US-China AI Hotline: Europe Watches from the Sidelines

Samantha: Welcome to The State of Tech, The European Edition, Tuesday May twelve, 2026. I'm Samantha Lawrence.

Bob: And I'm Bob Russell. Today: Washington and Beijing reopening an AI hotline over fears of a breakthrough, Microsoft launching three new in-house AI models, Kuaishou eyeing a twenty billion dollar spin-off of its Kling video AI, China's biggest ever chip foundry merger getting the green light, plus two things you can actually use today: a quantum cloud playground anyone can try, and Japan's diamond chips heading from lab to factory. Let's start with the story everyone in Brussels and Berlin is watching.

Fears of an AI breakthrough force the United States and China to revive their emergency hotline.

Samantha: The United States and China are quietly reopening an emergency communication channel specifically for artificial intelligence. These talks are happening ahead of President Trump's state visit to Beijing, and officials describe them as urgent. The trigger, according to multiple reports, is the perceived capability of the latest frontier models, including Anthropic's new Mythos system, which industry insiders are describing as something close to an unprecedented cyberweapon.

Bob: And that framing matters. We're not talking about a chatbot that writes better emails. We're talking about models that can reportedly probe and exploit digital infrastructure, government databases, financial systems, communication networks, at machine speed. That's why the two largest AI powers on the planet suddenly want a red phone. Nobody wants a misread signal turning into a real incident.

Samantha: The structure of the channel is reportedly modelled on nuclear-era hotlines. A direct, always-on line between designated officials, so if an AI-driven event happens, either accidental or hostile, both sides can verify what's going on before reacting. Anthropic, OpenAI and several Chinese labs are said to have informally briefed officials on what their systems can now do.

Bob: It's striking how quickly the tone has shifted. Eighteen months ago the conversation was about export controls and chip bans. Now it's about deconfliction. That tells you something has spooked both governments behind closed doors. And the word Mythos keeps coming up. Whatever Anthropic has shown to regulators, it appears to have moved the needle.

Samantha: For Europe, this is a delicate moment. The EU AI Act is now fully in force, and the Commission has positioned itself as the rule-setter on safety. But this hotline is bilateral. Europe isn't at that table.

Bob: That's the awkward part. Brussels can write the rulebook, but if Washington and Beijing carve out their own crisis mechanism, Europe risks becoming a regulator without a seat in the operations room. There are already murmurs from Paris and Berlin about wanting a trilateral version, or at least observer status. The counter-argument is that any agreement between the two superpowers on AI safety protocols benefits Europe by default, because it lowers the global tail risk.

Samantha: And European companies running critical infrastructure, banks, energy grids, telcos, they're the ones who'd be on the receiving end if an advanced model went rogue or got weaponised. So even an imperfect US-China channel is better than no channel.

Bob: The bottom line is, this is the first time AI is being treated like a nuclear-tier strategic technology by both governments simultaneously. That's a turning point, regardless of what happens next.

Microsoft launches three new in-house foundational AI models to challenge OpenAI and Google directly.

Bob: Microsoft has dropped three new foundational models built entirely in-house. There's MAI-Transcribe-1 for speech-to-text, MAI-Voice-1 for voice generation, and MAI-Image-2 for image creation. All three are live today through Microsoft Foundry and a new MAI Playground.

Samantha: And the headline claim is on transcription. MAI-Transcribe-1 is being pitched as best-in-class accuracy across twenty-five languages. That's a direct shot at Whisper from OpenAI and at Google's speech stack. For a company that has spent years being seen as OpenAI's distribution partner, this is Microsoft saying, we build our own too.

Bob: The strategic signal is louder than the technical one. Microsoft has poured tens of billions into OpenAI, and that relationship isn't going away. But with these models, Satya Nadella is hedging. If OpenAI ever stumbles, or pricing gets uncomfortable, Microsoft now has its own stack ready to swap in across Azure, Office and Copilot.

Samantha: The image model is interesting too. MAI-Image-2 reportedly focuses on photorealism and commercial-safe outputs, meaning training data Microsoft can defend in court. That matters in Europe specifically, where copyright cases against generative AI providers are piling up.

