When GPUs sing together: Solving the AI data centre power spike – We Talk IoT #87

When GPUs sing together: Solving the AI data centre power spike – We Talk IoT #87

AI data centres have a problem that no one talks about enough. When GPUs synchronise and fire together, power demand spikes exponentially, in milliseconds. Legacy backup systems were never built for this. Oversizing the grid is expensive. Throttling GPU utilisation wastes the very infrastructure operators are paying for. Ramesh Narasimhan, EVP, CCO and CFO at Nyobolt, explains how the Cambridge deep-tech startup evolved from a university materials lab into a full solutions provider. And how its Dynamic Response System addresses the power spike problem at the rack level, in real time. Ramesh discusses Nyobolt's journey from a vacuum cleaner prototype to a partnership with Symbotic's warehouse robots, the case for ultra-fast charging over high energy density, and why the company's core lesson applies as much to startups as to battery chemistry: don't confuse your product with your purpose. #battery #fastcharging #AIDataCentre #robotics #IoT #WeTalkIoT   Summary of this week's episode: 01:08 From automotive to deep-tech: Ramesh's background 02:00 Nyobolt's origin: Cambridge materials lab to commercial company 03:03 What forced the shift from materials to solutions provider 05:05 The secret sauce: what Nyobolt's battery does differently 06:47 Charging a car in under five minutes, and why range obsession misses the point 09:42 The AI data centre power problem explained 12:25 The Dynamic Response System vs. conventional UPS 13:55 Off-grid in Rajasthan: what the MoU actually means 14:50 Symbotic and the warehouse robot use case 16:02 Humanoids, trade-offs, and scaling the supply chain 17:55 Where Nyobolt focuses and why 19:05 Looking ahead: changing how industries think about operations 20:05 The talent ecosystem around Cambridge   Show notes: Ramesh Narasimhan: https://www.linkedin.com/in/ramesh-narasimhan-201ab3a/ [https://www.linkedin.com/in/ramesh-narasimhan-201ab3a/] Nyobolt Dynamic Response System: https://nyobolt.com/ai-data-centre-power-solutions/ [https://nyobolt.com/ai-data-centre-power-solutions/]   Listen to the "We Talk IoT" Soundtrack on: Spotify: https://open.spotify.com/playlist/05MOV4OV2MH2in2txsAGtG?si=ad08112cb8d443f4 [https://open.spotify.com/playlist/05MOV4OV2MH2in2txsAGtG?si=ad08112cb8d443f4] YouTube: https://www.youtube.com/watch?v=D-NvQ6VJYtE&list=PLLqgVFfZhDRVYmpEqbgajzDvGL4kACRDp [https://www.youtube.com/watch?v=D-NvQ6VJYtE&list=PLLqgVFfZhDRVYmpEqbgajzDvGL4kACRDp]   About Avnet Silica: This podcast is brought to you by Avnet Silica — the Engineers of Evolution. Subscribe to our newsletters here: https://my.avnet.com/silica/resources/newsletter/ [https://my.avnet.com/silica/resources/newsletter/] You can connect with us on LinkedIn: https://www.linkedin.com/company/silica-an-avnet-company/ [https://www.linkedin.com/company/silica-an-avnet-company/]. Or find us at www.avnet-silica.com [http://www.avnet-silica.com].

From the cab to the cloud: How AGCO Fendt is making smart farming real – We Talk IoT #86

