Shop Talk
Capturing this week’s zeitgeist
What’s really great is that Taiwan is a strategic partner in reindustrializing United States. Look, we’re trying to set up fabs in United States, and what do they do? They send a bunch of engineers. They invest a whole bunch of money. It’s not easy building up a world-class semiconductor manufacturing in the United States because there’s just not enough talent there.
Jensen Huang with Lex Fridman
Hansei-kai
How Apple turbocharged China’s development /NPR/
How Industrial Espionage Started America’s Cotton Revolution /Smithsonian/
Please spread the word of Exponential Industry by sharing the publication
Assembly Line
Thought provoking stories across the industrial world.
🛢️🖥️📉 Old Man Coffee: Decline Curve
✍️ Author: Chuck Yates
A shale well doesn’t decline like a glacier. Because of the massive fracs it takes to make a horizontal well produce at all, it declines like a kid’s birthday balloon. It comes on like a howitzer, but declines sixty or seventy percent in the first year, and about a third of everything that well will ever produce is gone before your first or second redetermination. Eventually every one of them bends over, flattens out, and settles into the same lazy seven-to-ten-percent decline a conventional well has. The whole game is where the bend happens. Early bend, great well. Late bend, you just bought a very expensive hole.
...
Chips get there from two directions at once. One: Nvidia keeps shipping a better, faster, more efficient chip every year, and each one turns the last one into the Nokia in your junk drawer. Two, and this is the shale one: everybody’s building every data center they can get permitted, at the highest prices chips have ever sold for, while the models keep learning to run on less. Half the compute. A cheaper chip. Your phone. Here’s the weird part: that actually extends the life of the chip. It’ll keep running. It just runs in a world with way more compute than anybody needs, which means the rent on compute goes to thirty-dollar oil. The hardware doesn’t die. The price does. And a lease payment doesn’t care which one.
Read more at Collide
The AI boom is threatened by an epidemic of unreliable numbers /Economist/
The Staggering Costs of a Single AI Data Center /Bloomberg/
⚙️ This Chicken Farmer Now Runs a Thriving CNC Machining Business
✍️ Author: Peter Zelinski
The contrast between chickens and CNC is striking—just this much makes Gary’s story compelling. But there is a larger point. Machining, programming and engineering are adjacent skills to running a successful business, but they are not the core skills needed. Indeed, a great machinist could prove to be a hapless machine shop owner. Something different is needed, and part of the talent we should be eager to welcome into manufacturing is talent for overseeing its enterprises. However, part of that talent finding welcome is having the right mindset and attitude for stepping in.
Read more at The Zelinski Report
🗜️📉 China Gigacasting Glut?
Recently, Guangdong Hongtu disclosed key production and operating data for its die-casting business during an institutional research session. The company’s overall capacity utilization currently remains above 70%. As newly built plants gradually ramp up to full production and capacity continues to climb, overall utilization is expected to rise above 80%, mainly to customers like Xpeng.
Read more at Industry Arsenal
Why the AI Boom Isn’t Showing Up in U.S. Fab Utilization /Richmond Fed/
The Battery Boom Moving From Electric Cars to the Power Grid
🗣️ A Class Project
Tim Latimer, a drilling engineer, came to Stanford to start a geothermal company using techniques from the shale boom. As the lone MBA in engineering classes, he won over a skeptical professor, who pointed him to Jack Norbeck, a PhD with expertise in the field and the co-founder Tim had been looking for. Just as the two got started, a stranger’s conference paper proposed bringing shale techniques to geothermal. “We’ve been scooped,” Tim told Jack. But the stranger, Christian Gradl, didn’t see a rival. He saw a team.
New Product Introduction
Highlighting new and innovative facilities, processes, products, and services
🖨️3️⃣ Introducing Form Studio 1 - a professional resin 3D printer, starting at $699
Form Studio 1 ships open. You get access to Formlabs materials and to third-party resins. Both resin tanks have a lifespan of 75,000+ layers (approximately 200 prints at 37 mm tall, at 100 µm layer height). Every tank comes with a reusable lid that protects resin from light and debris when it’s out of the printer, and lets you stack tanks for compact storage. If you run multiple resins, we recommend keeping a dedicated tank per material, so switching is as simple as swapping in a new cartridge and its matching tank.
