India chip plan 2026: Modi's Red Fort AI & fab push

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

Bob: And I'm Bob Russell. Today: India bets big on chips and AI skills from the Red Fort, Alibaba's Qwen models cross three billion downloads, a disturbing lawsuit over Grok and child abuse imagery, a major reshuffle at Google DeepMind, Bending Spoons snaps up Airtable at a deep discount, and to close, two things you can actually use today: the private search engine Brave Search and the free open-source drawing app Excalidraw. Let's start in Delhi.

India unveils a sweeping semiconductor and AI skilling plan aimed at rewiring its entire economy.

Samantha: From the Red Fort on Independence Day, Prime Minister Narendra Modi laid out one of the most ambitious tech plans India has ever announced. Three new chip plants are already in production, and the government wants five to eight more fabrication facilities over the next seven to eight years. On top of that, Modi promised AI training for ten million young Indians and a one lakh crore rupee innovation fund for startups. That's roughly one lakh crore rupees aimed at moving India from importing silicon and software to designing and building its own.

Bob: And the framing matters here. This isn't a single subsidy or a one-off announcement, it's being pitched as a decade-long industrial strategy. India has watched Taiwan, South Korea, and now the United States build entire ecosystems around chips, and Delhi clearly does not want to be left dependent on any of them. The AI skilling piece is arguably the more interesting half. Ten million trained young people is a workforce bigger than most European countries have in total. If even a fraction of that translates into working engineers, India becomes a serious talent pool for global AI companies.

Samantha: For European companies this cuts two ways. On one hand, more Indian fabs and trained engineers mean more partners, more suppliers, and potentially cheaper access to talent for European chip designers and AI firms. On the other hand, it's another region competing hard for the same equipment, the same rare materials, and the same global demand. European semiconductor plans, like the Chips Act, suddenly look modest next to what India is putting on the table.

Bob: The bottom line is that the map of chip manufacturing is being redrawn faster than Brussels tends to move. Whether India delivers on the timelines is another question, building fabs takes years and the yield problems are brutal, but the political will is now unambiguous. European policymakers should probably read the Red Fort speech carefully.

Alibaba's Qwen open models pass three billion downloads and claim the global open-source AI lead.

Samantha: Alibaba announced that its Qwen family of open-source AI models has crossed three billion downloads, and the company is claiming the top spot in the open-source AI world. That reportedly puts Qwen ahead of Meta's Llama models and Google's open offerings, with most of the growth coming in just the last six months. The pitch is straightforward: strong capabilities, permissive licensing, and it costs nothing to run.

Bob: The speed of this shift is what stands out. A year ago the conversation about open models was almost entirely about Meta. Now a Chinese lab is not only competitive but claiming the lead by download numbers. And permissive licensing matters more than it sounds. It means European developers, universities, and small companies can build products on top of Qwen without paying an American cloud provider per token, and without the legal grey zone that some earlier open models had.

Samantha: There's a geopolitical layer too. Washington has been pressing allies to pick a side between US and Chinese AI, and yet Qwen is being downloaded en masse by developers everywhere, including in Europe. That tension is not going away. For European businesses the practical question is straightforward: do you build on a model that runs on your own servers and cuts your cloud bill, or do you stick with closed systems from American labs for compliance and support reasons?

Bob: And the EU AI Act adds another wrinkle. Open models can be inspected, audited, and adapted, which fits the regulatory direction. But they also raise questions about who is responsible when something goes wrong. Qwen's rise makes that debate more urgent, because it is no longer theoretical for European CIOs.

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.

A Wyoming lawsuit alleges the Grok chatbot was used to create child sexual abuse imagery.

Bob: A federal lawsuit filed by a woman in Wyoming alleges that her stepfather used the Grok chatbot to turn a childhood photograph of her into child sexual abuse material. It's a deeply disturbing case, and it's now in the US federal court system. The suit raises direct questions about what safeguards were, or were not, in place on the image generation side of the product.

Samantha: This is exactly the scenario safety researchers have been warning about for two years. Once image-generation models are widely available and loosely moderated, the risk of misuse against real, identifiable people, especially children, is not hypothetical. The lawsuit will test whether an AI company can be held liable when its product is used to create illegal imagery of a real person. That's uncharted legal territory in the United States.

Bob: In Europe the legal picture is different and, in some ways, further along. The EU AI Act designates certain uses as prohibited or high-risk, and existing child protection laws already criminalise AI-generated abuse material in several member states. The UK passed specific legislation on AI-generated child sexual abuse imagery earlier this year. So while the Wyoming case plays out in America, European regulators are likely to point to it as evidence that voluntary safeguards from AI companies are not enough.

Samantha: The uncomfortable reality is that the tools to prevent this kind of misuse exist, watermarking, image hashing, refusal training, identity checks on uploaded photos, but they cost money and slow products down. This case is going to force a very public conversation about which of those safeguards should be mandatory rather than optional.

Google DeepMind restructures, shifting from research-first to shipping AI products faster.

Bob: Google DeepMind has been through a major reorganisation, and the direction is clear: less pure research, more product delivery. According to reports, co-founder Demis Hassabis has stepped aside as CEO, chief scientist Jeff Dean has left to start his own company, and co-founder Sergey Brin is taking a bigger role. Control is being consolidated more tightly with Google's Silicon Valley parent, and the mission is to get AI into commercial products faster.

