Chinese AI Chip, Bollywood AI Production & Meta Security Breach

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

Bob: And I'm Bob Russell. Today: a Chinese AI chip that reportedly blows Nvidia out of the water, AI transforming Bollywood film production at breakneck speed, Chinese firms trading AI-enriched intelligence on US troop movements, Meta hitting pause on a major AI data partner after a security breach, the White House dropping a new national AI policy framework, and a charging tech company making waves as fuel prices climb. Let's start with that Chinese chip.

A Chinese processor claims a 300% performance lead over Nvidia's best GPU

Samantha: So Bob, this one really caught my eye. A relatively obscure Chinese chipmaker has apparently benchmarked a processor that outperforms Nvidia's flagship GPU by three hundred percent. Three hundred. That's not an incremental improvement. That's a generational leap — if the numbers hold up.

Bob: And that's the key qualifier — if the numbers hold up. We've seen bold benchmark claims before that don't always translate into real-world performance. But the fact that this is even being discussed seriously tells you something about where China's semiconductor ambitions are right now. They've poured enormous resources into advanced chip R&D, and this could be the moment those investments start paying off in a very visible way. Silicon Valley is clearly rattled. Nvidia's stock took a hit on Friday just on the rumours alone.

Samantha: And the geopolitical dimension is impossible to ignore. The US has spent the last few years trying to restrict China's access to advanced chip technology through export controls. If China can leapfrog those restrictions with domestic innovation, the entire strategy needs rethinking.

Bob: For Europe, this is a real wake-up call. European companies — from automotive to industrial AI — are heavily dependent on Nvidia and a handful of other suppliers for high-performance chips. If the balance of power in semiconductors is genuinely shifting, European policymakers need to accelerate efforts around the EU Chips Act. Diversifying supply chains isn't just a nice talking point anymore. It's becoming urgent. And frankly, Europe's own chip design capabilities need serious investment if we don't want to be caught between two superpowers in a technology cold war.

Samantha: The question is whether Europe has the political will to move fast enough. These things take years.

Bob: Years we may not have, honestly.

AI slashes Bollywood production costs to one-fifth and cuts timelines by 75%

Samantha: Alright, let's shift to something a bit more colourful. Bollywood — India's massive film industry — is going all in on AI, and the numbers are staggering. Production costs for genres like mythology and fantasy are dropping to a fifth of what they used to be, and production timelines are shrinking by up to seventy-five percent.

Bob: Think about what that means in practical terms. A film that used to cost ten million dollars and take two years to produce could now be made for two million in six months. They're using AI to generate digital actors, create entire visual worlds, and even re-edit existing films with completely new storylines. There's a re-release of a film called Raanjhanaa that's been reworked with AI-altered plot elements. That's a fascinating and slightly unsettling development.

Samantha: It is. And predictably, there's pushback from artists who worry about creative integrity. If you can digitally recreate an actor's performance or rewrite a film's narrative with algorithms, where does human artistry fit in?

Bob: It's a legitimate concern. But the efficiency gains are so dramatic that the industry is embracing this regardless. And here's where it gets interesting for Europe. European film industries — think France, Germany, Italy, Scandinavia — have traditionally been more cautious about this kind of disruption. But when you see Bollywood cutting costs by eighty percent, European studios and streaming platforms are going to take notice. The competitive pressure alone will force adoption. Especially for effects-heavy productions where budgets have been a real constraint compared to Hollywood.

Samantha: It could actually democratise filmmaking in Europe. Smaller studios could suddenly punch above their weight.

Bob: That's the optimistic reading, and I think there's truth to it. But the regulatory conversation around AI-generated content in Europe is going to intensify. Questions about intellectual property, performers' rights, disclosure — all of that needs frameworks, and fast.

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.

Private Chinese companies are reportedly trading AI-enhanced intelligence on US military movements

Samantha: This next story has a very different tone. Private companies in China are reportedly buying and selling intelligence about US troop movements and military equipment, and they're using AI to do it. They combine open-source materials with commercial satellite imagery — including data from China's Jilin satellite constellation — to detect and compile detailed activity reports.

Bob: This is the sharp end of AI-powered open-source intelligence, or OSINT. What's new here isn't that satellites can photograph military bases — that's been possible for decades. It's the AI layer on top. These systems can automatically detect changes in troop positions, identify specific vehicle types, flag unusual activity patterns, and generate reports at a speed and scale that would have required a full intelligence agency just ten years ago. And now private companies are doing it commercially.

Samantha: Which raises a whole host of questions. Who are the buyers? What are the end uses? And how do you regulate something that's technically built on publicly available data?

Bob: That last point is the real headache. Individually, each piece of data might be innocuous — a commercial satellite image, a social media post, a shipping record. But AI can fuse all of that into something that looks a lot like classified intelligence. For European defence establishments, this should be deeply concerning. NATO allies share military infrastructure, training exercises happen across the continent, and European forces deploy globally. If private entities can track US troop movements this way, European military operations are equally exposed.