Smart farming has been a promise for decades. GPS guidance, variable-rate application, autonomous tractors, the technology has been on the horizon for so long that it is easy to miss how much of it has already arrived. In this episode, Andreas Gatti, manager digital products & FendtOne at AGCO Fendt, takes us through what precision agriculture actually delivers today,  and where the hard problems still remain. Andreas explains how the PTx Trimble joint venture, formed by AGCO and Trimble in 2024, is turning Fendt from a machine manufacturer into a full farm ecosystem provider. We discuss what it means to treat 25-centimetre sections of a field independently, why full autonomy is less about driving and more about work quality, and how smart machines are quietly eliminating hours of paperwork for farmers who never asked for a digital transformation. We also ask the harder questions: who owns the data, whether the business case holds without subsidies, and which adoption barrier the industry consistently underestimates. #smartfarming #precisionagriculture #agtech #IoT #WeTalkIoT   Summary of this week’s episode: 01:30 Andreas introduces himself and AGCO Fendt 03:45 What PTx Trimble changes for farmers 07:10 How Fendt defines smart farming — and what is already delivering results 11:20 Mixed fleets and interoperability in practice 14:30 The biggest benefit: driver, owner, or office? 16:00 Where farm autonomy stands today — and what is still in the way 22:10 Treating every 25 centimetres differently: precision application 27:00 Data ownership, security, and GDPR 31:30 Smart farming and climate change: mitigation or marketing? 35:00 Adoption barriers: cost, skills, connectivity — and the one nobody talks about 40:20 What Andreas is watching on the horizon   Show notes: Andreas Gatti: https://www.linkedin.com/in/andreas-gatti-b09195104/ [https://www.linkedin.com/in/andreas-gatti-b09195104/] Smart farming: https://www.fendt.com/de/smart-farming [https://www.fendt.com/de/smart-farming] Smart agriculture: https://my.avnet.com/silica/solutions/markets/smart-city/smart-agriculture/ [https://my.avnet.com/silica/solutions/markets/smart-city/smart-agriculture/]   Listen to the “We Talk IoT” Soundtrack on: Spotify: https://open.spotify.com/playlist/05MOV4OV2MH2in2txsAGtG?si=ad08112cb8d443f4 [https://open.spotify.com/playlist/05MOV4OV2MH2in2txsAGtG?si=ad08112cb8d443f4] YouTube: https://www.youtube.com/watch?v=D-NvQ6VJYtE&list=PLLqgVFfZhDRVYmpEqbgajzDvGL4kACRDp [https://www.youtube.com/watch?v=D-NvQ6VJYtE&list=PLLqgVFfZhDRVYmpEqbgajzDvGL4kACRDp] About Avnet Silica: This podcast is brought to you by Avnet Silica - the Engineers of Evolution. Subscribe to our newsletters here: https://my.avnet.com/silica/resources/newsletter/ [https://my.avnet.com/silica/resources/newsletter/] You can connect with us on LinkedIn: https://www.linkedin.com/company/silica-an-avnet-company/ [https://www.linkedin.com/company/silica-an-avnet-company/]. Or find us at www.avnet-silica.com [http://www.avnet-silica.com].

No app required: The NB-IoT pill dispenser changing medication management in care homes — We Talk IoT #85

Around 50% of medications are not taken on time. In ageing societies, the consequences range from hospital readmissions to preventable deaths. The costs run into millions of euros every year. In this episode, Daniel Böber, hardware developer at wirewire, explains how his three-person team tackled this problem with the ANABOX smart. A connected pill dispenser that runs on NB-IoT, requires no Wi-Fi, and is already deployed in nursing homes across Europe. The patient never touches an app. The carer does. Daniel discusses the deliberate design choice to keep the device analogue-looking and the digital layer invisible to the patient; the regulatory journey through EU MDR Class I certification and pharmacy approval; and what it takes to get a health insurer to add a hardware product to its catalogue. He also reflects honestly on what went wrong, including a full hardware redesign after elderly users couldn't open the compartments, and on wirewire's broader ambition to digitalise everyday objects unobtrusively. #NB-IoT #SmartHealth #IoT #MedicationManagement #HealthTech #AgeTech #WeTalkIoT Summary of this week's episode: 01:37 Wirewire and Mission 02:28 Why Doses Get Missed 03:52 Inside the Smart Pillbox 04:59 Why NB IoT Connectivity 06:30 Caregiver App Design 08:41 Patient Alerts and UX 10:31 Real World Rollout 14:16 Costs and Reimbursement 15:24 Break and Sponsor 15:50 Regulatory Approval Journey 17:06 Insurance Listing Explained 18:09 Scaling Across Europe 19:38 Analog Products Made Smart 21:06 Startup Lessons Learned 22:46 What’s Next and Partnerships 24:22 Hardware Advice and Support   Show Notes: * Daniel Böber (LinkedIn): https://www.linkedin.com/in/danielboe1/ [https://www.linkedin.com/in/danielboe1/] * ANABOX smart: https://anabox-smart.de/en [https://anabox-smart.de/en] * Anmed GmbH: https://anmed.de [https://anmed.de] * Avnet Silica: NB-IoT — Flexibility is Key: https://my.avnet.com/silica/resources/article/nb-iot-flexibility-is-key/ [https://my.avnet.com/silica/resources/article/nb-iot-flexibility-is-key/]   Listen to the "We Talk IoT" Soundtrack on: * Spotify: https://open.spotify.com/playlist/05MOV4OV2MH2in2txsAGtG?si=ad08112cb8d443f4 [https://open.spotify.com/playlist/05MOV4OV2MH2in2txsAGtG?si=ad08112cb8d443f4] * YouTube: https://www.youtube.com/watch?v=D-NvQ6VJYtE&list=PLLqgVFfZhDRVYmpEqbgajzDvGL4kACRDp [https://www.youtube.com/watch?v=D-NvQ6VJYtE&list=PLLqgVFfZhDRVYmpEqbgajzDvGL4kACRDp]   About Avnet Silica: This podcast is brought to you by Avnet Silica — the Engineers of Evolution. Subscribe to our newsletters here: https://my.avnet.com/silica/resources/newsletter/ [https://my.avnet.com/silica/resources/newsletter/] You can connect with us on LinkedIn: https://www.linkedin.com/company/silica-an-avnet-company/ [https://www.linkedin.com/company/silica-an-avnet-company/]. Or find us at www.avnet-silica.com [http://www.avnet-silica.com].