Samantha: This is a significant cultural shift. DeepMind was founded in London in 2010 with a research-first identity, and it produced landmark work like AlphaFold, which won a Nobel Prize in chemistry last year. Moving to a product-delivery focus makes commercial sense given how quickly OpenAI and Anthropic are shipping, but it changes what DeepMind is. The departure of Jeff Dean to start his own venture is notable, he was one of the most influential engineers at Google for over two decades.

Bob: For Europe there's a specific angle here. DeepMind's London office has been one of the flagship AI research operations on the continent, employing hundreds of researchers and drawing talent from European universities. A shift toward product delivery, run more tightly from California, raises questions about whether that London hub keeps its research weight or becomes more of an engineering satellite.

Samantha: And competitively, if Google speeds up product releases, European AI companies like Mistral and Aleph Alpha are up against an even faster-moving rival. The window for building distinctly European alternatives keeps narrowing. The bottom line is that the era of AI labs behaving like research institutes is ending, even at the most prestigious of them.

Bending Spoons buys Airtable for one point twenty-nine billion dollars, an eighty-one percent haircut from its peak.

Bob: Italian tech company Bending Spoons has acquired the workflow platform Airtable at an enterprise value of one point twenty-nine billion dollars. That's an eighty-one percent discount from Airtable's peak valuation of eleven point seven billion, and it values the company at 2.7 times its annual recurring revenue. For anyone who remembers the funding frenzy of a few years ago, this is a very sober number.

Samantha: Airtable was one of the darlings of the last software boom, a flexible spreadsheet-database tool used by teams to build lightweight internal apps. The peak valuation reflected an assumption that every knowledge worker would end up using something like it. That assumption did not fully play out, and the company had to raise capital in a much colder market. Bending Spoons is getting a well-known product at a fraction of its former price.

Bob: Bending Spoons has become a serious European player through this exact playbook. They previously acquired Evernote, Meetup, and WeTransfer, take mature software products, cut costs hard, and run them profitably. It's not glamorous but it works, and it's a rare example of a European company consolidating well-known Western consumer and business software.

Samantha: The wider signal is about valuations. Companies that raised at peak multiples in 2021 are now being sold or acquired at much lower numbers, and that recalibration is still working through European venture portfolios. Late-stage European startups looking for exits should not expect the pricing of three years ago. For users of Airtable the immediate question is what changes under new ownership, Bending Spoons has a reputation for tightening subscription pricing and trimming free tiers.

Try Brave Search, the independent European-friendly search engine that runs its own index.

Bob: To close out, two things you can actually use today. First up, Brave Search. It's the search engine built by the team behind the Brave browser, and the appealing part is that it runs on its own independent index. It is not quietly rebranding Google or Bing results underneath. That makes it one of the very few genuinely independent search engines available to European users right now.

Samantha: The pitch is straightforward. No ad tracking, no personalised profile built from your searches, and results that are not filtered through the same commercial optimisation as the big two. Users report that for technical searches and news queries the results hold up well, and for shopping queries it's improved noticeably over the past year. It also has an optional AI summariser called Answer with AI that gives you a short overview at the top of the results, which you can turn off if you prefer plain links.

Bob: Practically, you can start using it in about thirty seconds. Go to search dot brave dot com in any browser, or if you use the Brave browser it's already set as default. Setting it as the default search in Firefox, Safari, or Chrome takes about two clicks in settings. It's free, and there is no account required. For anyone who has been vaguely meaning to try an alternative to Google but never quite got round to it, this is the low-effort version.

Samantha: A useful tip: Brave Search has what it calls Goggles, which are custom ranking filters you can apply. There is one for academic sources, one for tech blogs, one for independent journalism. It's a genuinely different way to search, and it takes a minute to try. Available worldwide, works in every major European language, and, again, completely free.

Excalidraw turns any browser into a shared whiteboard for sketches and diagrams.

Bob: The second thing to try today is Excalidraw. It's a free, open-source drawing tool that runs entirely in your browser, and it makes rough sketches, flowcharts, and diagrams look like they were drawn by hand. That hand-drawn style is the point, it signals a work-in-progress rather than a polished slide, which is exactly what you want for early brainstorming.

Samantha: The reason people keep recommending it is speed. You open the site, you start drawing, no signup, no account, no template gallery to wade through. It has just enough shapes and arrows to build real diagrams, and you can collaborate live with colleagues by sharing a link. For remote teams in Europe running meetings across time zones, it's become a genuinely useful whiteboard replacement.

Bob: What to try today: go to excalidraw dot com in any browser. Draw a quick flowchart, save it as a PNG or a link, and share it in your next meeting. The whole thing takes about two minutes to learn. It's free, works on any device with a browser, including tablets with a stylus, and there's an open-source version you can host yourself if your company is strict about that.

Bob: Today we covered: India's Red Fort chip and AI plan, Alibaba Qwen crossing three billion downloads, the Grok child abuse imagery lawsuit, Google DeepMind's product-first restructuring, Bending Spoons buying Airtable at a steep discount, and two things to try yourself: Brave Search and Excalidraw.

Samantha: Today we covered: India's Red Fort chip and AI plan, Alibaba Qwen crossing three billion downloads, the Grok child abuse imagery lawsuit, Google DeepMind's product-first restructuring, Bending Spoons buying Airtable at a steep discount, and two things to try yourself: Brave Search and Excalidraw.

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.