

Bob: China has made its accusation against Washington more formal. The Chinese Ministry of Commerce, through spokesperson He Yongqian, says the US is severely undermining the stability of global semiconductor industrial and supply chains by abusing export controls.

Samantha: The trigger is a recent round of US measures aimed at closing what Washington calls regulatory loopholes around chip exports. Beijing's position is that those measures harm the legitimate rights of Chinese companies and disrupt international trade order.

Bob: And the demand is direct. China is urging the US to rectify its practices and stop what it calls discriminatory measures, in order to safeguard global supply chain stability.

Samantha: This is the latest round in a fight that has been escalating for years, but the language is sharper. We are now at the point where China is openly framing US chip policy as a threat to the global system, not just a bilateral dispute.

Bob: And the timing matters. We just had record numbers for semiconductor equipment billings globally. The industry is booming, demand for AI chips is enormous, and the geopolitics around who can buy what from whom is getting more, not less, tangled.

Samantha: For Europe this is the uncomfortable middle. European chip equipment makers, ASML being the obvious name, sit right in the line of fire. Their machines are essential to both sides, and export rules from Washington reach into Dutch boardrooms.

Bob: And Europe's own chip ambitions, the Chips Act, the new fabs being built, depend on a stable global supply chain. If the US-China standoff hardens further, the cost of European tech sovereignty goes up, not down.

Samantha: It also raises a question Brussels has been quietly wrestling with. How aligned does Europe want to be with US export policy when the bill partly lands on European companies?

Global semiconductor equipment billings hit a record thirty-six point five billion dollars in the first quarter, up fourteen percent.

Bob: And that brings us neatly to the numbers behind that fight. The industry association SEMI reports global semiconductor equipment billings reached 36.55 billion dollars in the first quarter of 2026. That is a 14 percent jump compared with the same quarter last year, and a record.

Samantha: The driver is, unsurprisingly, AI. Chipmakers are spending heavily on capacity expansion and on upgrading to the most advanced technology, the kind needed for leading-edge logic chips, for the memory that AI servers depend on, and for advanced packaging.

Bob: Advanced packaging is the bit where multiple chips are stitched together into one tightly integrated unit. It sounds boring, but it is one of the biggest bottlenecks in AI hardware right now.

Samantha: SEMI's chief executive Ajit Manocha framed this as continued confidence in AI's foundational hardware. Translation: the people who actually buy the machines that make the chips are still betting heavily that AI demand keeps growing.

Bob: And this is a useful reality check against the recurring question of whether we are in an AI bubble. Equipment billings are a leading indicator. Companies don't spend tens of billions on fab equipment unless they expect orders for years.

Samantha: For Europe there are two angles. One is supply. European AI projects, from research labs to industrial deployments, depend on access to these chips. Strong global investment is broadly good news for availability.

Bob: The other is competitive. Europe has ambitions in chip manufacturing, but the bulk of this 36 billion is being spent in Asia and the US. The gap with the leading chip regions is widening in absolute terms, even if Europe is growing.

Samantha: And it puts a sharper point on the China-US dispute we just discussed. The bigger the equipment market gets, the more consequential every export control decision becomes.

Quantum computing company Quantinuum prices an upsized IPO and starts trading on Nasdaq today.

Bob: To close out, two things you can actually look at today. First, Quantinuum. The full-stack quantum computing company has priced an upsized initial public offering of 28 million shares at 60 dollars per share. Shares begin trading on Nasdaq today under the ticker QNT.

Samantha: This is a Smart Find in the sense that it is something concrete a curious listener can pull up this afternoon, watch the opening price, follow the ticker. It is not investment advice, just an interesting moment to witness.

Bob: Quantinuum works on quantum computing across the full stack, meaning both the hardware and the software. They list market leaders in pharmaceuticals, materials science, financial services, and government as their customers.

Samantha: And the upsized part of the offering is the interesting signal. When a company increases the size of its IPO at pricing, that usually means demand was stronger than expected. Investor appetite for quantum is clearly there.

Bob: Quantum computing is one of those technologies people have been told is ten years away for about thirty years now. But there has been real progress recently, and quantum is increasingly framed as the next foundational layer underneath advanced AI.

Samantha: For Europe this matters because Europe has its own quantum players, in France, Germany, the Netherlands, and Finland, often spun out of universities. A successful IPO sets a public benchmark for what investors will pay for quantum at this stage of maturity.

Bob: And there is a practical takeaway for the curious listener. Open Nasdaq, type in QNT, and watch how the first trading day goes. It is a small window into how seriously the public markets are taking quantum right now.

Samantha: The other useful thing: Quantinuum's own website spells out which sectors they are working with. If you want a quick read on where quantum is actually being applied today, that is a free resource you can scan in ten minutes.

SpaceX warns water scarcity is becoming a critical risk for AI data centre expansion.

Bob: And the second thing to take with you. SpaceX, in an amendment to its IPO filing, has flagged water scarcity as a critical risk to the operation and expansion of large-scale AI infrastructure.

Samantha: This is a Smart Find in the awareness sense. The filing puts in writing something that has been whispered about for a while. Modern data centres, especially the ones running AI, use enormous amounts of water for cooling.

Bob: SpaceX cites droughts, limited water access, local competition for water, and regulatory restrictions as direct threats to future AI growth. So this is not an environmental campaigner saying it. This is a major tech operator telling investors, plainly, that water could be a hard ceiling.

Samantha: The practical takeaway for listeners: when you next read a story about a new data centre being proposed near your town or region, look up the water usage figures. Many operators now publish them, and several European countries require them.

Bob: In the Netherlands, in Ireland, in Spain, this debate is already live. Data centre permits are being challenged on the basis of water and energy use. SpaceX's warning gives that conversation more weight.

Samantha: And for anyone curious, type your country and data centre water use into a search engine. Public reports from national regulators and from companies like Microsoft and Google are surprisingly readable.

Bob: It is one of those rare cases where a corporate risk disclosure doubles as a public information tool. Free to read, takes ten minutes, and changes how you think about every AI query you make.

Samantha: Today we covered: a Faraday Future humanoid robot starting work at a Los Angeles dental clinic, experts warning of a ten percent chance of catastrophic AI outcomes within five years, China formally accusing the US of breaking the chip supply chain, record semiconductor equipment billings, and two things to check yourself: the Quantinuum IPO trading on Nasdaq under QNT, and SpaceX's IPO filing warning on water scarcity for AI data centres.

Bob: Want to know more or react? Visit stateoftech.eu or email us at info@doorzetters.net.

Bob: State of Tech, the tech world in 15 minutes.