620: Will Nvidia be the West's Open Source Champion?, Meta's Custom Silicon Bet, Mark Leonard Health News, Google AI's Flaws, American Health Puzzle, Ukraine's 7 Million Drones, and a Ride with Jensen
"cooking with magnets never stops being cool"
Many people die at twenty-five and aren’t buried until they are seventy-five.
—Commonly attributed to Benjamin Franklin, likely apocryphal
🧲🍳 I’ve now had my induction stove for long enough to update you on the experience:
It rocks!
I wouldn’t go back to anything else. It basically combines the best parts of gas and resistive electric, plus a few unique advantages.
My favorite benefit is something that seems small on paper, but that makes a big difference in everyday life: cleanup.
Because induction heats the cookware directly, the cooktop never gets anywhere near as hot as it does on gas or resistive electric, so if anything spills, it doesn’t bake into a carbonized crust that is a pain to remove. You just have to wipe it down with a damp towel and you’re done!
And, well, cooking with magnets never stops being cool 🤓
(The diagram above was a good opportunity to test the new Nano Banana 2. It did a pretty good job, based on my fact-checking)
🧊🌧️🚘🌊 Something really strange happened to me this week.
On Wednesday, the forecast was for 20-25cm of snow (8-10 inches), but it ended up coming down as rain before turning into freezing rain. This caused a bunch of tree branches to break, and there were widespread power outages across the area. 🔌🚫
My house lost power for roughly 26 hours. When I woke up yesterday, I couldn’t make myself my usual Sencha or espresso, so I decided to get mediocre coffee at the nearest Tim Hortons ☕
The car was all iced up. It took me a while to get it ready, but when I finally drove off… there was a very loud sloshing noise coming from all around me! 😬
It was so surprising that, for half a second, I wondered if I was on drugs.
The impression it gave was that the car’s body panels were half-filled with liquid, and every time I accelerated or braked, it sloshed around loudly. I stopped on the side of the road, walked around the car, and didn’t notice anything unusual.
So, of course, I asked AI what this could be.
✅ Most likely explanation: Meltwater trapped in one of the doors because the drain holes froze shut during the ice storm.
Car doors are not sealed watertight. Water normally enters through the window seals and drains out through small drain holes at the bottom of the door.
During an ice storm:
Ice can block those drain holes.
Meltwater from defrosting the car can run down into the door.
When you drive, you hear sloshing inside the door cavity.
Good to know!
Hopefully, if this ever happens to you, you won’t think you’re losing your mind 😵💫
🏦 💰 Liberty Capital 💳 💴
The West Needs an Open Source AI Champion. Meta Fumbled. OpenAI is Distracted. Will it be Nvidia?
There’s a real Western open-model gap, and Nvidia has unusually good incentives to fill it. Wired is reporting this:
NVIDIA WILL SPEND $26 billion over the next five years to build open source artificial intelligence models, according to a 2025 financial filing. Executives confirmed the news, which has not been previously reported, in interviews with WIRED.
I went and looked at the filing myself, and as far as I can tell, the $26 billion refers to multi-year cloud service commitments, not model-training spend specifically. Here’s the exact footnote:
Multi-year cloud service agreement commitments as of October 26, 2025, were $26 billion for which $1 billion, $6 billion, $6 billion, $5 billion, $4 billion, and $4 billion will be paid in fiscal years 2026 (fourth quarter), 2027, 2028, 2029, 2030, and 2031 & thereafter, respectively.
Some cloud service capacity may be reduced, terminated or sold to others by the CSPs, in which case our commitments will be reduced. We expect cloud service agreements to be used to support our research and development efforts and DGX Cloud offerings.
I don’t know what exactly Nvidia told Wired — if they gave more precision on how much of that capacity is actually earmarked for open models, it’s not in the piece. But let’s think through Nvidia’s open source strategy based on what we do know:
