KPMG Claude Rollout: OpenAI Enters Consulting

Samantha: Welcome to The State of Tech, The European Edition, Thursday May twenty-eight, 2026. I'm Samantha Lawrence.

Bob: And I'm Bob Russell. Today: KPMG rolls out Claude to its entire global workforce while OpenAI launches a consulting arm; Nvidia pours up to one hundred and fifty billion dollars a year into Taiwan; Amazon greenlights its first fully AI-generated children's shows; a robotic hand teaches itself music by ear; a new chip designed for AI in space; and a laser that keeps a warehouse robot running for twenty-four hours straight. Let's start with the biggest shake-up in consulting we have seen in years.

KPMG deploys Claude to 276,000 staff as OpenAI launches a four-billion-dollar consulting arm.

Samantha: Two announcements landed today that, taken together, signal a real turning point for how AI gets sold into big business. First, KPMG, one of the Big Four accounting and consulting firms, is rolling out Anthropic's Claude AI to its entire global workforce. That is over two hundred and seventy-six thousand professionals, across one hundred and thirty-eight countries. They are plugging it straight into their main client platform, called Digital Gateway, and aiming it at tax, legal, and private equity work.

Bob: And on the same day, OpenAI announced something called DeployCo. It is a brand-new consulting subsidiary, backed with four billion dollars from major investors. The idea is to embed OpenAI's own engineers inside client companies, helping them actually use AI in their day-to-day workflows, rather than just selling them access to a model.

Samantha: So in one corner you have a Big Four firm betting its entire delivery model on Claude. In the other, the maker of ChatGPT setting up shop as a consulting firm itself. That is a direct shot across the bow of the traditional consulting world.

Bob: It is. For years the big AI labs said: we make the models, the consultants integrate them. That line is now blurring fast. OpenAI clearly wants a piece of the implementation revenue, not just the licence fees. And KPMG is trying to get ahead by saying, look, our entire workforce already runs on frontier AI.

Samantha: This makes clear that the battleground is shifting from who builds the smartest model to who actually gets it working inside real companies. That is where the money lives.

Bob: For European listeners, this matters in two ways. One: if you work at a large bank, a law firm, an audit client, expect Claude-powered tools to start appearing in audits, tax filings, and due diligence reports within months. Two: it puts pressure on European consulting players and on Mistral, the French AI champion, because the heavyweight enterprise deals are being locked in by American labs right now.

Samantha: There are also questions Brussels will be watching. Confidentiality of client data, professional liability when AI co-authors an audit, and how Europe's AI Act applies when an OpenAI engineer is effectively sitting inside a European bank. Plenty for regulators to chew on.

Nvidia's Jensen Huang calls Taiwan the epicentre of AI and pledges up to 150 billion dollars a year.

Bob: Nvidia's CEO Jensen Huang was in Taiwan this week, and he did not hold back. He called the island the epicentre of the global AI revolution and announced annual investment that could reach between one hundred and one hundred and fifty billion dollars. That is on top of the roughly hundred billion Nvidia has already committed.

Samantha: To put that in perspective: Nvidia's annual spending in Taiwan alone could rival the entire defence budget of a mid-sized European country. Huang said Taiwan now sits at the centre of Nvidia's AI ecosystem and its chip manufacturing, and he warned that the island will need to massively expand its energy capacity to keep up.

Bob: That energy point is critical. Taiwan is already power-constrained, and AI chip production is enormously hungry. Huang basically told his hosts: if you want to stay the centre of this industry, build more power plants, fast.

Samantha: It also reinforces a geopolitical reality Europe cannot ignore. The advanced chips that train and run AI are overwhelmingly made on one island, in a tense corner of the world. Any disruption in the Taiwan Strait would ripple straight through European data centres, car factories, and defence contractors.

Bob: Brussels has been pushing the European Chips Act to attract more semiconductor capacity to the continent, but the gap is widening. While Europe debates billions, Nvidia and Taiwan are talking in hundreds of billions per year.

Samantha: For European tech firms the practical effect is simple: chip allocation, pricing, and delivery times for advanced AI hardware remain decided in Taipei, not in Brussels or Berlin. That dependency is not going away soon.

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.

Amazon greenlights its first three children's shows made entirely with generative AI.

Bob: Amazon MGM Studios has officially greenlit its first three children's shows developed through a new initiative called the GenAI Creators Fund. The fund backs filmmakers, digital creators, and startups using generative AI to produce content much faster than traditional studios can.

Samantha: And the numbers are striking. One director on the project said a pilot episode that would normally take two years to produce was finished in two months. That is a tenfold compression of the production cycle.

Bob: The target is clear: lower cost, faster output, more shows greenlit for a fraction of the traditional budget. Children's animation is a logical place to start, because the visual style is more forgiving than live action, and the volume of content needed is enormous.

Samantha: This is also the first time a Hollywood-scale studio has publicly committed to AI-generated children's programming as a strategy, not just an experiment. That makes it a real test case, and a contentious one. Hollywood unions have been warning for years that generative AI would displace animators, writers, and voice actors. This announcement does not exactly calm those fears.

Bob: For European audiences, two things stand out. First, expect to see these shows on Prime Video in Europe, with all the questions that brings around child-appropriate content moderation, especially under the EU's Digital Services Act. Regulators will want to know how the AI was trained and whether outputs are reviewed for harmful material before reaching children.