Change the OS, stop the train: Why industrial software stability is non-negotiable – We Talk IoT #84

Most people think they know Windows. Industrial OEMs know better. In this episode, Yann Stridde, Senior Business Development Manager for Microsoft Windows IoT at Avnet Silica, explains why the operating system running a factory robot, a crane in Dubai, or a medical imaging device is fundamentally different from the one on your office PC – and why that difference matters more than ever. Windows IoT Enterprise is not a new product. Its roots go back to Windows Embedded in the late nineties. But the stakes around it have never been higher. The EU Cyber Resilience Act is coming into force. Windows 10 IoT Enterprise has reached end of support. And industrial OEMs face a hard question: how do you manage a 10-year device lifecycle when AI is reshaping the landscape every 18 months? Yann walks us through the certification chain behind a train brake system to show why changing an OS mid-lifecycle can literally stop a train at the station. He explains what Microsoft's 10-year Long-Term Servicing Channel actually guarantees – and where its limits lie. And he makes the case that for small OEMs who have to build their own security patches, peace of mind has a price worth paying. We also explore LiRISE, the ARM-based edge AI solution from Liebherr-Components and Reycom, shown at Embedded World 2026. A real-world example of Windows IoT on ARM running in harsh environments without active cooling, and why power efficiency is now a competitive requirement, not a nice-to-have. Tune in to find out what the industrial world demands from an operating system, and why the answer is not the Windows you know. #windowsiot #embeddedsystems #cybersecurity #cyberresilienceact #iot #oem #arm #lirise #avnetsilica #wetalkiot   Summary of this week’s episode: 01:30 Yann's background: from Microsoft to Avnet Silica 03:00 Why IoT devices need their own version of Windows 06:30 Avnet Silica: advisory, licensing, and consulting for OEMs 08:30 Ten years of LTSC support: realistic in the AI era? 10:30 The train certification story: why OS stability is non-negotiable 15:00 Windows 10 IoT end of support: what OEMs must do now 17:30 Security and the EU Cyber Resilience Act 21:30 Choosing the right embedded OS: the key questions 23:30 LiRISE: Windows IoT on ARM at Embedded World 2026 26:30 ARM architecture, passive cooling, and the power efficiency argument 28:30 Use cases beyond industrial: kiosks, data centres, and more   Show notes: Yann Stridde: https://www.linkedin.com/in/yannstridde/ [https://www.linkedin.com/in/yannstridde/] Avnet Silica – Windows IoT Embedded Software: https://my.avnet.com/msembedded/embedded-software/overview [https://my.avnet.com/msembedded/embedded-software/overview] LiRISE® – Liebherr-Components and Reycom IoT solution: https://www.liebherr.com/en-us/components/solutions/iot-solutions/e2e-iot-edge-gateway-solution-6654887 [https://www.liebherr.com/en-us/components/solutions/iot-solutions/e2e-iot-edge-gateway-solution-6654887]   Listen to the “We Talk IoT” Soundtrack on: Spotify: https://open.spotify.com/playlist/05MOV4OV2MH2in2txsAGtG?si=ad08112cb8d443f4 [https://open.spotify.com/playlist/05MOV4OV2MH2in2txsAGtG?si=ad08112cb8d443f4] YouTube: https://www.youtube.com/watch?v=D-NvQ6VJYtE&list=PLLqgVFfZhDRVYmpEqbgajzDvGL4kACRDp [https://www.youtube.com/watch?v=D-NvQ6VJYtE&list=PLLqgVFfZhDRVYmpEqbgajzDvGL4kACRDp]   About Avnet Silica: This podcast is brought to you by Avnet Silica—the Engineers of Evolution. Subscribe to our newsletters here: https://my.avnet.com/silica/resources/newsletter/ [https://my.avnet.com/silica/resources/newsletter/] You can connect with us on LinkedIn: https://www.linkedin.com/company/silica-an-avnet-company/ [https://www.linkedin.com/company/silica-an-avnet-company/]. Or find us at www.avnet-silica.com [https://www.avnet-silica.com].

