20 september 2026 · 14:43
Anthropic vs OpenAI: New Model Amid IPO Pressure
Anthropic is reportedly weighing the launch of a new AI model to counter OpenAI's GPT-6 Astra, putting its safety-first positioning under direct commercial pressure ahead of a potential IPO. This episode unpacks what that tension means for European enterprise buyers, the AI Act, and Anthropic's credibility with investors. We also cover drugmakers co-writing regulatory frameworks for pharmaceutical manufacturing in orbit, Spirit AI's 2027 humanoid brain forecast, and two open-source tools worth installing today.
Beluister deze aflevering:
Transcript
Samantha: Welcome to The State of Tech, The European Edition, Sunday September twenty, 2026. I'm Samantha Lawrence.
Bob: And I'm Bob Russell. Today: Anthropic weighs a new model to counter OpenAI, Donald Trump wants to rename artificial intelligence, a Chinese startup predicts a humanoid robot brain breakthrough, drugmakers push pharmaceutical research into orbit, and to close out, two things you can actually use today: the free science reader Kiwix Medicine Library, and the open-source home tool OpenSprinkler. Let's start with Anthropic.
Anthropic weighs launching a new AI model to counter OpenAI's GPT-6 Astra momentum.
Samantha: Anthropic is reportedly weighing the release of a new AI model to answer OpenAI's recently launched GPT-6 Astra. The internal deliberation comes amid competitive pressure and, according to the reporting, ahead of a potential initial public offering. What makes it awkward is the timing. Anthropic CEO Dario Amodei publicly called for a global slowdown in AI development on safety grounds only days ago, and now his own company is said to be preparing a counter-punch.
Bob: The trigger, according to sources cited in the reporting, is the strong market traction of GPT-6 Astra. Some investors have started re-evaluating whether Anthropic still holds its position as a premier enterprise AI tool provider. The company is said to be evaluating safety protocols for its next model as part of the discussions, though no launch date has been confirmed. So this is very much internal debate, not an announcement.
Samantha: The tension here is obvious. You cannot credibly call for the industry to slow down and then rush a model out to defend market share. Either the safety concerns are real and the timeline gets adjusted, or the safety talk starts to look like positioning. Investors reading the same reporting will draw their own conclusions, and enterprise buyers who signed with Anthropic on the safety pitch will want to see how the two threads are reconciled.
Bob: For European customers, and there are many banks, telcos and public sector bodies using Claude under enterprise contracts, this matters practically. If Anthropic ships a stronger model soon, they get faster capability upgrades. If the company holds back, OpenAI's lead widens and procurement teams may reopen contracts. Either outcome shifts negotiation leverage. And European regulators watching the AI Act rollout will note the gap between public safety rhetoric and commercial reality.
Samantha: The bottom line: this is one of the sharpest tests yet of whether the safety-first branding survives contact with a serious commercial threat. The next few weeks of Anthropic communication will tell us a lot about which way the company actually leans when the pressure is on.
Donald Trump launches a public poll to rename artificial intelligence.
Bob: Former US President Donald Trump has publicly suggested renaming artificial intelligence, arguing the current term is inaccurate and, in his words, ineloquent. In a post on Truth Social he launched a poll offering alternatives: Superior Intelligence, Extreme Intelligence, or Supreme Intelligence. He described the current label as failing to capture the pace and nature of the technology.
Samantha: The proposal is largely symbolic, but the timing is interesting. It lands in the middle of a broader debate about the economic and societal impact of AI, and significant new investment in AI infrastructure and computing capacity. Renaming a technology does not change what it does, but it can shift how the public talks about it, and how policymakers frame their arguments. Words carry weight in regulation.
Bob: There is a substantive point buried in the stunt. Plenty of researchers have argued for years that artificial intelligence is a misleading label, since current systems are statistical pattern matchers, not intelligent in any human sense. Whether Superior or Supreme is an improvement is another question. Both suggest the technology is already above us, which is exactly the kind of framing safety researchers have warned against.
