Nvidia & Marvell AI Alliance, Quantum Bitcoin Threat, Iran Tech Strike

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

Bob: And I'm Bob Russell. Today: Nvidia pours two billion dollars into a new AI alliance with Marvell. A brand new monitoring tool launches to keep AI agents in check. AI is coming for the corner office, not the factory floor. Quantum computers could crack Bitcoin way sooner than we thought. Iran threatens strikes on major US tech giants. And Onsemi hikes semiconductor prices starting today. Let's start with that massive Nvidia deal.

Nvidia and Marvell join forces in a $2 billion AI infrastructure deal

Samantha: So Nvidia is writing a very large cheque. Two billion dollars into Marvell Technology, and they're calling it an AI alliance. That's not pocket change, even for Jensen Huang.

Bob: It's a strategic move, and the focus is really interesting. This isn't just about GPUs. They're going deep into silicon photonics — that's the technology that uses light instead of electrical signals to move data around. And when you're running massive AI workloads, the bottleneck is often not the compute itself but getting data between chips and between servers fast enough. Silicon photonics solves that. High speed, low energy. It's the plumbing of the AI revolution, if you want to think of it that way.

Samantha: Plumbing that costs two billion dollars. But Jensen Huang framed this as being about helping customers build specialised AI compute. What does that actually mean in practice?

Bob: It means Nvidia wants to own more of the stack. Right now, if you're a hyperscaler or a telecom company, you're piecing together infrastructure from multiple vendors. This alliance essentially lets Nvidia and Marvell offer a more integrated package — compute, networking, data transmission — all optimised to work together. And the telecom angle matters too. 5G networks, edge computing — all of that needs this kind of high-speed, efficient connectivity.

Samantha: And for Europe, this is directly relevant. The EU has been pushing hard on sovereign AI, building its own compute capacity, funding supercomputers. But Europe still depends heavily on American and Asian companies for the foundational chips and infrastructure.

Bob: That's the tension. Europe wants digital sovereignty, but the building blocks keep getting consolidated among a handful of non-European players. A deal like this could actually benefit European AI initiatives if it drives down costs and improves availability. But it also deepens the dependency. European policymakers will be watching closely to see whether this kind of alliance creates open ecosystems or walled gardens.

Samantha: And whether Europe's own chip ambitions under the European Chips Act can keep pace. Alright, let's move on.

New tool launches to monitor AI agents as agentic AI takes off

Samantha: So we're in this world now where AI agents are doing real work — processing invoices, handling customer service, planning projects. And a company called Codenotary has just launched something called AgentMon. It's basically a monitoring tool to watch what these AI agents are actually doing.

Bob: And the timing makes perfect sense. We've been talking about agentic AI for months, but now these systems are actually deployed in production environments. They're making decisions, accessing databases, interacting with customers. And the question everyone's asking is — how do you know they're behaving properly? Are they leaking data? Are they racking up costs you didn't expect? Are they compliant with your security policies?

Samantha: It's like hiring a thousand new employees and having zero managers.

Bob: That's a great analogy. And AgentMon is essentially the manager. It tracks agent behaviour, flags anomalies, monitors for data leaks and compliance violations. There's also some interesting research from Georgia Tech showing that older adults trust AI agents more when the AI explains its reasoning. Which sounds obvious, but a lot of these systems are still black boxes.

Samantha: And then there's Samsara, which is showing off what they call physical AI — autonomous trucks, robots working alongside humans. So the agent concept is moving beyond software into the physical world.

Bob: Which raises the stakes considerably. A software agent that makes a bad decision might cost you money. A physical agent that makes a bad decision could hurt someone.

Samantha: For Europe, this is huge. The EU AI Act is arguably the most comprehensive AI regulation in the world, and it has very specific requirements around transparency, accountability, and risk management. Tools like AgentMon aren't just nice to have for European companies — they're practically mandatory if you want to deploy agentic AI and stay compliant. European businesses are already using these autonomous systems for invoice processing and supply chain management, so the governance infrastructure needs to keep up.

Bob: And the explainability research feeds directly into the AI Act's transparency requirements. Europe has essentially legislated what Georgia Tech is now proving empirically — that people need to understand why an AI made a decision.

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.

AI is outperforming executives at strategic decision-making

Bob: Here's a story that should make a few boardrooms uncomfortable. The Harvard Business Review ran an experiment where AI went head-to-head with human participants on strategic decision-making for virtual companies. And the AI won.

Samantha: Not just won. Outperformed. We're not talking about AI replacing warehouse workers or data entry clerks. This is AI doing the job of executives, strategists, and consultants. The people who thought they were safe.

Bob: There's been this assumption baked into a lot of economic predictions that automation would primarily affect manual and routine cognitive work. Factory jobs, call centres, basic accounting. And AI would trickle up slowly. But what we're seeing is almost the reverse. AI is really good at synthesising large amounts of information, identifying patterns, and making strategic recommendations. That's literally what consultants and strategists get paid to do.

Samantha: And the big tech companies are leaning into this. They're building autonomous AI agents specifically designed to plan and execute complex business tasks with minimal human oversight. So it's not just a research finding — it's becoming a product category.

Bob: The labour market implications are significant. If you're a mid-career professional with a strategy or consulting background, the value proposition is shifting. It's no longer enough to analyse and recommend. You need to be able to do things AI can't — build relationships, navigate politics, exercise ethical judgment in ambiguous situations.