Bob: For European developers and enterprises, the practical impact is immediate. If you're already on Azure, you can call these models today, in European data centres, with the compliance paperwork Microsoft has spent years building out around GDPR and the AI Act. That's a real advantage over having to route requests through US-hosted APIs.

Samantha: The catch is the twenty-five languages claim. Microsoft hasn't yet published which twenty-five, and European languages are a long tail. Dutch, Portuguese, Greek, the Nordic languages, those are where transcription quality tends to fall off. We'll see if Microsoft has actually closed that gap or if it's mostly English, Mandarin and Spanish carrying the benchmark.

Bob: Either way, the era of Microsoft as just a reseller of someone else's AI is over.

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.

Kuaishou explores a twenty billion dollar spin-off and IPO for its Kling AI video unit.

Bob: Kuaishou Technology, the Chinese short-video giant, is reportedly preparing to spin off its AI video-generation unit, Kling AI, for a separate listing next year. Market sources put the valuation at around twenty billion dollars, and pre-IPO conversations with investors have already started.

Samantha: Kling launched in 2024 and has rapidly become the most credible Chinese answer to OpenAI's Sora. It generates short video clips from text prompts, and the quality has caught a lot of people off guard. Twenty billion is a striking number for a unit that didn't exist two years ago.

Bob: It tells you where the money is flowing in generative AI right now. Video is the next frontier. Text is solved, images are commoditising fast, but high-quality video generation is still expensive, still hard, and still has enormous commercial potential, advertising, film, gaming, social content.

Samantha: The spin-off structure is also interesting. By separating Kling from the parent company, Kuaishou can raise dedicated AI capital without diluting its core advertising business. And it gives Kling its own currency, shares, to recruit talent in a market where AI researchers are getting bid up to absurd levels.

Bob: For European investors and European AI startups, this is a signal. If a Chinese video AI can command twenty billion, it changes the comparables for everyone. Companies like France's Mistral, or smaller European video AI players, suddenly have a much bigger benchmark to point at when raising their own rounds.

Samantha: There's a regulatory wrinkle though. European broadcasters and rights holders are already nervous about AI-generated video flooding platforms. If Kling becomes a publicly traded twenty-billion-dollar entity actively pushing into Western markets, expect Brussels to take a much closer look at provenance rules and content labelling.

Bob: And the geopolitics. A Chinese AI video company listing publicly, taking foreign capital, is going to draw scrutiny in Washington too. Don't be surprised if the IPO route ends up being Hong Kong rather than New York.

SMIC wins approval for a five point nine billion dollar deal, China's largest ever chip foundry merger.

Samantha: Semiconductor Manufacturing International Corporation, SMIC, has received Shanghai Stock Exchange approval for a share issuance and asset acquisition worth five point nine billion dollars, around forty point six billion renminbi. It's the largest M&A deal in China's wafer foundry industry, full stop.

Bob: The deal consolidates SMIC's grip on SMIC North, the Beijing-based subsidiary focused on twelve-inch wafer fabrication. SMIC is buying out the remaining forty-nine percent it didn't already own, turning SMIC North into a wholly-owned entity. That gives SMIC full control over a critical chunk of China's domestic chip-making capacity.

Samantha: This is the response to years of US export controls. Beijing has poured state money into building a self-sufficient semiconductor ecosystem, and this consolidation lets SMIC move faster, invest more aggressively, and capture all the upside from a facility that's central to that strategy.

Bob: Twelve-inch wafers matter because that's the standard for advanced logic and memory chips. If SMIC can scale this capacity, it reduces China's dependence on TSMC and Samsung for a meaningful slice of its domestic demand.

Samantha: For Europe, the implications cut two ways. On one side, a stronger Chinese foundry sector means more global chip supply, which is good for European automakers and industrial customers who got burned during the 2021 shortage. On the other side, it accelerates the decoupling that European policymakers are still figuring out how to navigate.

Bob: The European Chips Act has thirty billion euros earmarked for boosting domestic fabrication. Compared to what China is now consolidating in a single deal, that suddenly looks modest. Europe doesn't have a national champion at SMIC's scale. STMicroelectronics, Infineon, NXP are strong in specific niches, but no one is doing leading-edge logic at volume.