Samantha: Second, European public broadcasters, like the BBC, France Télévisions, or Germany's ARD, are watching closely. If Amazon can make kids' shows in two months instead of two years, that is going to put real pressure on traditional production budgets across the continent.

Bob: There is also a copyright question. European law on AI-generated works and on training data is still being shaped. Studios that depend on AI pipelines will need to prove their training sets are clean, and that is far from settled.

A robotic hand at USC teaches itself to play a melody by ear in just two minutes.

Samantha: Researchers at the University of Southern California have built a robotic hand that can learn a melody by ear, with no sheet music and no detailed programming. After about two minutes of self-taught practice, it can reproduce a tune. They call it the Musician Hand.

Bob: What is genuinely new here is the method. Instead of training on enormous datasets, the way most AI systems do, this robot uses what the researchers call motor babbling. It is the same way infants learn to control their limbs: try a movement, listen to the result, adjust. The robot perceives, experiments, and self-corrects.

Samantha: The musical part is the eye-catching demo, but the real prize is medical. The USC team believes this same framework could be used to build personalised assistive devices for people with conditions like Parkinson's disease, where symptoms and movements change over time. A robotic aid that adapts to the patient, rather than the other way around.

Bob: That is a meaningful shift. Most assistive robotics today rely on heavy pre-programming and large training datasets, which makes them expensive and slow to personalise. A system that learns on the fly, from the patient in front of it, could be far cheaper and more useful in practice.

Samantha: For European healthcare systems, which are under serious budget pressure and dealing with ageing populations, adaptive rehabilitation robots could become an important tool. Hospitals in Germany, the Netherlands, and Scandinavia have already been piloting robotic rehab. This kind of self-learning approach could make the next generation of devices far more accessible.

Bob: It is still university research, so we are years from a commercial product. But it points to a direction where AI in healthcare looks less like a giant cloud model and more like a small, embedded system that learns from one person.

Bob: To close out, two things you can actually keep an eye out for today, one a fresh consumer-level hardware story and one a real-world tech demo that might change how robots get powered.

New nanomagnetic AI chips designed for satellites, defence, and space-based systems start trading today.

Samantha: A company called Rocket One officially started trading on the Nasdaq today, under the ticker R-K-T-O. It used to be a biotech firm called Hoth Therapeutics, but it has fully repositioned around a next-generation AI chip architecture. The chips use what is called a nanomagnetic design, built specifically for ultra-low-power and radiation-tolerant computing.

Bob: In plain English: these are chips meant for places where normal silicon struggles. Satellites, defence platforms, deep-space probes. Up there, you have intense radiation, very limited power, and no chance of swapping a broken component. Standard AI chips burn too much energy and degrade too quickly. A chip designed from the ground up for that environment is a real niche, but a growing one.

Samantha: For listeners who follow space tech, this is worth bookmarking. The European Space Agency, Airbus Defence and Space, and companies like ICEYE in Finland are all pushing more AI processing onto satellites themselves, so data can be analysed in orbit instead of beamed back to ground stations.

Bob: If you want to look into it today, you can check Rocket One's filings and investor materials on the Nasdaq website, which is free to access. It will not change your afternoon directly, but if you are curious about where edge AI hardware is heading beyond the usual Nvidia headlines, this is a name to watch.

Samantha: And it underlines a broader point. AI hardware is fragmenting. There is no longer one chip for every job. There are chips for data centres, chips for phones, chips for cars, and now chips for orbit. Each market with its own players.

A laser keeps a moving warehouse robot powered for twenty-four hours straight, no battery swap needed.

Bob: A company called Aquila Earth, with roots in New Zealand and operations in Australia, has just demonstrated something genuinely impressive. They kept a moving warehouse robot powered for twenty-four hours straight using nothing but a laser beam. No cables, no battery swaps, no docking station.

Samantha: They are calling it a world first in wireless power transmission. The laser delivered a steady four kilowatt-hours of energy, which they say sets new records for both the total laser power transferred and the longest continuous duration to a moving platform.

Bob: The idea is what they call perpetual robots and drones. Instead of building bigger batteries, you beam power to the machine as it works. For warehouses, ports, and large industrial sites, that could remove one of the biggest constraints on automation: downtime for charging.

Samantha: There are obvious questions, of course. Laser safety in a crowded warehouse, line-of-sight requirements, efficiency losses over distance. None of that is solved yet. But the demo itself is a real proof point, not a press render.

Bob: For European logistics, this is the kind of technology to keep an eye on. Operators like Maersk, DHL, and the big Dutch and German port automation companies are constantly looking for ways to extend robot uptime. If laser charging matures, it could reshape warehouse design over the next decade.

Samantha: You cannot try this one at home, obviously. But you can look up Aquila Earth's demonstration video online today, and it is worth a couple of minutes if you have any interest in robotics or industrial automation. It is one of those moments where something that sounded like science fiction last year is suddenly running in a real warehouse.

Samantha: Today we covered: KPMG and OpenAI reshaping the consulting world, Nvidia's enormous bet on Taiwan, Amazon's first AI-made children's shows, a robotic hand that learns music by ear, new chips built for AI in space, and a laser that keeps a robot running for a full day.

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.