Samantha: For European audiences the immediate impact is limited, but not zero. Public perception drives political appetite for regulation, and a high-profile rebranding push from a former US president tends to ripple across the Atlantic. Expect the terminology debate to surface in European media in the coming days, and possibly in policy circles too, where the AI Act uses the current term throughout its text.
Bob: Practically, no one is renaming their product line tomorrow. But if the poll gets traction, expect marketing departments at the big US labs to test alternative framings quietly. Watch whether any of them start using different language in their next product launches. That would be the tell that the rebrand is more than a Truth Social post.
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.
Chinese startup Spirit AI predicts a humanoid robot brain breakthrough by mid-2027.
Bob: Spirit AI, a leading Chinese embodied AI firm, is forecasting a significant breakthrough in humanoid robot brains as early as mid-2027. Chinese humanoids have grabbed headlines this year with hardware feats: sprinting, dancing, backflips. Spirit AI argues the impressive body is now waiting for a matching mind, and that the next crucial step is in software, not mechanics.
Samantha: The company is careful to separate the cognitive leap from actual deployment. Even if the brain breakthrough arrives on schedule, Spirit AI says widespread rollout of advanced humanoids in homes could still take at least eight years. The blocker is what they call a persistent data bottleneck: there simply is not enough real-world interaction data to train these systems to behave reliably in messy human environments.
Bob: That eight-year timeline is worth pausing on. It cuts against the marketing you see from some robotics firms suggesting your household helper is around the corner. Spirit AI is one of the more advanced players in this space, and their own message is: the smart part is coming, the useful part is not. Kitchens, staircases and pets are harder than dance routines.
Samantha: For Europe this prediction lands at an interesting moment. Several European industrial groups have started their own humanoid programmes, and research funding is flowing. If the Chinese cognitive breakthrough arrives in 2027, European firms have a narrow window to either catch up or focus on the areas Spirit AI itself flags as bottlenecks, particularly data collection in real environments and safety certification for shared spaces.
Bob: There is also a supply chain angle. The hardware for these robots leans heavily on Chinese components today. If the software leap happens inside a Chinese firm first, the entire stack tilts further east. European policymakers thinking about strategic autonomy in robotics will want to read the Spirit AI forecast carefully, and decide whether the response is investment, partnership, or standards.
Drugmakers pioneer commercial pharmaceutical research in orbit.
Samantha: As the International Space Station approaches the end of its operational life, pharmaceutical companies are preparing for the next phase of drug research and manufacturing in low Earth orbit. Drugmakers are working with regulators to establish guidelines for producing medicines in space, aiming to show reliable and cost-effective manufacturing outside Earth's gravity. This is a genuine shift from experimental science to commercial planning.
Bob: The scientific case is well established. Researchers have long recognised that microgravity offers unique advantages for cellular and particle distribution in biological materials that are difficult to replicate on Earth. Certain protein crystals grow more perfectly, certain tissue structures form more evenly. What is new is the commercial infrastructure being built around these advantages, with regulators drawn in from the start rather than after the fact.
Samantha: Two companies stand out in the reporting. LambdaVision is developing artificial retinal implants using space-based manufacturing, aimed at restoring sight lost to degenerative disease. Auxilium Biotechnologies is using 3D bioprinting in orbit to produce transplantable tissues. Both are betting that the quality gains from microgravity manufacturing justify the logistics of getting materials up and finished products back down.
Bob: The regulatory piece is the interesting part. Producing a medicine in orbit raises questions no framework fully answers today: jurisdiction, sterility standards, transport validation, batch traceability across an atmospheric re-entry. The fact that drugmakers are actively co-writing guidelines with regulators suggests the industry sees a real path to approved products, not just experiments.
Samantha: For Europe this opens two doors. European pharmaceutical companies have deep expertise in biologics and cell therapies, exactly the areas where microgravity offers the biggest gains. And the European Medicines Agency will need to decide whether to build its own space-manufacturing rulebook or align with what US regulators develop. Harmonisation now would save years of duplicated work later, and give European firms a fair shot at the market.