Samantha: And the trend is clear. National self-sufficiency in chips is no longer aspiration, it's industrial policy in every major bloc. The question is whether Europe can move fast enough to matter.

Bob: To close out, two things you can actually use today.

Quantum computing leaves the lab as European startup eleQtron opens cloud access to its trapped-ion machines.

Samantha: Quantum computing is having a genuinely practical moment. D-Wave Quantum reported its first-quarter fiscal 2026 results this week, Rigetti Computing announced its 108-qubit system is now generally available, and Germany's eleQtron just closed a fifty-seven million euro Series A to scale production and, crucially for our listeners, expand cloud access to its trapped-ion quantum computers.

Bob: That cloud access is the interesting bit for anyone curious. You don't need to buy a quantum computer or fly to a lab. Several providers, including D-Wave, IBM Quantum, Rigetti, and now eleQtron, let you log in through a browser and run actual quantum circuits today. IBM Quantum has a free tier you can sign up for in about five minutes.

Samantha: If you want to try it this afternoon, IBM's Quantum Platform is probably the easiest entry point. You create an account, you get access to real quantum hardware, and there are pre-built tutorials in Python using their Qiskit library. Even if you've never touched quantum code, the visual circuit composer lets you drag and drop quantum gates and see what comes out.

Bob: D-Wave's Leap platform is the other free option, and it takes a different approach, optimisation problems rather than gate-based circuits. So if you've got a logistics puzzle, a scheduling problem, or a portfolio allocation question, Leap is genuinely useful, not just a toy.

Samantha: The eleQtron angle is specifically European. They're based in Siegen, Germany, and the cloud expansion means European researchers and developers can run workloads on European hardware, in European data centres, which matters if you're working with sensitive data or under public-sector procurement rules.

Bob: And finance and logistics are the early adopters. Banks are testing quantum for risk modelling, shipping companies for route optimisation, pharma for molecular simulation. None of this is hypothetical anymore. So if you're technically curious, pick a provider, spend an hour with a tutorial this week, and you'll be ahead of ninety-nine percent of people in your office.

Samantha: It's free, it's browser-based, and it's a genuinely new skill on your CV.

Japan's diamond chip startups move from prototype to factory, opening up extreme-environment electronics.

Bob: Japan's diamond semiconductor sector is going commercial. University-backed startups are now accelerating factory construction, sample production and live device demonstrations. Diamond chips handle higher frequencies, more power, and harsher environments than silicon, and they're heading from academic curiosity to actual product.

Samantha: Now this isn't something you can download or install, but it is something you can look into today if you work in any industry that touches power electronics, electric vehicles, aerospace, industrial sensors, or telecoms infrastructure. Several of these Japanese startups, including Power Diamond Systems and Orbray, are publishing sample specs and demo videos online right now.

Bob: The practical takeaway for listeners: if you're in procurement, R&D or product development, this is the moment to start tracking diamond chip suppliers. The first commercial samples are reportedly shipping this year. Getting on an evaluation list early is the kind of move that pays off in eighteen months when these components hit production volume.

Samantha: For European industrial players, particularly German automotive suppliers and French aerospace firms, diamond chips solve a real problem. Silicon carbide and gallium nitride are great, but they hit ceilings at extreme temperatures and voltages. Diamond keeps going. For an EV battery management system or a satellite power converter, that's a meaningful upgrade.

Bob: And spend twenty minutes searching for diamond semiconductor sample programs. Most of these Japanese startups have English-language websites, and several are actively looking for European pilot customers. It's free to request a datasheet, it costs nothing to start the conversation, and you might end up with a design-in advantage your competitors won't have for two years.

Samantha: A very concrete homework assignment to close on.

Samantha: Today we covered: the US-China AI hotline being revived, Microsoft launching three new in-house AI models, Kuaishou spinning off Kling AI for a twenty billion dollar IPO, SMIC's record Chinese chip merger, quantum computing you can try in your browser today, and Japan's diamond chips moving into production.

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.