EU AI Act Enforcement: What August 2026 Means for You

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

Bob: And I'm Bob Russell. Today: global AI spending hits two and a half trillion dollars, the EU AI Act enforcement clock starts ticking, Apple ditches its cash neutral policy to chase AI, Applied Materials buys NEXX for advanced packaging, Nvidia's China share collapses to zero, and an AI tool helps astronomers confirm over a hundred exoplanets. Let's start with the money.

Global AI spending projected to hit two and a half trillion dollars in 2026, more than the entire previous decade combined.

Samantha: Two and a half trillion dollars. That's the forecast for global AI spending in 2026 alone, and it's bigger than every dollar spent on AI in the last ten years put together. Amazon, Microsoft and Alphabet are each committing hundreds of billions to AI infrastructure this year. The capital is flooding in, and the demand is broadening from hyperscalers to enterprises in every sector.

Bob: And the curve is genuinely vertical. A decade of cumulative spending, eclipsed in twelve months. What stands out is the breadth: it's not just chips and data centres, it's energy contracts, cooling, networking, custom silicon, software platforms, and the services layer on top. Every supplier in that chain is suddenly strategic.

Samantha: Which is why we're seeing power purchase agreements with nuclear operators and grid upgrades being negotiated like cloud contracts. The bottleneck is shifting from GPUs to electrons.

Bob: And to permits. Building a gigawatt-scale data centre in Europe takes years longer than in Texas or Virginia, and that lead time is becoming a competitive factor in itself.

Samantha: The forecast comes amid Q1 earnings showing the big four hyperscalers raising capex guidance again. Alphabet alone is on track for over ninety billion in capital spending this year, most of it AI related.

Bob: And Meta, which isn't a cloud provider in the traditional sense, is matching them dollar for dollar on infrastructure. That tells you the spending isn't just about selling AI services, it's about owning the stack to train frontier models.

Samantha: The bottom line is that this scale of investment creates an entirely new industrial base. It also concentrates power in a handful of US firms, which has knock-on effects for everyone else.

Bob: And it raises the bar for anyone trying to compete. If your rival is spending a hundred billion on compute, your ten billion R&D budget suddenly looks quaint.

Samantha: For Europe, this is opportunity and exposure in equal measure. European AI startups, chip designers and research labs will see real demand. But the lion's share of that two and a half trillion flows through American balance sheets, and Europe's sovereign cloud ambitions look modest next to it.

Bob: Brussels is going to have to decide whether to subsidise capacity, regulate access, or both. The InvestAI fund and the gigafactory plans are a start, but the gap with US capex is widening, not closing.

EU AI Act enforcement begins August 2026 with mandatory risk classification, red-team testing, and transparency reporting.

Bob: Speaking of regulation, the EU AI Act enforcement begins this August. From that date, organisations operating in Europe must classify their AI systems by risk level, build oversight plans, run red-team tests for vulnerabilities, and publish detailed transparency information. High-risk applications in employment, credit, education, and public services face strict conformity assessments and ongoing post-market monitoring.

Samantha: This is the moment AI regulation stops being a policy debate and becomes a compliance line item. Every general counsel in Europe has been preparing for this, and a lot of mid-size firms are still behind.

Bob: The red-team requirement is the one I'd flag. It's not just paperwork, it requires genuine adversarial testing of your model, documented, repeatable, and reviewable by regulators. That's a real engineering commitment.

Samantha: Fines can reach up to seven percent of global turnover for the most serious violations. That's higher than GDPR, and it applies to providers and deployers, not just developers.

Bob: Which means a French bank deploying an American foundation model is on the hook for compliance, even if the model was trained in California. That extraterritorial reach is exactly the GDPR playbook.

Samantha: Key takeaway: the AI Act will likely shape how AI is deployed worldwide, the same way GDPR reshaped global privacy practice. Companies tend to standardise on the strictest rules they face.

Bob: For European firms, the short-term pain is the compliance cost. The longer-term upside is trust. If your AI product is AI Act compliant by design, that becomes a selling point, especially in regulated sectors.

Samantha: And the smaller players need help. The Commission has promised guidance and sandboxes, but small and medium businesses are nervous about the cost of conformity assessments.

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.

Apple drops its net cash neutral policy after fifteen years, signalling aggressive AI infrastructure spending and acquisitions.

Bob: Apple has dropped its net cash neutral policy. For nearly fifteen years, Apple ran a strategy of returning roughly all surplus cash to shareholders through buybacks and dividends, more than a trillion dollars in total. That's now over. Incoming CEO John Ternus gets the green light to spend.

Samantha: The shift was announced alongside Q2 2026 results, and analysts read it as preparation for big AI infrastructure investments and possibly a major acquisition. With Meta and Alphabet committing hundreds of billions to AI, Apple's restraint had become a competitive weakness.

Bob: The market has been pricing Apple as the AI laggard for over a year. Apple Intelligence underwhelmed, the Siri overhaul slipped, and rivals shipped features Apple was still demoing. This policy change is essentially an admission that the old model can't compete.

Samantha: Apple is sitting on roughly fifty billion in net cash and generates close to a hundred billion in free cash flow annually. That's enormous firepower if they choose to deploy it.

