ASML chip tool shortfall, XPENG pitches Europe on AI cars

Samantha: Welcome to The State of Tech, The European Edition, Thursday October fifteen, 2026. I'm Samantha Lawrence.

Bob: And I'm Bob Russell. Today: ASML warns the world is running short on the machines that make AI chips, Snowflake pulls AI and data into one platform, XPENG pitches Europe on deeper car and robot partnerships, and a smart home merger reshapes a crowded market. Then two things to try yourself: the open-source photo organiser Mylio-free alternative Tropy running on your own machine, and the free focus app Serene. Let's start with ASML.

ASML warns it cannot build enough chip machines to keep up with the AI boom.

Samantha: ASML, the Dutch company that makes the machines that make advanced chips, is warning of a supply shortfall over the next two years. The reason is simple: the global spending on artificial intelligence has gone through the roof, and demand for cutting-edge chips is outrunning what ASML can physically produce. The company lifted its own sales forecast for this year, but says it is now actively looking for ways to push capacity for 2028 beyond what it previously told investors. In plain terms: the single most important bottleneck in the AI supply chain just flashed red.

Bob: And this really matters because ASML sits in a unique spot. It is the only company in the world that builds the most advanced lithography machines, the ones that print the finest circuits onto silicon. If ASML cannot ship fast enough, the whole chain behind it slows down. That includes TSMC in Taiwan, Samsung in Korea, Intel in the United States, and every AI data centre that depends on those chips. The warning is not that the machines will stop coming, it is that the queue is getting longer.

Samantha: What this signals is a shift in the story around AI. For two years the narrative has been about model breakthroughs and software races. Now the limiting factor is hardware, and specifically a handful of very physical things: lithography tools, cooling systems, power grids. ASML is effectively saying the industry wants more than physics and factories can deliver on the current timeline. That changes who holds leverage in negotiations and which customers get served first.

Bob: There is also a strategic layer. ASML is navigating export rules on which tools can go to China, pressure from Washington and Brussels, and domestic political debate in the Netherlands about keeping research and talent at home. A shortfall sharpens all of that. Governments start asking who gets priority. Customers start locking in slots years in advance. And smaller players, universities, startups, risk being pushed to the back of the line while hyperscalers write the biggest cheques.

Samantha: For Europe this is a double-edged story. On one hand, ASML is a crown jewel, proof that the continent still owns a critical piece of the global tech stack. On the other hand, if the machines are rationed, European chip ambitions, including the Chips Act factories being built in Germany and elsewhere, could face delays. Brussels has staked a lot on homegrown semiconductor capacity by the end of the decade, and that plan assumes ASML can equip those fabs on schedule. Today's warning puts a question mark next to that assumption.

Snowflake pulls data, AI training and deployment into one governed platform.

Bob: Snowflake is bringing together everything a company needs to build and run AI into a single environment. That means the data itself, the training of machine learning models, and the deployment of those models into real products, all inside one governed platform. The idea is to shorten the painful journey from AI experiment to AI in production, which today often involves stitching together half a dozen different tools and losing track of who touched what.

Samantha: The company's chief data and AI officer, Anahita Tafvizi, is talking about this at TechCrunch Disrupt, and the framing is telling. For the past two years, enterprise AI has been stuck in pilot purgatory: lots of demos, few deployments that actually make it to customers or employees. Snowflake is betting that the blocker is not model quality anymore, it is plumbing and governance. Who can access which data, how is a model monitored once it is live, where are the audit trails when something goes wrong.

Bob: It is also a competitive move. Databricks, Microsoft, Google and Amazon are all pushing the same message, that the winning platform will be the one where data and AI live together with proper controls. For buyers that is good news, because it should mean less duct tape and lower integration costs. The risk is lock-in: once your data, your models and your governance all sit with one vendor, switching gets expensive. Companies will have to weigh convenience against flexibility.

Samantha: In Europe the governance piece is not optional. The AI Act is now phasing in real obligations around transparency, risk classification and documentation, and GDPR still sets a high bar for anything touching personal data. A platform that can show, line by line, which dataset trained which model, and who approved its release, becomes genuinely useful for compliance teams. The flip side is sovereignty: European regulators and customers will want to know exactly where that data sits, and whether it ever leaves the continent.

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.

XPENG's chief executive tells Paris that Europe should build cars and robots with Chinese partners.

Bob: At the Paris Automotive Summit, XPENG's chairman and chief executive He Xiaopeng became the first Chinese car boss to deliver a keynote at the event. His pitch was straightforward: the industry should collaborate more deeply on artificial intelligence across vehicles, software and even humanoid robots. XPENG wants to open up what it calls its full-stack engineering, from vehicle platforms to what the company labels physical AI, meaning the software and sensors that let a machine move through the real world.

Samantha: Two concrete pieces came with the speech. First, XPENG's assisted-driving system, which it calls Navigation Guided Pilot, is being lined up for a global push in 2027. Second, the company gave an update on its humanoid robot IRON, with mass production targeted for late this year. Taken together, that is a Chinese carmaker saying it does not just want to export vehicles, it wants to export the underlying AI stack and share it with European partners.

