AI flies real F-16: DARPA test & Alibaba Qwen 3.8-Max

Samantha: Welcome to The State of Tech, The European Edition, Sunday August ninth, 2026. I'm Samantha Lawrence.

Bob: And I'm Bob Russell. Today: an AI takes the stick of a real US fighter jet, Alibaba drops a new flagship AI model aimed straight at Silicon Valley, a room-temperature quantum breakthrough out of Louisiana that could shrink quantum tech onto a chip, Google's fifteen-billion-dollar Indian data centre runs into an environmental wall, and to close out, two things you can actually use today: the open-source knowledge tool Logseq, and the free European navigation app HERE WeGo. Let's start with the fighter jet.

The US Air Force flies a real F-16 fighter jet with artificial intelligence fully in control.

Samantha: DARPA has flown an F-16 in the real sky, not a simulator, with artificial intelligence handling the controls from take-off onwards. A human pilot sat in the cockpit as a safety net, but the aircraft itself was flown by software known as the VENOM Autonomy Kit. This is the first time an operational American fighter has been handed over to an AI in an actual flight test rather than a computer model, and it lands squarely in the middle of the debate about how much authority we give machines in combat.

Bob: The strategic logic behind this is simple and uncomfortable. Modern fighter jets cost more than a hundred million each, and pilots take years to train. If you can pair one human pilot with a swarm of AI-flown wingmen, you scale air power without scaling human risk. The US calls it collaborative combat aircraft, and this test is a real-world milestone on that path. What it doesn't answer is the harder question: at what point can an autonomous jet decide to fire, and who is legally responsible when it does. Right now the rule is a human stays in the loop for weapons release, but the pressure to loosen that will grow the moment adversaries move faster.

Samantha: For Europe this cuts two ways. Britain, France and Germany are all working on their own next-generation fighter programmes, GCAP and FCAS, and both assume autonomous drone wingmen. Watching the Americans fly a manned jet under AI control puts real pressure on those programmes to catch up, or risk buying American again. At the same time, European defence ministries have to answer to parliaments that are far more sceptical about lethal autonomous systems than Washington is.

Bob: And there's a regulatory gap here that nobody has closed. The EU AI Act explicitly carves out military use, which means the strictest AI rulebook in the world simply doesn't apply the moment a system is worn in uniform. That leaves individual member states writing their own guardrails, and most haven't. Expect this test to accelerate a conversation in Brussels and NATO about common standards for autonomous combat systems, because the technology is now clearly ahead of the policy.

Alibaba launches Qwen 3.8-Max and opens the weights, challenging OpenAI and Anthropic head-on.

Samantha: Alibaba has unveiled Qwen 3.8-Max, its newest large AI model, aimed particularly at software development and team collaboration. The interesting part isn't the benchmark scores, which every lab claims to lead on these days. It's that Alibaba is releasing the model weights, which means companies can download it and run it on their own servers instead of paying a subscription to a US provider. That's a very different business model from OpenAI or Anthropic.

Bob: And it fits a pattern we've been tracking. Chinese labs have decided that if they can't out-spend the Americans, they'll out-open them. DeepSeek did it earlier this year, Moonshot did it with Kimi, and now Alibaba is putting its top-tier model into the same category. For European businesses that don't want their code and customer data flowing through American cloud services, that's genuinely attractive. A hospital or a bank can run this behind its own firewall and keep sensitive information on the continent.

Samantha: The catch is that "open weights" is not the same as "no strings attached". Companies still need to check the licence terms, and any model coming out of China raises questions about training data and what the model might or might not say on politically sensitive topics. European regulators are also going to want to know how a downloaded model fits into the AI Act's transparency requirements, which formally kicked in earlier this year.

Bob: The bottom line is that the choice for European developers is widening. Twelve months ago, if you wanted a top-end AI model you basically picked between three American providers. Today you can add Chinese options, French Mistral, and a growing set of European open-weight projects. More competition tends to mean lower prices and better terms for customers, which is the part that will matter for most listeners.

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.

Louisiana physicists build a quantum material that works at room temperature, no giant fridge required.

Bob: Physicists at Louisiana State University have created the first quantum material that can sort and transport different states of light while sitting on a normal desk at room temperature. Their paper is in Nature this week. The device is an ultrathin structure they're calling a metacrystal, and it acts like a filter that steers quantum light along separate paths while preserving the delicate information encoded inside it.

Samantha: The reason this matters is the practical bottleneck of quantum tech so far. Most quantum devices need to be cooled to near absolute zero, which means bulky, expensive refrigeration and specialist facilities. That's why quantum computers still live in labs, not offices. If you can get equivalent behaviour at room temperature, the whole engineering picture changes. Suddenly you can imagine quantum components inside ordinary equipment, from data centres to secure communication kit.

Bob: And the timing is significant for Europe. The EU has poured billions into its Quantum Flagship programme, and companies like IQM in Finland and Pasqal in France are building actual machines. Most of them still depend on cryogenic systems. A breakthrough like this doesn't replace their approach, but it opens a parallel route, where certain quantum tasks can be done with photonic chips at normal temperatures. That could be especially useful for quantum communication, where you're sending information along fibre-optic lines that already run at room temperature.

