AI Materials Discovery, Solid-State Batteries & G7 AI Framework

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

Bob: And I'm Bob Russell. Today: AI platforms are turbocharging the hunt for new sustainable materials, solid-state batteries promise incredible EV range but face serious skepticism, banks are fighting fraud with real-time AI, multimodal AI models hit a trillion parameters, G7 nations agree on a new framework for responsible AI reporting, and decentralized identity systems take aim at deepfakes. Let's start with materials science.

AI is compressing years of materials research into weeks — and the implications for clean energy are enormous.

Samantha: So here's something that genuinely excites me. Researchers are using generative AI and machine learning to discover entirely new materials for energy storage. And we're not talking about shaving a few months off the timeline. We're talking about going from years of lab work down to weeks.

Bob: The key development here is that these AI-driven platforms can predict how a material will behave before anyone synthesizes it in a lab. You feed the model thousands of potential chemical combinations — different alloys, different compounds — and it narrows down which ones are worth actually making. Yale hosted a seminar on this just yesterday, and the researchers involved are using explainable AI layered on top of supercomputing infrastructure. So it's not just a black box spitting out answers. Scientists can actually understand why the model is recommending a particular compound.

Samantha: And that matters because trust in the output is everything when you're trying to build a next-generation battery or a new type of solar cell. You can't just take a guess from a neural network and pour millions into manufacturing.

Bob: The immediate applications are in battery technology — think solid-state electrolytes, new cathode materials. But this extends to superconductors, lightweight structural materials, you name it. It's a fundamental shift in how materials science works.

Samantha: And the speed is the real story. If you can identify a superior material in weeks instead of years, you can iterate on battery designs, solar panel efficiency, hydrogen storage — all of it — at a pace that was unimaginable five years ago.

Bob: For Europe, this is directly relevant to the Green Deal. European research institutions — places like the Fraunhofer Institutes, CNRS in France, Max Planck in Germany — they're heavily invested in advanced manufacturing and green tech. If AI-driven discovery becomes the standard, and European labs adopt it aggressively, it could reduce the continent's dependence on imported materials and give European manufacturers a genuine edge in clean energy components. The EU has poured billions into battery gigafactories. Making sure those factories have access to the best possible materials is the next piece of the puzzle.

Samantha: It's one of those cases where AI isn't replacing scientists — it's giving them superpowers. Alright, let's talk batteries.

Solid-state batteries could transform EVs — if anyone can actually mass-produce them.

Bob: This is one of those stories where the hype and the reality are in a very interesting tug-of-war. Solid-state batteries have been the holy grail of EV technology for a decade. Higher energy density, faster charging, longer lifespan, safer because there's no liquid electrolyte sloshing around. And now multiple companies are claiming they're close to cracking it.

Samantha: Finnish startup Donut Lab announced what they're calling a production-ready solid-state battery. Bold claim. But the established players — Toyota, Samsung SDI, the big names — they're raising eyebrows. Scalability is the issue. Making one cell in a lab is very different from making millions on a production line.

Bob: Meanwhile, Chinese automakers are not waiting around. Changan has unveiled what they call the Golden Bell battery — claiming a range of 932 miles on a single charge. They're planning vehicle testing by the third quarter of this year. And semi-solid-state batteries, which are sort of a halfway step, are already going into mass production in China for commercial vehicles and some passenger EVs.

Samantha: And that's the competitive tension. If Chinese manufacturers get to scale first with even semi-solid technology, they extend an already significant lead in EV battery production.

Bob: European carmakers are watching this very closely. Volkswagen has its partnership with QuantumScape, BMW is working with Solid Power, but the timelines keep slipping. Europe has spent enormous political and financial capital building out battery manufacturing capacity. Northvolt in Sweden, ACC in France and Germany. But if the underlying chemistry shifts to solid-state and Europe isn't ready, those gigafactories could be producing yesterday's technology. The EU needs to ensure its battery strategy is flexible enough to pivot as the chemistry evolves.

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.

Banks are deploying AI that can spot fraud in real time — and it's saving billions.

Samantha: Next up, the financial world is going all-in on AI to fight fraud. And the numbers are staggering.

Bob: JPMorgan Chase is a good example. They've rolled out an enterprise-wide AI platform called OmniAI that analyzes transactions in real time. We're talking about every swipe, every transfer, every online payment being evaluated by machine learning models that are trained to spot anomalies. And critically, these systems are getting much better at reducing false positives — those annoying moments when your legitimate purchase gets blocked because the algorithm flagged it incorrectly.

Samantha: False positives are actually a huge cost center for banks. Every flagged transaction requires human review. So if AI can cut false positives significantly, you're saving money and making customers happier at the same time.

Bob: The other side of this is that fraud itself is becoming AI-powered. Criminals are using generative AI to create convincing phishing emails, synthetic identities, deepfake voice calls. So it's essentially AI versus AI at this point. Financial institutions that don't adopt these tools are going to fall behind very quickly.

