US $874M AI Chip Push: Europe's Gap Widens

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

Bob: And I'm Bob Russell. Today: the United States pours nearly nine hundred million dollars into chip research, AI patent filings smash through a hundred thousand in a single year, a German equipment maker bets forty-five million euros on advanced chip packaging, OpenAI opens its top models free to a hundred thousand researchers, and to close out, two things you can try today: the encrypted European messenger Threema and the free open-source photo tool Hugin. Let's start with that big chip funding announcement from Washington.

US Commerce Department commits eight hundred seventy-four million dollars to accelerate AI chip research at home.

Samantha: The US Department of Commerce has signed seven letters of intent worth eight hundred seventy-four million dollars in federal money under the CHIPS and Science Act. The target: research into the chips that power AI systems, from the raw silicon to the way data moves between processors. One of the biggest chunks, up to three hundred million dollars, goes to GlobalFoundries for something called silicon photonics. That's a technique that uses light instead of electricity to shuttle data around inside a chip, which matters because moving data is where a huge share of energy and time gets burned in modern AI systems.

Bob: And this is not a one-off cheque. It fits a pattern where Washington keeps stacking public money behind the specific bottlenecks in AI hardware. Faster interconnects, better packaging, more efficient power delivery. The bet is that whoever owns those bottlenecks owns the next decade of AI infrastructure. What's notable is the shift from just funding factories to funding the research pipeline that feeds those factories. That's a longer horizon, and it's aimed squarely at keeping cutting-edge work anchored in the US.

Samantha: For Europe the contrast is uncomfortable. The European Chips Act was pitched with big numbers, but the actual research budgets going into AI-specific hardware here are a fraction of this. Industry groups have argued repeatedly that Europe needs to match not the total dollar figure, but the intensity, the willingness to fund early-stage work at the level that shifts industrial gravity. Meanwhile the practical effect for European chip designers and equipment makers is that their most ambitious customers, and their toughest competitors, are being subsidised on the other side of the Atlantic.

Bob: The other thing worth flagging: silicon photonics is one of the areas where European research groups, particularly in Belgium and the Netherlands, are genuinely world-class. So a US programme aimed at that field will inevitably pull talent, partnerships, and licensing deals westward unless European funders respond in kind. The bottom line is that eight hundred seventy-four million dollars sounds modest against private AI investment, but as targeted state money into a narrow technical frontier, it lands hard.

Global AI patent grants pass one hundred thousand in a single year for the first time.

Samantha: For the first time ever, AI patent grants worldwide broke through the hundred-thousand mark in a single year. The figure for 2025 came in at one hundred seven thousand two hundred seventy-nine, according to the annual report from IFI CLAIMS Patent Services. That's a jump big enough to reset expectations about how fast intellectual property is being generated in this field. The report highlights a particular surge in filings around so-called agentic AI, systems that don't just answer questions but actually carry out multi-step tasks on their own.

Bob: What this really shows is the industrialisation of AI research. A patent grant is not the same as a working product, but at this volume it signals that companies are staking out territory at a scale we simply have not seen before. Chinese and American filers dominate the league tables, and the gap with European applicants has been widening for years. Industry associations have been warning that Europe files strong fundamental research but converts less of it into protected commercial IP, and these numbers put a hard figure on that concern.

Samantha: There is a counter-argument, that patent counts measure activity, not quality, and a lot of these filings will never become meaningful products. Fair enough. But the raw volume still matters for licensing negotiations, for defensive portfolios, and for who gets to sit at the table when standards are set. If European firms want a voice in how agentic AI is regulated and commercialised globally, having a patent portfolio in the game is part of the price of entry.

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.

German equipment maker SUSS MicroTec invests forty-five million euros in advanced chip packaging.

Bob: SUSS MicroTec, a German maker of semiconductor equipment, is putting roughly forty-five million euros into a new application and development centre in Karlsruhe. The focus is advanced packaging, which is the craft of stitching multiple chips together into a single high-performance system. It sounds mundane, but it's become one of the most important frontiers in the industry, because you can only shrink individual chips so far. After that, the gains come from how cleverly you combine them.

Samantha: And this matters for Europe in a very concrete way. The AI boom is driving demand for high-bandwidth memory and tightly integrated processor stacks, and almost all of the leading-edge packaging capacity today sits in Taiwan and South Korea. A European equipment maker expanding its research base in Germany is one of the few genuine industrial anchors the continent has in that supply chain. SUSS MicroTec supplies tools that end up in fabs all over the world, so what happens in Karlsruhe ripples outward.

Bob: Compared to the hundreds of billions being deployed in the US and Asia, forty-five million euros is a modest sum. But focused investment in a narrow specialty is exactly how European industry has historically punched above its weight, think ASML in lithography. The question is whether policy support and customer demand will let SUSS MicroTec and similar firms scale that specialty into serious market share, rather than remaining a niche supplier while the volume growth happens elsewhere.

