← The State of Tech RSS Feed

21 juli 2026 · 12:49

China AI infrastructure drives Q2 growth: Gartner at $64B

China's AI infrastructure boom is no longer a footnote: electronics and information technology accounted for more than half of the country's Q2 economic expansion, with AI-related exports tripling their share of GDP growth year on year. In today's episode, Samantha and Bob unpack what that concentration means for Europe, pair it with Gartner's fresh $64 billion forecast for the AI platform market, and cover Moonshot AI's latest model, Lightcraft's Spark Story virtual production beta, and the NexaHealth Triage app rolling out across 20 countries.

AI China Gartner Moonshot AI Virtual Production Lightcraft

Beluister deze aflevering:

Transcript

Samantha: Welcome to The State of Tech, The European Edition, Tuesday July twenty-one, 2026. I'm Samantha Lawrence.

Bob: And I'm Bob Russell. Today: China's economy leaning hard on AI infrastructure, Gartner sees the AI platform market ballooning past sixty billion dollars, Moonshot's new model shakes up the global AI race, Lightcraft opens up a virtual production tool for filmmakers, and to close out, two things you can actually use today: the Spark Story beta from Lightcraft, and the healthcare screening app NexaHealth Triage. Let's start with China.

China's economy leans on an AI infrastructure boom as broader growth wobbles.

Samantha: The big picture out of Beijing this morning: artificial intelligence has quietly become one of the main things holding China's economy up. In the April to June quarter, electronics and information technology contributed more than half of China's economic expansion. That's not a rounding error, that's the engine.

Bob: And the export side tells the same story. AI-related exports alone accounted for one point one percentage points of nominal GDP growth in the first four months of the year. That's nearly triple the share they had a year earlier. So chips, servers, networking gear, all the physical stuff underneath the AI boom, is flowing out of Chinese factories at a pace that is starting to matter for the headline growth number. Analysts at Capital Economics say this could give China some economic resilience through 2027.

Samantha: The caveat they add is important though. Capital Economics warns that AI is not a fix-all for the wider Chinese economy. Property is still weak, consumer demand is patchy, and if the global AI investment cycle cools, that pillar wobbles too. So it's a real growth driver, but a concentrated one, and concentrated growth drivers tend to age badly when the cycle turns.

Bob: For Europe, there are two lenses on this. First, the strategic one: China is showing what happens when you treat AI infrastructure as an industrial policy, not just a software story. Europe is spending on the AI Gigafactories and on chip capacity, but the sums are still an order of magnitude smaller than what China and the US are putting in. Industry groups in Brussels argue that gap is widening, not closing. Second, the trade lens. If Chinese AI hardware exports keep climbing, European buyers get more choice and lower prices in the short term, but it also feeds the debate about dependency on Chinese kit inside critical infrastructure. Expect that conversation to get louder in the autumn, when the Commission revisits its economic security toolbox.


Gartner forecasts the worldwide AI platforms market to hit sixty-four billion dollars in 2026.

Samantha: Staying with the money side. Gartner is out with a new forecast that puts worldwide end-user spending on AI models and platforms at sixty-four billion dollars in 2026. That's up sixty-three percent from thirty-nine billion in 2025. Generative AI models specifically are set to grow one hundred and seventeen percent. So the market is not just big, it's still doubling in places.

Bob: But the tone underneath the numbers has shifted. Gartner explicitly says enterprise AI budgets are now under closer scrutiny. Companies are asking harder questions about efficiency, cost control, and measurable outcomes. That's a real change from the last two years, when a lot of buyers were essentially spending to be seen spending. Now finance teams want to see the ROI on the invoice.

Samantha: And the interesting number sitting inside that forecast is the two hundred and ten percent projected growth for domain-specific and specialised models. In plain terms: smaller AI systems built for one job, like legal contracts, radiology scans, or industrial inspection, are growing more than three times faster than the generic chatbots. That's a signal that the market is maturing from "buy the biggest model" to "buy the model that actually fits the problem."

Bob: For European companies, this matters in two ways. One, there's a real opening for smaller European AI vendors who build specialised tools, because the buyer's mood has moved in their direction. Two, CFOs at big European enterprises now have Gartner-branded cover to push back on generic AI spend and ask for evidence. The era of the blank cheque for AI pilots looks to be ending, and that's probably a healthy thing for the market.

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.


China's Moonshot AI unveils a new large language model and reignites the global AI race.

Bob: Third story, and it connects directly to what we just covered. Chinese AI developer Moonshot has unveiled its latest large language model, and the reaction on markets has been sharp. Investors are re-opening the debate about AI infrastructure spending, semiconductor demand, and whether the top-tier models from the US labs still command the premium they used to.

Samantha: The pattern here is becoming familiar. Every few months, a Chinese lab, DeepSeek earlier, now Moonshot, releases a model that performs close to the top US systems for a fraction of the training cost, and Western AI stocks wobble for a day or two. What's different this time is that it's happening while Gartner is publishing that sixty-three percent growth forecast we just discussed. So the market has to hold two ideas at once: the pie is getting much bigger, but the slices may be redistributed.

Bob: That redistribution question is the awkward one for investors. If Chinese models keep closing the quality gap and undercutting on price, then the valuations built around a handful of US labs and their chip suppliers get harder to defend. Australian investors in particular were flagged as watching this closely, because their tech exposure runs heavily through those same names.