Samantha: The scale of the problem is enormous. We're talking about billions in losses annually across the global financial system.

Bob: European banks have to thread a very particular needle here. GDPR imposes strict limits on how personal data can be processed. The upcoming AI Act adds another layer of compliance requirements for high-risk AI systems — and fraud detection absolutely qualifies as high-risk. So European financial institutions need AI that's not only effective but also explainable and auditable. There's also a concentration concern. The market for enterprise AI fraud solutions is dominated by a handful of providers. European regulators are keeping a close eye on that, because over-reliance on one or two vendors creates its own systemic risk.

Samantha: Alright, let's get into the AI model wars.

A trillion parameters, native multimodal support — AI models are getting scarily capable.

Bob: DeepSeek just dropped V4 — a trillion-parameter model with native multimodal capabilities. It processes text, images, video, and audio in one unified architecture. And it's competitive with the best proprietary models from OpenAI and Google.

Samantha: And Google isn't sitting still. They've announced Gemini Embedding 2, which is natively multimodal for embeddings. That means you can map a photo, a paragraph of text, and an audio clip all into the same mathematical space. For developers, that unlocks incredibly powerful search and classification tools.

Bob: What's remarkable is the pace. We went from text-only large language models to fully multimodal systems in what — eighteen months? The development cycles are getting shorter and shorter. These models are enabling things like real-time video understanding, sophisticated content generation that blends media types, and much more natural conversational interfaces.

Samantha: This changes everything from customer service to medical imaging to industrial automation. When AI can see, hear, read, and understand context across all those inputs simultaneously, the range of applications explodes.

Bob: European tech companies and startups now have access to incredibly powerful open and semi-open models. That's good for innovation. But the AI Act classifies many of these applications — especially in healthcare, finance, and public services — as high-risk. So deploying a trillion-parameter multimodal model in a European hospital, for instance, comes with serious compliance obligations around transparency and bias testing. It's a balancing act. Europe wants innovation, but it also wants guardrails.

Samantha: Speaking of guardrails — the G7 has a new plan.

The world's leading economies just agreed on how AI developers should report risks.

Bob: G7 nations have unveiled a new international reporting framework for responsible AI. This builds on the Hiroshima AI Process from 2023, and it provides a standardized way for organizations developing advanced AI to demonstrate they're following a shared code of conduct. It covers risk management, incident reporting, and best practices for assessment.

Samantha: So think of it as a common language for AI accountability. Instead of every country having completely different requirements, there's now a template that companies can use to show they're being responsible — regardless of which jurisdiction they're operating in.

Bob: That fragmentation risk is real. If the US, EU, Japan, and the UK all have incompatible AI governance frameworks, multinational companies face a nightmare of compliance. This G7 framework doesn't replace national laws — it sits alongside them. But it creates a common floor.

Samantha: Europe has arguably the most developed AI regulatory framework in the world right now with the AI Act. So this G7 initiative actually aligns nicely with what Brussels has already been pushing. For European companies operating globally, international alignment means less friction. And for Europe as a bloc, it means its regulatory philosophy — transparency, human oversight, risk-based classification — is influencing the global conversation rather than becoming an isolated outlier.

Bob: It's a smart diplomatic play. Europe gets to export its values while also making life easier for its own companies abroad.

Samantha: Last story for today.

Your digital identity might soon belong to you — not to a tech platform.

Bob: Decentralized identity systems are gaining real momentum as a response to the deepfake crisis. The idea is simple but powerful — instead of relying on a tech platform to verify who you are, you control your own digital identity using cryptographic credentials, often anchored to a blockchain.

Samantha: And the timing makes sense. Deepfakes are getting so good that traditional verification methods — a photo ID, a video call — just aren't reliable anymore. You need something cryptographically verifiable. Something a generative AI model can't fake.

Bob: Several social platforms are experimenting with integrating these self-sovereign identity systems. You'd have a verifiable credential that proves you're a real person without necessarily revealing your name, your address, or any other personal information. It's privacy-preserving authentication.

Samantha: If people can't trust that the person they're talking to online is real, the entire digital economy suffers. This isn't just about social media. It's about remote work, e-commerce, digital government services — everything.

Bob: The EU has been working on digital identity wallets — the eIDAS 2.0 regulation is pushing exactly this direction. Decentralized identity aligns perfectly with GDPR's principles of data minimization and user control. For European citizens, this could mean a future where you prove your age, your qualifications, or your identity online without handing over your entire personal profile to every service you interact with. It's a fundamentally more European approach to the internet — user-centric rather than platform-centric.

Samantha: Today we covered: AI speeding up materials discovery for clean energy, the solid-state battery race heating up, banks fighting fraud with real-time AI, trillion-parameter multimodal models pushing new boundaries, the G7 agreeing on an AI reporting framework, and decentralized identity taking on deepfakes.

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.