I share my personal, unfiltered thoughts on Crypto, AI, NFTs, and Tech.
Nov 17, 2026
•
2 min read
So who keeps financing the AI buildout?
Nov 11, 2026
Now what?
Nov 9, 2026
Oct 27, 2026
5 min read
The physical record tells a different story.
Oct 21, 2026
3 min read
A 20-company review of the repayments & customer credits that shape future margins and cash flow.
Research
Oct 15, 2026
Meet Tessara.
Oct 8, 2026
Contract structure may determine who captures the next leg.
Oct 1, 2026
Our forecast scorecard and the economics behind the next quarter.
Sep 29, 2026
Our model points to $56.16B. Here is the assumption that could break it.
The Chokepoint
Sep 3, 2026
6 min read
AI boards are getting thicker, denser and harder to drill. The tiny tools making the holes are wearing out faster.
Aug 28, 2026
AI needs power faster than the grid can deliver it. Fuel cells are becoming the workaround.
Aug 20, 2026
7 min read
MLCC cost per rack is up 182% into VR200. The rack itself only doubled.
Earnings
Aug 3, 2026
Here is what we missed.
Jul 28, 2026
Tessara's numbers, assumptions and probability range
Jul 27, 2026
1 min read
What will actually matter on the call
Jul 22, 2026
4 min read
Both report today. Our frame on each, and the calls we’ve put on the record before the prints.
Jul 16, 2026
Our proprietary read on TSMC before the print + our ASML grade
Jul 14, 2026
Our proprietary read on ASML before the print
Jul 8, 2026
The AI shortage is spreading to chips designed forty years ago, made in fabs nobody builds anymore.
Jul 3, 2026
16 min read
The chips, advanced packaging, substrates, cooling, and racks all run through one island. Plus 7 growing names that ride it
Jul 1, 2026
The 2nm bottleneck is somewhere else. One etch step, three suppliers.
Jun 24, 2026
Three companies make the world's big gas turbines. Two blade manufacturers decide how fast they can build them.
Jun 23, 2026
A live test of Tessara’s constraint model for the AI buildout
Jun 17, 2026
The AI fiber shortage is a glass-rod shortage and you can't make the rods on a GPU schedule.
Jun 15, 2026
9 min read
Why AI capacity is not the same as GPU shipments - 8 layers between wafer and live cluster.