Apple AI, Neuro-Symbolic Energy, US Intelligence & Nuclear Fusion Rockets

Samantha: Welcome to The State of Tech — The European Edition. I'm Samantha Lawrence.

Bob: And I'm Bob Russell. Today: Apple goes all-in on AI ahead of WWDC, promising a completely new Siri. Researchers crack a neuro-symbolic approach that could cut AI energy use by a factor of one hundred. The US Intelligence Community names artificial intelligence its number one national security threat. Fujitsu launches a generative AI service that claims to modernize legacy code ninety-seven percent faster. MegazoneCloud and AVITA team up to put AI into physical robots. And a UK company fires up a nuclear fusion rocket engine prototype. Let's start with Apple.

Apple finally admits it needs to catch up on AI — and bets everything on WWDC 2026.

Samantha: So Apple has officially announced WWDC 2026, running June eighth through the twelfth, and the big headline is AI. They're not being subtle about it either. The company is basically saying — yes, we fell behind, and here's how we plan to fix that.

Bob: The centrepiece is Siri. And not a minor Siri update — we're talking about a ground-up rethink. The expectation is that Siri will get onscreen awareness, meaning it can actually see what you're doing on your device and respond to it. Personal context, so it understands your habits, your preferences, your schedule. And deeper app integration, potentially including full chatbot capabilities similar to what you get from ChatGPT or Google's Gemini.

Samantha: Which is fascinating because Apple has historically been the company that says, "We do things our way, on our timeline." But the AI race clearly spooked them. When your competitors have chatbots that can write essays and plan vacations and your assistant still struggles with setting a timer, you've got a problem.

Bob: And they'll be unveiling iOS 27 alongside all of this, so the AI features should be baked directly into the operating system rather than bolted on as an afterthought.

Samantha: This is important because Apple doesn't do small rollouts. When they ship something, it goes to over a billion devices. If Siri genuinely becomes competitive with the best AI assistants out there, it shifts the entire landscape.

Bob: Now for European users, this gets interesting fast. Apple will have to navigate the AI Act, GDPR, and the Digital Markets Act all at once. The kind of personal context Siri would need — your messages, your browsing habits, your location patterns — that's exactly the type of data European regulators watch like hawks. So the version of Siri that Europeans get might actually look different from what ships in the US. Apple has already had to make concessions in Europe before. Remember the sideloading changes? This could be round two.

Samantha: And honestly, European consumers might end up with a better product because of that scrutiny. Privacy by design tends to produce more thoughtful features.

Bob: Wouldn't be the first time regulation accidentally drove innovation.

---

What if AI could do the same work using one percent of the energy?

Samantha: Next up — a research breakthrough that sounds almost too good to be true, but the numbers are striking. Scientists working in Matthias Scheutz's lab have developed a proof-of-concept neuro-symbolic AI system that uses up to one hundred times less energy than standard approaches while actually being more accurate on certain tasks.

Bob: So the idea here is combining two things that usually live apart. Neural networks — which are great at pattern recognition but are energy hogs — and symbolic reasoning, which is more like traditional logic-based computing. By blending the two, you get a system that mimics how humans actually solve problems. We don't brute-force everything. We use shortcuts, rules of thumb, logical deduction. This approach does something similar.

Samantha: And the timing could not be more relevant. We keep hearing about how AI data centres are consuming enormous amounts of electricity. Some estimates say AI workloads could account for a significant percentage of global energy consumption within a few years. A hundredfold reduction in energy per task — even if it only applies to specific workloads — that's a game changer.

Bob: The bottom line is that scaling AI the way we've been doing it, just making models bigger and throwing more GPUs at them, has a ceiling. And that ceiling is partly environmental. This kind of research opens a door to a different trajectory.

Samantha: For Europe, this aligns perfectly with the EU's Green Deal and its broader sustainability goals. If neuro-symbolic approaches prove scalable, European policymakers could push for energy-efficiency standards in AI, similar to what we already have for appliances and vehicles. And European research institutions are well positioned here. There's strong academic talent in both symbolic AI and neural network research across the continent.

Bob: It could also become a competitive advantage. If European companies adopt more efficient AI earlier, they lower costs and reduce regulatory risk at the same time.

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.

---

The US Intelligence Community says AI is now its biggest worry — above conventional military threats.

Bob: The US Intelligence Community has released its annual threat assessment for 2026, and the headline is blunt. Artificial intelligence is the number one national security concern. Not nuclear weapons, not cyberattacks in the traditional sense — AI.

Samantha: And the specifics are alarming. The report zeroes in on generative AI's ability to produce realistic fake images, video, audio, and text at scale. We're talking about disinformation that's nearly impossible to distinguish from reality. And it highlights China's advancing capabilities in using AI for military applications, including autonomous weapons systems and intelligence gathering.

Bob: What strikes me is the psychological dimension they emphasise. It's not just about battlefield applications. It's about confusion, erosion of trust, and what they call psychological distress at a societal level. When you can't trust what you see or hear, the fabric of public discourse starts to fray.

Samantha: And this is significant because the US intelligence assessment often sets the tone for allied nations' security postures. When Washington says this is the priority, European capitals listen.