Read more at Formlabs
Is Resin 3D Printing Safe? Material Scientist Answers Your Questions /YouTube/
📐 USD BRep v1.0.0
USD BRep is a standalone C++ and Python library for NURBS-based boundary representation (BRep) solid modeling, with native import and export of BRep data to and from OpenUSD /GitHub/. It has two public APIs, each with its own reference below: the usd_brep Python package and the C++ headers.
So what is it, really, and (why) does it matter? 👉 /István Csanády on X/
Read more at NVIDIA
Frontier Models for Every Robot /π/ Announcing our API beta. You upload data, launch fine-tuning on compute and infrastructure we manage, and access fine-tuned models via a hosted gateway to run inference. The latest model available through the API already provides more data-efficient fine-tuning and improved capability for our partners. Since Pi can update the models underpinning the API rapidly, we plan to continue enabling access to our latest and most capable models. Launching the beta is our first step in providing the physical intelligence foundation the community can build upon and enabling end-to-end autonomy.
🏎️🪽 The FLYING Tesla Roadster explained by the man who engineered it
Tesla's Flying Roadster Revealed: Everything That Happened in 10 Minutes /YouTube/
Tesla-SpaceX Deal News Possible This Year, Barclays Analyst Says /Bloomberg/
Business Transactions
This week’s top funding events, acquisitions, and partnerships across industrial value chains.
🤝 Honeywell Technologies Launches Forge Cognition with NVIDIA and Dell Technologies, Advancing Physical AI Across Buildings and Industrial Operations /Honeywell/
🇺🇸 🧪🧠 Albert Raises $55M Series B to Scale the First Chemistry-Native AI
Albert, which has built the first chemistry-native AI operating system for R&D, announced a $55 million Series B led by TCV with participation from new investors Delta-v Capital, Socium Ventures, HighSage Ventures, Coral Capital, and Legion VC. They are joined by existing investors Coatue, J.P. Morgan Growth Equity Partners, Index Ventures, and Applied Ventures, LLC. The new funding will be used to expand Albert’s chemistry-native AI capabilities, shortening development cycles and reducing time to market for the everyday products used by billions of people.
Albert is also announcing the industry defining success of its chemistry-native model “Valence” which new research shows is 45x more accurate than general-purpose AI at retrieving the correct answers to scientific queries /Albert/. Valence achieves this improved performance because it was trained to discriminate chemical composition and function while reading text, the way an experienced chemist would.
Read more at Business Wire
🇬🇧🧪 HydRegen Raises $8M (£6M) Series A to Bring Biocatalysis to Industrial Chemical Manufacturing /HydRegen/
🇦🇷🧪 Giraffe Bio raises $6.4M Seed to scale next-generation reagents for copper mining /GlobeNewswire/
🇮🇳⚗️ Scimplify raises $25M to speed up critical materials production with AI
Indian specialty chemicals startup Scimplify has raised a $25 million all-equity funding round led by Hitachi Ventures with participation from corporate customers Henkel and Teknor Apex alongside existing investors, bringing its total equity raised to $80 million at a reported $250 million post-money valuation. Operating primarily out of India, the company manufactures critical materials and specialty chemicals including tungsten derivatives from scrap and magnet rare earths using a rapid-turnaround production model that delivers customer requirements in two months rather than the typical 12 to 18. Scimplify plans to use the newly raised capital to expand its international manufacturing footprint beyond India into Japan and the United States (allocating $7.5 million specifically to the US), deliberately bypassing European production capacity to build alternative non-Chinese supply chains for the energy, electronics, and defense sectors.
Read more at TNW
🇨🇭⚗️ UniSieve raises $9M to scale high-precision membrane solution for industrial separations /UniSieve/