Samantha: The other angle is talent. A new development centre means engineering jobs, PhD placements, and industrial partnerships with local universities. That kind of anchor helps keep chip expertise in Europe, at a moment when both the US and Asia are actively recruiting from exactly this pool. It's small money by industrial standards, but the strategic footprint is larger than the headline number suggests.

OpenAI opens its frontier models free to one hundred thousand academic researchers worldwide.

Bob: OpenAI has launched a programme called ChatGPT for Academic Researchers, giving free access to its most advanced models across the main ChatGPT product and its coding tool Codex. The pitch is aimed at scientists, mathematicians and engineers, roughly a hundred thousand of them, with higher usage limits, larger context windows for working through long documents, and expanded research features. The goal, according to OpenAI, is to accelerate scientific discovery by putting frontier AI directly into working labs.

Samantha: What this really is, underneath the framing, is a distribution play. Academic researchers are the people who will shape the next generation of AI applications in medicine, materials science, and physics. Getting them onto OpenAI's stack now means the papers, the methods, and the follow-on companies grow up inside that ecosystem. Google, Anthropic and Meta have all run similar programmes at various points, and the competition for research mindshare is genuinely intense.

Bob: For European universities, the offer is attractive and complicated at the same time. Attractive, because access to frontier models is expensive and most research budgets on this side of the Atlantic are stretched thin. Complicated, because sending research data through an American AI provider raises the same questions European institutions have been wrestling with for years around data sovereignty and the EU AI Act's transparency rules that took effect last week. Universities will have to weigh the scientific upside against the compliance homework.

Samantha: And there is a longer-term concern that policy analysts keep raising: free access today can become expensive dependency tomorrow. If a generation of European researchers builds their workflows around one provider's tools, switching costs down the line become real. Some European research consortia have argued for building shared, publicly funded compute infrastructure precisely to avoid this pattern. Whether that happens fast enough to compete with a hundred thousand free frontier seats is another matter entirely.

The encrypted Swiss messenger Threema gives Europeans a chat app that needs no phone number.

Bob: To close out, two things worth trying today. First up, Threema, a paid messenger app made in Switzerland that has quietly built a following across Europe as a serious WhatsApp alternative. The distinctive thing about it is that you don't need to hand over a phone number or an email address to sign up. Instead you get a random identifier, and that becomes your identity. Everything is end-to-end encrypted, and the servers sit in Switzerland under Swiss data protection law.

Samantha: It's not free, there's a one-off purchase of a few euros, which the developers argue is precisely the point. If you're not paying, you're usually the product, and Threema's whole pitch is that its business model doesn't depend on knowing anything about you. The app supports group chats, voice and video calls, file sharing, and it has a separate work version that businesses and public institutions across Germany, Austria and Switzerland already use, including some government agencies that specifically wanted to keep sensitive communication out of American hands.

Bob: Users report that setting it up takes about five minutes on iPhone or Android, and there's a desktop version if you want to keep chatting from your laptop. Reviewers often praise the metadata protection: even the fact that two people are talking to each other is not stored on Threema's servers. The main downside is the same one every alternative messenger faces, which is convincing the people you actually want to chat with to install it too. But if you have a small group of contacts who care about privacy, or if you want a work channel separate from your personal WhatsApp, it's a genuinely usable option.

Samantha: It runs on iPhone, Android, Windows, Mac, and Linux, and payment is a one-time purchase rather than a subscription. For anyone who has been looking for a European-hosted alternative to the big American messengers, Threema is one of the more established names on that list.

The free open-source tool Hugin stitches your holiday photos into stunning panoramas.

Bob: The second thing to try today is Hugin, a free open-source tool that turns a series of overlapping photos into a single panoramic image. If you've ever tried to capture a wide landscape on holiday and ended up with three or four separate shots that don't quite line up, Hugin is the tool that stitches them together properly. It runs on Windows, Mac and Linux, and it has been developed by a community of volunteers for more than fifteen years.

Samantha: What makes it interesting is how much control it gives you. The automatic mode will handle a straightforward panorama in a couple of clicks, but if you have tricky lighting, moving objects, or you want to build something ambitious like a full three hundred sixty degree virtual tour, the manual controls are genuinely powerful. Photography enthusiasts have used it for everything from wedding albums to architectural documentation.

Bob: Reviewers often note that the interface looks dated, which is fair, it does. But once you get past the first few menus, the results speak for themselves. And crucially, everything runs on your own computer. Your photos never touch a cloud service, which matters if you're stitching together images of your home, your family, or anywhere you'd rather not upload to a big American platform.

Samantha: It's a good example of the kind of quiet, mature open-source software that keeps working year after year without asking for a subscription or an account. Download it from the project's website, point it at a folder of overlapping photos, and you'll usually have a finished panorama within a few minutes.

Samantha: Today we covered: the eight hundred seventy-four million dollar US chip research push, AI patents breaking the hundred thousand mark, SUSS MicroTec's forty-five million euro packaging bet in Karlsruhe, OpenAI opening its frontier models to a hundred thousand researchers, and two things to try yourself: the Swiss messenger Threema, and the free panorama tool Hugin.

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.