Bob: For Europe, this report should accelerate conversations that are already happening within NATO and the EU. European countries have been building out cybersecurity frameworks, but AI-driven threats move faster than traditional cyber threats. You need AI to fight AI, essentially. And that means investing in defensive AI capabilities, training intelligence analysts to work with these tools, and building resilience against disinformation campaigns — especially with elections happening regularly across the bloc.

Samantha: And let's not forget that Europe has its own AI Act coming into full effect. There's a real question about whether security applications get enough room to operate within that regulatory framework, or whether the rules create blind spots.

Bob: That tension between regulation and security is going to define European AI policy for the next decade, I think.

---

Fujitsu says it can turn your ancient COBOL systems into modern code — ninety-seven percent faster.

Samantha: Also today, Fujitsu has launched a new generative AI service called Fujitsu Application Transform, available as software-as-a-service starting today in Japan. And the pitch is bold. It claims to reduce legacy code modernization time by ninety-seven percent.

Bob: Here's what it does. It takes existing source code — particularly COBOL, which is still running in banks, insurance companies, and government agencies worldwide — and uses AI to analyse it and automatically generate design documents. Normally, understanding legacy code and creating documentation for migration takes months or even years. Fujitsu says their system does it almost instantly, and you don't need to be a COBOL expert to use it.

Samantha: They're using something called Knowledge Graph-Enhanced RAG to reduce hallucinations, which is clever. Because the last thing you want when modernising a banking system is an AI that makes things up.

Bob: Key takeaway — there are trillions of lines of COBOL still running critical infrastructure globally. Every year that code stays un-modernised is another year of accumulated technical debt and security risk.

Samantha: European organisations are some of the biggest users of legacy systems in the world. Think about the major European banks, national tax authorities, healthcare systems. Many of them are still running on code written in the seventies and eighties. If Fujitsu brings this service to Europe — and they almost certainly will — it could unlock a wave of digital transformation that's been stuck for years.

Bob: And the compliance angle matters too. EU digital standards keep tightening. Modernised systems are much easier to audit and secure than ancient COBOL running on mainframes nobody fully understands anymore.

---

AI is leaving the screen and entering the room — literally.

Bob: Let's talk about physical AI. MegazoneCloud, which is a major cloud and digital transformation company, has partnered with AVITA, a Japanese AI avatar specialist, to develop robots that can see, think, talk, and act independently.

Samantha: So we're not talking about chatbots on a screen. These are actual robots in physical spaces — retail stores, bank branches, factories, public service centres — equipped with AI avatars that can hold natural conversations, perceive their surroundings, and make decisions on their own.

Bob: The technology stack combines AVITA's avatar tech with autonomous robotics and cloud infrastructure. The idea is that these robots can do tasks like customer guidance, consultation, and information services without constant human oversight.

Samantha: This matters because we've been talking about AI mostly as a software phenomenon. Chatbots, image generators, code assistants. But the real economic transformation happens when AI starts operating in physical environments alongside people.

Bob: For Europe, this raises immediate regulatory questions. The AI Act already has provisions for high-risk AI systems, and autonomous robots interacting with the public in banks and shops would almost certainly qualify. Safety standards, liability frameworks, data protection — all of that needs to be sorted before these robots can operate at scale in European markets.

Samantha: And there's a workforce dimension too. European labour unions and policymakers will want to understand how physical AI affects employment — particularly in service sectors where human interaction has traditionally been valued.

Bob: It's going to be a delicate balance. The technology is clearly coming. The question for Europe is whether it shapes the rules before the robots arrive, or scrambles to catch up after.

---

A company in Bletchley just fired up a nuclear fusion rocket engine — and it worked.

Samantha: And finally, something that feels straight out of science fiction. Pulsar Fusion, a UK-based space propulsion company located in Bletchley, England, has successfully tested a prototype exhaust system for a nuclear fusion rocket engine.

Bob: They achieved what's called first plasma, using inert krypton gas as fuel. Now, this is a proof of concept — we're not launching fusion-powered rockets to Mars next year. But it's a genuine milestone. Nuclear fusion propulsion could theoretically cut transit times to Mars dramatically compared to conventional chemical rockets.

Samantha: And the fact that this happened in Bletchley of all places — the same town where they cracked the Enigma code — feels almost poetic. Britain has a habit of doing extraordinary science in unassuming places.

Bob: The significance here is that fusion propulsion has been theoretical for decades. Getting actual plasma flowing through actual hardware — that's the kind of step that separates speculation from engineering.

Samantha: For Europe broadly, this is a signal that the continent has genuine capabilities in cutting-edge aerospace. The European Space Agency and national space programmes should take note. There's an opportunity to build partnerships with Pulsar Fusion and invest in a technology that could define the next era of space exploration. And post-Brexit, this kind of scientific achievement helps keep the UK connected to European research networks.

Bob: Space is one area where collaboration tends to win over competition. If fusion propulsion works at scale, everyone benefits from working together on it.

Samantha: Today we covered: Apple betting big on AI for WWDC, a neuro-symbolic breakthrough slashing AI energy use, the US naming AI its top security threat, Fujitsu's AI-powered legacy code modernization, physical AI robots from MegazoneCloud and AVITA, and Pulsar Fusion's nuclear rocket engine test.

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.