100 TOPS, Zero Production: Why Edge AI Projects Die Between Demo and Deployment – We Talk IoT #83

Edge AI promised intelligence everywhere. The reality? Most projects die between proof of concept and production. In this episode, Amir Sherman from DeepX and Michaël Uyttersprot from Avnet Silica reveal why moving AI from comfortable development kits to demanding industrial environments isn't just difficult. It's a maze of incompatible metrics, hidden power costs, and integration nightmares that catch companies off guard. We discuss why TOPS ratings mislead engineers, how ChatGPT triggered a wave of failed internal deployments, and what it takes to run vision AI in factories, delivery robots, and smart cities where five-watt power budgets matter more than marketing specifications. From Hyundai's factory robots to Baidu's Chinese character recognition systems, Amir and Michaël share real deployments that work, and explain the 50 years of embedded experience that AI code generators cannot replace. If you've wondered why edge AI keeps hitting walls nobody discusses in vendor presentations, this conversation delivers the answers. #EdgeAI #DeepX #IndustrialIoT #VisionAI #NPU #WeTalkIoT   Summary of this week's episode: 01:45 Guest Introductions: Amir Sherman and Michaël Uyttersprot 03:29 The Edge AI Market Landscape: TinyML to High-End NPUs 06:47 Why Edge AI Projects Fail in Production 09:18 The TOPS Trap: Why Performance Metrics Mislead 11:50 Choosing Hardware: Why It's More Complex Than It Looks 16:28 Real Deployments: Delivery Robots and Chinese OCR 19:29 Vision AI in European Factories and AGVs 21:32 Generative AI at the Edge: 20 Billion Parameters in Five Watts 23:52 Smart Cities: Traffic Lights and Vision Language Models 27:22 What Engineers Need to Understand About Edge AI 30:48 Why 50 Years of Experience Still Matters 33:13 Bohemian Rhapsody and AI-Generated Llama Songs   Show notes: Amir Sherman: https://www.linkedin.com/in/sherman-bizdev/ [https://www.linkedin.com/in/sherman-bizdev/] Michaël Uyttersprot: https://www.linkedin.com/in/micha%C3%ABl-uyttersprot-aaa971211/ [https://www.linkedin.com/in/micha%C3%ABl-uyttersprot-aaa971211/]   DeepX: https://deepx.ai/ [https://deepx.ai/] DeepX at Avnet Silica: https://my.avnet.com/silica/manufacturers/m/deepx/ [https://my.avnet.com/silica/manufacturers/m/deepx/]   Listen to the "We Talk IoT" Soundtrack on: Spotify: https://open.spotify.com/playlist/05MOV4OV2MH2in2txsAGtG?si=ad08112cb8d443f4 [https://open.spotify.com/playlist/05MOV4OV2MH2in2txsAGtG?si=ad08112cb8d443f4] YouTube: https://www.youtube.com/watch?v=D-NvQ6VJYtE&list=PLLqgVFfZhDRVYmpEqbgajzDvGL4kACRDp [https://www.youtube.com/watch?v=D-NvQ6VJYtE&list=PLLqgVFfZhDRVYmpEqbgajzDvGL4kACRDp]   About Avnet Silica: This podcast is brought to you by Avnet Silica—the Engineers of Evolution. Subscribe to our newsletters here: https://my.avnet.com/silica/resources/newsletter/ [https://my.avnet.com/silica/resources/newsletter/] You can connect with us on LinkedIn: https://www.linkedin.com/company/silica-an-avnet-company/ [https://www.linkedin.com/company/silica-an-avnet-company/] Or find us at www.avnet-silica.com [http://www.avnet-silica.com]