Bob: The acquisition question is the interesting one. Apple historically buys small. A ten-billion-dollar deal would be a record by a wide margin. But to close the AI gap, that may be exactly what's needed.

Samantha: With Apple in the bidding, AI talent and startups get another deep-pocketed buyer, and prices go up across the board.

Bob: European AI firms with strong on-device or privacy-preserving technology fit Apple's profile particularly well. There are several candidates in France, the UK and the Nordics that would suddenly be very attractive.

Samantha: And for European users, faster AI features in iPhones and Macs, but the data and consent questions get harder, especially under the AI Act we just discussed.

Applied Materials acquires NEXX to expand panel-level advanced packaging for AI accelerators.

Samantha: Applied Materials has signed a definitive agreement to buy NEXX, a key supplier of large-area advanced packaging deposition equipment. The deal expands Applied's panel-level packaging portfolio, which is becoming critical as AI accelerators move beyond the limits of traditional silicon wafers.

Bob: The technical story matters here. Modern AI chips are chiplet designs, multiple GPUs and high-bandwidth memory stacks integrated into one package. As those packages get bigger, the industry is shifting from round wafers to rectangular panels, which are more efficient at scale. NEXX brings electrochemical deposition expertise for exactly that transition.

Samantha: This is the unglamorous middle of the AI stack, but it's where bottlenecks are forming. Advanced packaging capacity at TSMC has been the limiting factor on Nvidia's shipments for nearly two years.

Bob: Exactly the kind of deal that doesn't make the front page but moves the industry. Whoever controls advanced packaging tooling controls how fast AI chips can scale.

Samantha: More efficient packaging means more powerful, more energy-efficient AI accelerators, and that flows through to every cloud provider and every AI product.

Bob: Europe doesn't have a major advanced packaging player, but ASML remains the linchpin upstream in lithography. The combination of US tooling consolidation and Dutch lithography dominance is what makes the AI chip industry possible.

Samantha: And it underscores how dependent European AI ambitions are on a global supply chain that runs through Eindhoven, Taipei and Santa Clara.

Nvidia's China AI accelerator share falls to zero as US export controls bite, while Asian partners rally.

Bob: Nvidia's market share for AI accelerators in China has fallen to effectively zero. Tightening US export controls have closed off the market entirely, after years of restrictions, redesigns and workarounds. At the same time, Nvidia's Asian manufacturing partners are rallying as physical AI and robotics push expands, with those partners now accounting for around ninety percent of Nvidia's production costs.

Samantha: Zero is a remarkable number. Two years ago China was a quarter of Nvidia's data centre revenue. The redesigned H20 chip kept some business alive, but the latest tightening shut the door entirely.

Bob: And the Chinese response has been to accelerate domestic alternatives. Huawei's Ascend line is closing the gap faster than expected, and Beijing is funnelling state money into the domestic AI chip industry.

Samantha: We're watching a genuine bifurcation. Two parallel AI hardware stacks, one Western, one Chinese, with limited interoperability.

Bob: The bottom line is fragmentation is expensive. Duplicated R&D, smaller addressable markets, slower standards. The whole industry pays for it eventually.

Samantha: For Europe, the lesson is to invest in domestic capability where possible and diversify partnerships where not. The European Chips Act is a step, but the scale gap with the US and Asia remains huge.

Bob: And European companies exporting to China face their own export-control headaches, since US technology is embedded in almost everything they make.

AI tool RAVEN helps Warwick astronomers confirm over a hundred exoplanets, including thirty-one new worlds.

Samantha: And to close on a lighter note, astronomers at the University of Warwick have used a new AI system called RAVEN to confirm more than a hundred exoplanets, including thirty-one previously unknown worlds. RAVEN analysed four years of data from NASA's TESS mission, which watches millions of stars for tiny brightness dips that betray a planet passing in front.

Bob: The findings are published in MNRAS, focusing on short-period planets, the ones that whip around their stars in days. Confirming exoplanets is painstaking work. RAVEN dramatically speeds up that process by sifting through observations at a scale humans simply can't match.

Samantha: TESS has been generating candidates faster than astronomers can vet them. RAVEN closes that gap, and it does it with remarkable accuracy on the short-period candidates where false positives are common.

Bob: It's a quiet but important shift. AI is moving from hype headlines to genuine scientific productivity, and exoplanet hunting is one of the cleanest examples.

Samantha: Every new world we confirm sharpens our picture of how planetary systems form. Thirty-one new ones in a single paper is a meaningful jump.

Bob: ESA's CHEOPS is already running, and PLATO launches in 2026 to hunt Earth-sized planets around Sun-like stars. Tools like RAVEN slot directly into those pipelines and could deliver discoveries on day one.

Samantha: Warwick leading on this is a nice reminder that European universities remain genuinely world-class on AI applied to science, even when the trillion-dollar headlines are elsewhere.

Bob: And it's the kind of AI story that doesn't worry anyone. No layoffs, no copyright lawsuits, just more planets.

Samantha: Today we covered: the two and a half trillion dollar AI spending forecast, the EU AI Act enforcement starting in August, Apple ending its net cash neutral policy, Applied Materials buying NEXX, Nvidia's China share collapsing to zero, and the RAVEN AI tool finding new exoplanets.

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.