Samantha: So what's the practical response?

Bob: European regulators need to think hard about controls on commercial geospatial data and the AI tools that process it. And European militaries need to invest in counter-surveillance and operational security measures that account for this new reality. The dual-use nature of AI isn't theoretical anymore. It's operational, it's commercial, and it's being used right now.

Meta suspends work with AI startup Mercor after a breach that may have exposed data from OpenAI and Anthropic

Bob: Next up — Meta has indefinitely paused its partnership with Mercor, an AI data contracting startup with a ten-billion-dollar valuation. The reason? A confirmed security breach at Mercor that may have exposed proprietary training data from some of the biggest names in AI, including OpenAI and Anthropic.

Samantha: Ten-billion-dollar valuation and they get breached. That's not a small operation — this is a company that major AI labs trusted with their most sensitive assets. Training data is the crown jewels of any AI company. It's what gives models their competitive edge. If that data has been compromised, the implications are enormous.

Bob: It exposes a fundamental vulnerability in the AI ecosystem. These large labs don't do everything in-house. They rely on networks of contractors for data labelling, curation, and processing. Each of those contractors becomes a potential attack surface. Mercor was essentially a trusted intermediary handling data for multiple competitors simultaneously. A single breach there could affect the entire industry.

Samantha: And for Europe, this is extremely relevant. European AI companies and research institutions often work with similar third-party data providers. Under GDPR, the obligations around data protection are already stringent, but this shows that contractual compliance isn't enough. You need to verify and continuously audit the security practices of every partner in your data supply chain.

Bob: I'd expect European regulators to use this as a case study. The AI Act is coming into force, and incidents like this will shape how enforcement priorities are set. If you're a European AI company, the message is clear — vet your contractors as rigorously as you vet your own systems. Because their breach becomes your breach.

The White House unveils a comprehensive national AI policy with a strong focus on workforce training

Samantha: Staying in the US — the White House has released what it's calling a National AI Policy Framework. The headline focus is on workforce training and education, which is a bit of a departure from the more regulation-heavy approaches we've seen elsewhere.

Bob: It's interesting because it signals a particular philosophy. Rather than leading with restrictions on what AI can and can't do, the US is essentially saying — let's make sure our people are ready for the AI economy. That means retraining programmes, updated educational curricula, partnerships between government and industry to identify the skills that will matter most. It's a supply-side approach to AI governance.

Samantha: Which contrasts pretty sharply with Europe's approach. The EU AI Act is fundamentally about risk classification and compliance. It's about guardrails. The American framework is more about acceleration and adaptation.

Bob: And neither approach is wrong on its own. But there's a lesson here for Europe. The AI Act is robust on safety and ethics, but critics have argued it doesn't do enough on the workforce side. If European workers aren't equipped to build, deploy, and work alongside AI systems, all the regulation in the world won't make Europe competitive. The US framework should push European policymakers to invest more seriously in AI education — from schools all the way through to mid-career retraining. The talent pipeline is just as important as the regulatory pipeline.

Samantha: And honestly, it's probably more urgent. You can write rules relatively quickly. Building a skilled workforce takes a generation.

Elektros promotes patented multi-plug EV charging tech as fuel prices climb

Samantha: Alright, last story. Elektros, a NASDAQ-listed company, is pushing its patented multi-plug EV charging technology. The pitch is simple — faster charging times for electric vehicles. And they're making this push as global gasoline prices keep climbing, which naturally drives more consumer interest in EVs.

Bob: The charging problem has always been one of the biggest barriers to mass EV adoption. People compare filling a petrol tank in three minutes to waiting thirty or forty-five minutes at a charger, and the maths doesn't work for a lot of drivers. What Elektros claims to offer is a multi-plug system that delivers power more efficiently, cutting those wait times significantly.

Samantha: For Europe, this is directly relevant. The EU has set aggressive targets for EV adoption, and member states are investing billions in charging infrastructure. But building chargers isn't enough — they need to be fast, reliable, and widely available. If technologies like what Elektros is developing can genuinely reduce charging times, that accelerates the whole transition.

Bob: The caveat is that we've seen a lot of promises in the EV charging space that haven't always delivered. Elektros is a relatively small company, and the technology still needs to prove itself at scale. But the direction of travel is clear. Rising fuel prices are pushing consumers toward EVs faster than many forecasts predicted, and the companies that solve the charging convenience problem are going to capture enormous value. European charging network operators should be watching this space closely and evaluating whether this kind of multi-plug approach could be integrated into the next generation of European infrastructure.

Samantha: Competition and innovation in this space can only be good for European drivers.

Bob: Couldn't agree more.

Samantha: That's our show for today. We covered: China's AI chip challenge to Nvidia, AI reshaping Bollywood film production, Chinese firms trading AI-powered military intelligence, Meta's security breach with Mercor, the White House AI policy framework, and Elektros pushing faster EV charging.

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.