Samantha: For European economies, this is a real challenge. Europe has a large services sector. Consulting, financial advisory, corporate strategy — these are big employers, especially in cities like London, Paris, Frankfurt, and Amsterdam. If AI starts eating into those roles, you're looking at displacement in precisely the demographic that tends to have high salaries, high mortgages, and high expectations.

Bob: European policymakers have been focused on retraining programmes for manufacturing and lower-skilled workers. They may need to pivot and start thinking about retraining and upskilling for white-collar professionals too. And honestly, European universities and business schools need to rethink their curricula. Training people to do what AI already does better is not a great business model.

Samantha: Uncomfortable but true. Next up — quantum computing and Bitcoin.

Quantum computers may crack Bitcoin encryption far sooner than expected

Bob: Researchers at Caltech and a company called Oratomic have published findings suggesting that a quantum computer capable of breaking Bitcoin's encryption might need only about ten thousand qubits. Previous estimates said you'd need over a million.

Samantha: That's a massive difference. We went from "this is a distant theoretical threat" to "we might need to worry about this within a few years."

Bob: The paper came out on March 31st, and Google researchers have separately been sounding the alarm, urging the crypto world to shift to post-quantum encryption. The core issue is that Bitcoin and most blockchain systems rely on elliptic curve cryptography, which is mathematically secure against classical computers but vulnerable to sufficiently powerful quantum machines.

Samantha: So if someone builds a ten-thousand-qubit fault-tolerant quantum computer, they could theoretically forge transactions, steal coins, or undermine the entire trust model of blockchain.

Bob: And we're not that far from ten thousand qubits. IBM, Google, and several European players are on aggressive roadmaps. We could be looking at that capability within five to eight years, maybe sooner.

Samantha: Europe has a massive stake here. The EU has poured billions into becoming a quantum superpower through the Quantum Technologies Flagship programme. European companies and research labs are at the forefront of quantum hardware development. So Europe is simultaneously building the tool that could break crypto and needing to defend against it.

Bob: European financial institutions — banks, exchanges, custodians — need to start migrating to post-quantum cryptographic standards now. Not next year, now. The transition takes time, and if you wait until quantum computers actually break something, it's too late. National security agencies across Europe are already aware of this, but the private sector needs to catch up fast.

Samantha: Alright, let's shift gears to geopolitics.

Iran's Revolutionary Guard threatens to strike US tech companies' regional operations

Samantha: Iran's IRGC has directly threatened to target around eighteen American tech companies operating in the Middle East. We're talking Microsoft, Google, Apple, Tesla. They said the campaign would begin at eight PM Tehran time — today.

Bob: The framing from the IRGC is that this is retaliation for what they call assassination operations against Iranian leaders. Whether that means cyberattacks, physical disruption, or both remains unclear. But the intent is to put economic pressure on American commercial interests across the Gulf region.

Samantha: And the date is notable — April 1st. There's naturally some scepticism about the timing. But security analysts are taking it seriously because the IRGC has a track record of following through on threats, even if on a different timeline than announced.

Bob: The immediate concern is for these companies' data centres, offices, and supply chains in the region. The Gulf states host significant tech infrastructure. Dubai, for instance, is a major hub.

Samantha: For European companies, the ripple effects could be significant. Many European businesses rely on the same cloud infrastructure from Microsoft and Google in the Middle East. If those services get disrupted, European companies with operations in the region are affected too. And European firms with their own presence in the Gulf need to be thinking about contingency plans.

Bob: There's also the cybersecurity dimension. If the IRGC launches cyberattacks on American tech platforms, those platforms serve European customers too. A breach or outage doesn't respect geographic boundaries. European cybersecurity agencies should be on heightened alert.

Samantha: And our last story today — semiconductor prices going up.

Onsemi raises chip prices effective today, citing surging costs

Bob: Onsemi has announced a price increase on select semiconductor products, and it kicks in today, April 1st. They're pointing to rising raw material costs, energy prices, manufacturing expenses, and infrastructure investment as the drivers.

Samantha: This is one of those stories that sounds dry but touches everything. Semiconductors go into cars, industrial equipment, consumer electronics, data centres. If chip prices go up, the cost cascades through the entire economy.

Bob: Onsemi is a major supplier in the power semiconductor space. They're big in automotive, which is especially relevant for Europe. European carmakers — Volkswagen, BMW, Stellantis — are all deep into electrification, and electric vehicles are incredibly chip-hungry. Power semiconductors are critical components in EV drivetrains, charging systems, and battery management.

Samantha: So European consumers could see higher car prices, higher appliance prices, higher costs for industrial goods. At a time when inflation has been a persistent headache across the eurozone.

Bob: And this highlights the ongoing vulnerability of European industry to global semiconductor supply chain dynamics. The European Chips Act was supposed to address this by building domestic production capacity, but those fabs take years to come online. In the meantime, European manufacturers are at the mercy of pricing decisions made in Phoenix, Arizona, or wherever Onsemi sets its rates.

Samantha: It strengthens the case for diversified sourcing. European companies should be actively cultivating relationships with multiple suppliers and investing in alternatives. Relying on a single source for critical components is a recipe for exactly this kind of squeeze.

Bob: The broader lesson is that semiconductor economics are now macroeconomics. Every price hike in this space has consequences that extend far beyond the tech sector.

Samantha: That wraps us up for today. We covered: Nvidia's two-billion-dollar AI alliance with Marvell, monitoring tools for AI agents, AI outperforming executives at strategy, quantum threats to Bitcoin, Iran targeting US tech companies, and Onsemi raising chip prices.

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.