Bob: The context here is tense. The European Union has imposed tariffs on Chinese electric vehicles, there are ongoing investigations into subsidies, and several national governments are nervous about data flowing back to China from connected cars. XPENG's answer is essentially to lean into partnership rather than confrontation: come work with us, use our platforms, build locally. Whether European carmakers, who are themselves struggling with the AI software layer, see that as an opportunity or a threat is the open question.

Samantha: For European consumers, the practical effect could be more Chinese-designed driver-assist features showing up in cars sold here, potentially through joint ventures. For European regulators, it raises familiar questions about data, cybersecurity and who certifies these AI systems as safe on public roads. And for Europe's own car industry, the message from Paris is uncomfortable: the companies that used to set the pace are now being invited to partner with a newcomer that writes its own AI from the ground up.

SKYX closes its merger with Deako to build a bigger smart home player.

Bob: SKYX, which makes smart home ceiling fixtures and lighting, has closed its merger with Deako, another smart home company with a strong foothold among American home builders. The combined entity gives existing SKYX shareholders about eighty-four percent ownership, and brings in Deako's builder relationships, meaning its products are already being installed when new houses go up rather than sold afterwards as upgrades.

Samantha: This is a consolidation story in a market that has been fragmented for a long time. Smart home products from different brands have historically struggled to work together, and consumers have been left juggling multiple apps and hubs. A larger combined player can push for standardisation, negotiate better with chip and sensor suppliers, and crucially, get its hardware baked into new homes from day one rather than fighting for shelf space at retail.

Bob: The question is whether scale actually translates into a better experience. Smart home mergers have a mixed track record: companies combine, product lines overlap, and consumers end up with discontinued devices and orphaned apps. The real test will be whether SKYX and Deako can keep both customer bases supported while building a coherent next-generation platform, especially as rivals like Apple, Amazon and Google keep pouring resources into their own ecosystems.

Samantha: European buyers mostly interact with smart home gear through Matter, the shared standard backed by Apple, Google, Amazon and Samsung, plus local players like Bosch and Signify. A larger US-focused SKYX does not instantly change what shoppers in Berlin or Barcelona see on the shelf. But it does signal where the market is heading: fewer, bigger brands, more integration with new-build housing, and more pressure on independent European smart home startups to find a niche or get acquired.

Tropy is a free research photo organiser that turns thousands of archive and family photographs into a searchable collection on your own computer.

Bob: To close out, two things you can actually use today. The first is Tropy, a free open-source tool for organising photographs, built for researchers and historians but just as handy for anyone with a mountain of pictures. You import your photos, group them into items, and add notes and metadata so that everything can be found again later. No subscription, no account, and your collection stays on your own machine.

Samantha: Tropy comes from the Roy Rosenzweig Center for History and New Media at George Mason University, and it is funded by the Andrew W. Mellon Foundation, so it is a serious, well-supported project rather than a weekend experiment. It was designed for people who photograph documents in archives, but it works equally well for family photos. You can describe each item with fields like date, place and source, add tags and notes, and then search across the whole collection. It also lets you export your work, so you are never locked in.

Bob: The practical appeal is simple. Many households have twenty years of photos scattered across old phones, hard drives and half-forgotten cloud accounts. Tropy gives you one place to gather the important ones, with the context that usually gets lost: who is in the picture, where it was taken, which letter or album it belonged to. Instead of a pile of files called IMG underscore 4072, you get a collection you can actually search and trust.

Samantha: For European listeners the appeal is concrete. The program runs locally, so your photos and notes live on your own computer, in your own home, rather than on a server somewhere you do not control. It is free to download, and installation takes only a few minutes. It works on Windows, Mac and Linux, and the project provides clear documentation and a friendly user forum if you get stuck. Historians, archivists and students across Europe already use it, so there is plenty of advice out there for anyone starting a family archive.

The free focus app Serene blocks distracting sites and apps for a set work session.

Bob: The second pick is Serene, a focus app for Mac and Windows that helps you actually finish a single task. You set a goal for a session, pick how long you want to work, and Serene blocks the websites and apps you have flagged as distractions for that period. It also has a soft timer, gentle background sounds, and a daily planning view that nudges you to pick one or two meaningful goals rather than a shopping list of twenty.

Samantha: What makes Serene different from the dozens of other focus tools is the opinionated design. It assumes you do not need another productivity dashboard, you need help starting and finishing one thing at a time. The free tier covers the basics most people want: site blocking, session timers and daily goals. Reviewers consistently praise how calm the interface feels, which matters, because a loud productivity app is itself a distraction.

Bob: It is a good complement to the Tropy tip, because both are about reclaiming a bit of control. One keeps your memories in your own hands, the other keeps your attention there. Serene is a quick install, works offline, and you can test the full experience in an afternoon without signing up for anything long-term.

Samantha: Today we covered: ASML's warning about a chip-tool shortfall, Snowflake unifying AI and data on one platform, XPENG's push for European car and robot partnerships, the SKYX and Deako smart home merger, and two things to try yourself: the self-hosted photo gallery Tropy and the focus app Serene.

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.