Samantha: It's worth being honest about the timeline, though. A lab result in Nature is a long way from a shipping product. What LSU has shown is that the physics works. Turning it into manufacturable components takes years, and this is where Europe has historically stumbled: strong basic research, weaker industrialisation. The question for Brussels and Berlin is whether they can move faster from paper to product this time.

Google's fifteen-billion-dollar Indian AI data centre runs into environmental protests over water and wildlife.

Bob: Google's plan to build a fifteen-billion-dollar data centre campus in India is hitting serious resistance from environmental groups. The project has government backing, but activists say it was fast-tracked without proper impact assessments, and they're pointing at two specific concerns: enormous water consumption for cooling, and disruption to local wildlife habitats. This is becoming a test case for how developing countries handle the AI infrastructure gold rush.

Samantha: The numbers behind modern AI data centres are genuinely staggering. A single large facility can drink millions of litres of water a day for evaporative cooling, and consume as much electricity as a small city. Multiply that by the dozens of sites being planned worldwide, and you start to understand why environmental groups are pushing back. India has particular reasons to worry, because large parts of the country already face serious water stress in summer.

Bob: For Europe, this is a familiar story arriving in a new place. Ireland effectively paused new data centre approvals near Dublin because grid operators warned the centres could consume nearly a third of national electricity. The Netherlands has similar debates around Amsterdam and Zeeland. What's happening in India suggests the same tension is now global: everyone wants the jobs and investment, nobody wants their aquifers drained.

Samantha: There is a technical answer emerging, which is liquid cooling and closed-loop water systems that reuse the same water instead of evaporating it. Some newer European projects, particularly in the Nordics, use seawater or even industrial waste heat recycling to warm nearby buildings. But retrofitting old designs is expensive, and hyperscale operators still tend to pick the cheapest option unless regulators force otherwise.

Bob: The bigger question is whether governments start treating water and electricity budgets for data centres the way they treat carbon budgets. Brussels is already discussing minimum energy-efficiency standards for large data centres under the Energy Efficiency Directive. Expect water to become the next front in that conversation, and India's protests are a preview of what happens when it doesn't.

The free knowledge tool Logseq turns your notes into a real thinking network on your own device.

Samantha: To close out, two things you can actually use today, starting with Logseq. It's a free, open-source knowledge tool that runs on Windows, Mac, Linux, iPhone and Android. On the surface it looks like a notes app, but it works more like a personal wiki: every note is a block you can link to any other block, and the app quietly builds a network of connections in the background.

Bob: The reason people reach for something like this rather than a plain document is that our thinking rarely happens in tidy files. You jot something down in a meeting, and three weeks later you need to connect it to a decision you made last month. Logseq lets you tag a topic once and then find every mention across your entire history, without any cloud service seeing your data. Users often describe it as an outliner that grew a memory.

Samantha: There are two features worth flagging. There's a daily journal that automatically creates a page for today, so you always have a place to start typing without deciding on a filename. And there's a graph view that draws the connections between your notes visually, which turns out to be surprisingly useful when you want to see what you've been thinking about across weeks or months.

Bob: For European listeners the appeal is straightforward. Your notes stay in a folder on your own device by default, and the file format is plain markdown, which means if you ever stop using Logseq you can still read your notes in any text editor. That's the opposite of the trap where your years of thinking are locked inside a company's cloud. It's free, and it takes about twenty minutes to set up something you'll actually keep using.

Samantha: One tip: don't try to import all your old notes at once. Users who stick with tools like this tend to start with today's journal entry and let the network grow naturally. Trying to migrate a decade of documents upfront is what makes people quit after a weekend.

HERE WeGo is the free European navigation app with genuinely useful offline maps for all of Europe.

Bob: The second thing to try today is HERE WeGo, a free navigation app from the European mapping company HERE Technologies. It's available on iPhone and Android, and it does something Google Maps still handles awkwardly: it lets you download complete country maps for offline use, then routes you through them without any data connection.

Samantha: HERE has an unusual backstory. The company is based in the Netherlands and its owners include several major European carmakers, which is why its map data ends up inside a lot of built-in car navigation systems. For an everyday phone user, that translates into map coverage that's often more up to date on European roads and small towns than the American alternatives.

Bob: The practical use case is holidays and business trips. You download the country you're travelling to over WiFi before you leave, and once you're abroad you can navigate by car, bike, on foot or by public transport without burning through mobile data or worrying about coverage in mountain areas. It also handles turn-by-turn voice guidance offline, which many free apps quietly don't.

Samantha: It's not perfect. The interface takes a bit of getting used to if you're coming from Google Maps, and the live traffic layer is thinner in some regions. But as a genuinely European alternative that keeps working when your phone signal drops in a Swiss valley or an Italian tunnel, it's hard to beat, and the price is zero.

Samantha: Today we covered: an AI-flown F-16 fighter jet, Alibaba's open-weight Qwen 3.8-Max model, a room-temperature quantum breakthrough from Louisiana State University, Google's fifteen-billion-dollar Indian data centre hitting environmental resistance, and two things to try yourself: the knowledge tool Logseq and the offline navigation app HERE WeGo.

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.