AI doesn’t run on dreams, it runs on power, and lots of it. With Trump calling for coal to fuel data centers, the conversation around energy and AI just got a lot hotter. As an engineer, I get the grid reliability argument—especially when demand is exploding—but I also see why the hyperscalers aren’t biting. There’s a tension here: keep the lights on, or keep the ESG folks happy. Whether you like the idea or not, coal’s not dead yet—and if you’re designing for the edge of the cloud, that matters more than ever.
Top Stories This Week
- Trump Wants Coal To Power AI Data Centers
- Cornerstone Robotics Raises Over $70 Million To Commercialize Its Robotic Surgery System
- Nvidia Reportedly Develops New Lower-cost AI Chip For China To Skirt US Export Curbs
- Wolfspeed’s Collapse Threatens SiC Market: US$2B Renesas Deal In Jeopardy
- On-orbit Computing Startup Sophia Space Raises $3.5m
- Taiwan Removes Lithium Battery-powered Air Quality Sensors Over Safety Concerns
- Hardware AI Turns Street Cams Into Rainfall Sensors
- Scientists Create Soft Robots Able To Walk Out Of Printers Which Made Them
- Sustainable Fully Inkjet-Printed Humidity Sensor Based On Ionic Liquid And Hydroxypropyl Cellulose
- Reconfigurable Flexible Complementary Circuits Based On Polarity-Configurable Carbon Nanotube Transistors
- Innatera Unveils Pulsar, The World’s First Mass-market Neuromorphic Microcontroller For The Sensor Edge
Hardware Business News
Cornerstone Robotics Raises Over $70 Million To Commercialize Its Robotic Surgery System
It’s not often I find surgical robotics particularly relatable, but Cornerstone Robotics has caught my attention—and for good reason. With over $70M in fresh funding and a practical, no-nonsense approach to in-house R&D, they’re proving that advanced medical systems don’t have to come with bloated price tags or empty promises. Their Sentire platform is already making waves in China, and now they’re gearing up to scale globally. While some startups chase trends, Cornerstone seems focused on delivering real engineering that matters.
Nvidia Reportedly Develops New Lower-cost AI Chip For China To Skirt US Export Curbs
With U.S. restrictions tightening, Nvidia isn’t backing down—they’re adapting. A new, lower-spec Blackwell-based GPU is reportedly in the pipeline for China, priced between $6,500–$8,000 and using GDDR7 memory to stay within export limits. While it’s not a powerhouse like the H20, it’s a smart play to maintain relevance in China’s $50B datacenter market. I admire the strategic pivot; it’s a reminder that great engineering isn’t just about specs—it’s about surviving the geopolitical game. And with CUDA still leading the stack, Nvidia knows exactly what cards it’s holding.
Wolfspeed’s Collapse Threatens SiC Market: US$2B Renesas Deal In Jeopardy
Wolfspeed’s financial troubles could send serious voltage spikes through the global silicon carbide industry. With Renesas already $2B deep in a wafer supply deal, the timing couldn’t be worse—especially as EV momentum wavers and Chinese firms ramp up production. I respect Wolfspeed’s bold vision, but in power electronics, as in business, grounding matters. This isn’t just about one company; it’s about the fragility of cutting-edge supply chains under economic stress. For those watching SiC closely, now’s the time to check your insulation.
On-orbit Computing Startup Sophia Space Raises $3.5m
Seattle’s Sophia Space just secured $3.5M to develop radiation-hardened, AI-ready orbital compute modules—what they call “TILEs.” It’s an ambitious pitch: space-based Edge computing that reduces reliance on Earth’s power and water. It’s fascinating to see the boldness of these orbital infrastructure plays, but I remain sceptical. Space is a brutal place for hardware, and while the cold vacuum sounds ideal for cooling, it’s not exactly maintenance-friendly. Still, with ex-NASA and big tech veterans onboard, it’s one to watch—if only to see just how far from Earth this Edge trend can go.
Hardware Engineering News
Taiwan Removes Lithium Battery-powered Air Quality Sensors Over Safety Concerns
Here’s a story that blends environmental tech with unintended consequences. Taiwan’s science council just pulled 16 lithium battery-powered air quality sensors after one allegedly sparked a wildfire in Yangmingshan National Park. While I get the appeal of lithium for off-grid deployments, in dry rugged terrain, this might not have been the best choice when considering that thermal runaway is no joke. This incident is a reminder that our push for smarter monitoring must also respect the environments we’re protecting. There’s a balance to be struck between innovation and risk—especially when nature is on the line.
Hardware AI Turns Street Cams Into Rainfall Sensors
Why build a new sensor network when you’ve already got cameras on every corner? A team in Tianjin figured out how to turn existing surveillance cameras into accurate rainfall detectors using a clever mix of AI, image analysis, and deep learning. It’s a textbook example of doing more with less—no new hardware, just smarter software. I’m always impressed by this kind of practical engineering that skips the hype and gets results. And with urban flooding on the rise, having real-time, hyperlocal rain data might be more valuable than ever.
Scientists Create Soft Robots Able To Walk Out Of Printers Which Made Them
This one really caught my eye—Edinburgh researchers have built a sub-£400 desktop printer that makes soft robots, which actually walk right off the print bed. It’s pneumatic, electronics-free, and surprisingly accessible for hobbyists or labs. Cutting out the complex tooling and long dev cycles could open up soft robotics to real-world applications—especially in places like space or hazardous environments where resilience and simplicity matter more than bells and whistles.
Hardware R&D News
Sustainable Fully Inkjet-Printed Humidity Sensor Based On Ionic Liquid And Hydroxypropyl Cellulose
Here’s a clever bit of materials science—researchers have developed a fully inkjet-printed humidity sensor that not only performs well but also makes a visible color change as it reacts. Made with ionic liquids and hydroxypropyl cellulose on a flexible base, it’s fast, accurate, and cheap to produce. But what really caught my attention is its sustainability angle. With e-waste on everyone’s radar, this kind of low-impact, high-utility design could be key for future IoT deployments. Smart, simple, and actually usable—just how it should be.
Reconfigurable Flexible Complementary Circuits Based On Polarity-Configurable Carbon Nanotube Transistors
A new class of reconfigurable thin-film transistors based on carbon nanotubes is pushing flexible electronics to the next level. By leveraging reversible electrostatic doping via a polymer layer on a Y?O?-passivated channel, researchers have created TFTs that can flip between P- and N-type behavior on demand—without sacrificing performance or stability. These ambipolar devices not only simplify CMOS logic design but also enable compact, programmable circuits. The standout demo? A single module that can morph into NAND, NOR, XOR, or XNOR gates. This kind of real-time configurability could revolutionize hardware security and adaptive electronics.
Open-Source Hardware News
Innatera Unveils Pulsar, The World’s First Mass-market Neuromorphic Microcontroller For The Sensor Edge
What happens when AI stops trying to mimic servers and starts thinking more like a brain? You get something like Innatera’s Pulsar—a neuromorphic microcontroller that processes sensor data locally with almost no power draw. It’s not about chasing flashy benchmarks; it’s about enabling real-world, edge AI where every microwatt counts. This kind of chip doesn’t just improve efficiency—it redefines where and how we can deploy intelligence.