Bob: The commercial timeline is still uncertain, but the direction is clear. Space is moving from research platform to production floor for high-value, low-volume medicines. Retinal implants and transplantable tissues fit that profile perfectly. Watch for the first approved product coming out of orbit within this decade.
The free science reader Kiwix Medicine Library puts trusted medical references on your laptop offline.
Samantha: To close out, two things you can actually use today. The first is aimed at anyone who wants reliable reference material without depending on a search engine. Kiwix Medicine Library is a free download that packages curated medical and scientific references into a single reader you keep on your own device. It works entirely offline once installed, which matters when you are travelling, on a weak connection, or simply want the material available regardless of what any website decides to change.
Bob: Kiwix itself is an open-source project that has been packaging offline versions of large reference collections for years. The medicine bundle draws from openly licensed medical wikis and reference works. Users report that having a serious reference library sitting on the laptop, searchable in seconds, changes how you approach any question that involves a technical term. It is not a replacement for a doctor, but it is a solid starting point for understanding what a term means.
Samantha: Installation is straightforward. You download the Kiwix reader for Windows, Mac, Linux, Android or iOS, then pick the medicine bundle from the built-in library catalogue. The full medical collection runs to several gigabytes, so plan the download over a decent connection. After that, everything works offline, no account required, no tracking, no subscription.
Bob: The wider Kiwix catalogue is worth exploring while you are there. There are offline packages for Wikipedia in dozens of languages, technical manuals, cooking references and educational content aimed at schools without reliable internet. For anyone travelling in Europe this autumn, or heading somewhere with patchy coverage, dropping a few of these onto a laptop or phone before you leave is a small habit that repays itself repeatedly.
Samantha: One practical note: the medical content is a reference, not personalised advice. Users describe it as most useful for understanding terminology in a diagnosis letter, or for reading around a subject before a consultation, rather than for self-diagnosis. Treat it as a well-organised library, not a doctor in your pocket.
OpenSprinkler turns your garden watering into a private, weather-aware system you fully control.
Bob: The second thing to try today is for anyone with a garden, an allotment, or even just a few pots on a balcony. OpenSprinkler is an open-source irrigation controller that turns garden watering into a smart, weather-aware system without handing your data to a big cloud service. It runs on affordable hardware you can order online, and the software is free.
Samantha: The idea is simple. Instead of a fixed timer that waters whether it rained overnight or not, OpenSprinkler pulls weather data and adjusts the schedule automatically. If a storm is forecast, it skips the cycle. If it has been dry and hot, it extends. Users report water savings of thirty to fifty percent compared with a dumb timer, which matters both for the bill and for the increasingly frequent European summer droughts.
Bob: The hardware sits between your existing garden tap valves and the controller software, which you access through a web browser on any device. There is no mandatory app store, no subscription, no account with a company that might change its terms next year. The community around the project publishes plug-ins, weather integrations and detailed setup guides in several European languages.
Samantha: For allotment holders and small growers the appeal is control. You can schedule different zones for different plants, override remotely when you are away, and see exactly how much water each zone has received. Reviewers particularly praise the reliability: the system keeps running even if your home internet drops, because the schedule lives on the controller itself.
Bob: Today we covered: Anthropic weighing a new model to counter OpenAI, Donald Trump's push to rename artificial intelligence, Spirit AI's forecast of a humanoid robot brain breakthrough, drugmakers preparing to manufacture medicines in orbit, and two things to try yourself: Kiwix Medicine Library and OpenSprinkler.
Samantha: Today we covered: Anthropic weighing a new model to counter OpenAI, Donald Trump's push to rename artificial intelligence, Spirit AI's forecast of a humanoid robot brain breakthrough by mid-2027, drugmakers preparing to manufacture medicines in orbit, and two things to try yourself: Kiwix Medicine Library and OpenSprinkler.
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.