Samantha: For Europe, the takeaway is twofold. European buyers get more options and more pricing pressure, which is good news if you're procuring AI. But European AI companies, Mistral, Aleph Alpha, the smaller labs, are now squeezed between well-funded American giants on one side and fast-moving, cheaper Chinese entrants on the other. The regulatory piece matters too: the AI Act's transparency and safety requirements apply regardless of where the model was trained, so any Chinese model deployed in the EU has to meet the same bar. That's a lever European regulators do have.


Lightcraft opens up Spark Story, a virtual production tool that lets filmmakers pre-visualise entire scenes in 3D.

Bob: Now for something completely different, and genuinely non-AI. Lightcraft Technology has opened a rolling beta for Spark Story, a virtual production tool aimed at film pre-production. It was demonstrated at SIGGRAPH 2026, the big computer graphics conference, and the idea is to let filmmakers workshop stories and visualise scenes in full 3D from day one, using realistic digital environments and physically accurate motion.

Samantha: What that means in practice: instead of storyboards on paper or expensive location scouts, a director, a cinematographer, and a producer sitting in different cities can jump into the same virtual set, move a camera around, try out shots, and agree on a look before anyone books a crew. The new features include a 3D asset and environment library, one-click real-time compositing on iPad, and the ability to load USD files directly. USD is the 3D file format that most of the big studios have standardised on, so Spark Story is fitting into an existing pipeline rather than asking studios to rebuild theirs.

Bob: The business case is straightforward. Pre-production mistakes are cheap to fix in software and expensive to fix on set. If you can catch a bad camera angle or a scene that doesn't work in a virtual review, you save real money. That's why virtual production has moved from a niche technique used on shows like The Mandalorian to something that mainstream film and streaming productions are adopting.

Samantha: For Europe, this is genuinely relevant. The continent has serious film and TV production hubs in London, Paris, Berlin, Prague, and the Nordics, and virtual production stages are being built out in several of them. A tool like Spark Story lowers the cost of entry for smaller studios and independent producers who cannot afford a full-scale LED volume. It's also a reminder that not every tech story is an AI story. This is 3D graphics, real-time rendering, and collaboration software doing the useful work.


To close out, two things you can actually use today: the Spark Story beta, and the NexaHealth AI triage app.

Bob: To close out, two things you can actually try today. The first one follows naturally from that last story: the Spark Story beta from Lightcraft. It's not just for Hollywood. If you're an indie filmmaker, a student on a film course, or you run a small production company doing corporate video or short-form work in Europe, this is worth looking at.

Samantha: The beta is described as rolling, which means Lightcraft is letting people in gradually rather than opening the floodgates. You sign up through the Lightcraft website. Once you're in, you get access to that 3D asset and environment library, so you don't need to build your own sets from scratch. Reviewers at SIGGRAPH highlighted the one-click compositing on iPad as the most immediately practical feature: you can point an iPad at your desk, and see a scene composited in real time with the digital environment behind it. That turns any room into a rough previs stage.

Bob: The practical thing to know is that Spark Story is aimed at the collaborative side of pre-production. So it's most useful if you actually have a team, even a small one, scattered across different cities. Users report that where it saves the most time is in those first script-to-screen conversations, when a director and a director of photography need to agree on what a scene is going to look like.

Samantha: One honest caveat: rolling beta means bugs, and it means the feature set will change under you. So this isn't the tool to bet a paid client project on next week. It's the tool to spend a Saturday afternoon experimenting with, to see whether it earns a place in your workflow. If you shoot on iPhone or iPad already, that's the easiest on-ramp, because the real-time compositing feature runs there natively.


NexaHealth's AI triage app is rolling out across twenty countries as a first-line health check.

Bob: Second one, and it's a very different kind of tool. NexaHealth Triage. It's an AI-powered symptom checker and triage app that's being rolled out to public health systems in more than twenty countries. NexaHealth was named as one of Techtalk Magazine's Companies to Watch this month, and the triage app is the piece of what they do that ordinary people can actually download and try.

Samantha: The way it works, based on what users describe, is that you enter your symptoms in plain language, answer a few follow-up questions, and the app suggests whether you should self-care, book a regular GP appointment, or seek urgent care. It's not a diagnosis, and NexaHealth is careful to make that clear. It's a routing tool, designed to help you decide what kind of care you probably need before you spend three hours in an A&E waiting room.

Bob: The reason to mention it as a practical find rather than just company news is that it's rolling into public health systems, which means in a growing list of countries it's essentially free at the point of use through your national health service's own app store or portal. Reviewers point out that its strength is on the routing side, not on rare conditions. So use it the way it's designed: to answer the question "how urgent is this, actually?" That's the moment most people either panic or under-react, and a calm second opinion is genuinely useful.

Samantha: The obvious European note: health data is sensitive, and GDPR applies. If you use it through your national health service's rollout, the data handling is covered by that agreement. If you download it directly, read the privacy notice before you type in your symptoms. And of course, if something feels seriously wrong, call your national emergency number first, an app is never a substitute for that.


Samantha: Today we covered: AI infrastructure driving China's economic growth, Gartner's forecast of a sixty-four-billion-dollar AI platform market, Moonshot AI reshuffling the global AI race, Lightcraft's Spark Story opening up virtual film production, and two things to try yourself: the Spark Story beta and the NexaHealth Triage app.

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.