Blog
Written for people to read and for AI to cite, so every claim keeps its source, its test and the way it could be proven wrong.
Building a Moving Espresso Machine in a Day: Drawing It Part by Part, and Simulating a Shot With Four Formulas
On October 14, 2026 I replaced the blog's espresso lab with a 110-part machine that moves with the sliders. Here is how it was drawn, how the code checks its own drawing for mistakes, and the four simplified formulas behind the extraction simulation, along with where they go wrong. A learning note, not advice.
TechWave EP159: Musk Bought 800 MHz — and Then What?
Notes on TechWave EP159 (12 October 2026): SpaceX acquired US 800 MHz low-band spectrum on 8 October (817–824 MHz uplink, 862–869 MHz downlink), US carrier shares fell 8–9% that day, and host Harry works through path loss and beam capacity to explain why Starlink fits better as a backup network. Educational listening notes only — no investment advice, no price targets, and the judgements here may be wrong.
Betting Before the Evidence: What Muse Wonder's Launch Reveals About Signing the Check First
Notes after listening to the 12 October 2026 Cheeky Pint with Mark Zuckerberg and Alex Wang. Why Meta committed hundred-billion-dollar spending before seeing a model, why they moved evaluation in-house, and where that argument is weakest. Educational discussion of method; no investment advice and no view on any stock.
When the data center gets big enough, the corner parts become a business — notes on The Circuit's Marvell analyst day episode
Listening notes on The Circuit, 2026-10-12: why Marvell's five-year targets came in above sell-side models, and where heterogeneous racks, BMCs and grid interconnection set the limits. Educational industry observation, no individual stock advice.
Ten Thousand People Serving 140 Million: David Vélez on Banking's Second Half
Notes after listening to Invest Like the Best EP.495. David Vélez's numbers: global consumer finance is a $10T industry with ~90% still flowing through branches; Nubank serves 140M customers with 10,000 employees at roughly 3% of an incumbent's cost to serve. Educational discussion of his reasoning and where my own read was wrong — not investment advice, no stock recommendations or price targets.
Three Shots in One Clip, 24 fps Doubled to 48: An AI Short Film That Took Four Runs
A small experiment log: making the 27-second AI short 'A Little More Every Day' with MiniMax H3 on a Google Colab A100. New this time: cutting three shots inside one clip, pinning storyboard stills to exact frames, and doubling 24 fps to 48 with open-source RIFE. Timings for each step, what broke in each of the four runs, the compute units spent, and whether the dialogue actually came out. The character is an AI creation.
He Bought a Mac Studio and Won't Run Local Models on It: Xiaotian's “Dumplings for a Dab of Vinegar,” and What I Found Timing Seven Days of My Own AI Work
In an October 12, 2026 video, Xiaotian (小天fotos) says he bought the 96 GB Ultra Mac Studio and doesn't plan to run local models on it. When he broke down one After Effects project an AI agent did for him, model output was only 22% of the time; most of the rest was waiting on tools and renders. I broke down seven days of my own AI work and the model's share came to 29%. This post covers his vinegar-and-dumplings idea, lines up six big-memory machines by their official specs, and asks whether it's worth buying a big-memory computer just to run models. A technical learning note, not business advice.
A Lithography Machine That Never Sold, Reborn Two Decades Later as an ASML Inspection Tool
Asianometry's 11 October 2026 episode on ASML's eScan 2200 multibeam inspection tool, whose technology came from Mapper Lithography, the Dutch startup that went bankrupt in 2018. Listening notes: the twenty-year timeline plus my own reading. Educational only — no investment advice, price targets or stock recommendations.
Making AI Short Clips with MiniMax H3 on Colab: Is 2K Worth It?
On October 12, 2026 I ran the MiniMax H3 video model through ComfyUI on a Google Colab Pro A100, generated 3-second clips at 1344×768, and upscaled them to 2K with Real-ESRGAN. Each clip took 206 seconds end to end with a GPU memory peak around 52GB. After comparing both versions, 768 was enough; the story is what decides whether a clip works. Includes the Colab gotchas I hit, the license terms worth reading, and three routes to 2K side by side. A personal learning note, not business advice.
Reshooting an AI Short Film: Testing Three Quality Settings for MiniMax H3 on a Colab A100
A lab note: nobody understood the first cut of our AI short film Daijoubu, so I rewrote the script around four problems, generated a fresh starting photo for every shot, and tested three quality settings on two shots on a Google Colab A100 before shooting all 13 clips. Timings and compute units for each setting, what the full shoot cost, a first run that produced nothing, and how the characters got to say their own lines this time. All characters are AI.
Sixty Years Pegged to the Dollar: The Bahamas' Central Bank Governor on Exchange Rate Anchors, Reserve Buffers, and the World's First Nationwide CBDC
Listening notes on the October 12, 2026 episode of Macro Musings with Bahamas central bank governor John Rolle. Covers the fixed exchange rate, reserve buffers, procyclical policy, and the Sand Dollar's design, plus the strongest objections and the conditions under which the setup breaks. Educational macro notes, no individual stocks, no investment advice.
Grown, Not Written: Notes on Nate Soares and AI Risk
Notes on the 2026-10-12 Odd Lots interview with MIRI president Nate Soares, unpacking his argument that reactive safety fails structurally for this technology, plus the conditions under which it breaks. Educational, not investment advice.
One Week of Public Posts, Laid Side by Side: What "Sold Out Through 2029" Actually Commits To
An independent researcher covering the AI infrastructure upstream posted eleven times on X between 2026-10-05 and 10-10. Lined up together, the posts put the tight spot in lasers and substrates — and put the date it loosens into public documents.
Podcast Notes Weekly: Oct 05–Oct 11, 2026
The takeaways from 22 podcast notes I wrote this week (Asianometry, Macro Musings with David Beckworth, Bloomberg Odd Lots, The Circuit, Invest Like the Best, Tech Wave (科技浪) and more). Two to four sentences per episode, with a link to each full post.
Gooaye EP704 Notes: When $70B Became $50B, and Why It Didn't Move His Position
Listening notes on Gooaye EP704 (2026-10-10). How to tell an accounting-definition dispute from something actually breaking, using OpenAI's revised annualized revenue and Marvell's investor day guidance. Educational, not investment advice; no tickers recommended, no price targets.
A Company Earning 30% Margins Chose to Leave the Stock Market: Nakano on His MBO (Investor's Sunday)
The October 11, 2026 episode of Investor's Sunday, part two: Ihara Science chairman Tokuo Nakano on the 2023 MBO — why a profitable company went private, how a ¥30bn loan got approved, and the problem that got harder after the shareholders left. Listener notes for education only; not investment advice.
Yields at a 20-Year High, Stocks at a Record High — Notes on MacroMicro After Meeting EP. 217
Listening notes on MacroMicro's After Meeting EP. 217 (11 October 2026): the US 10-year yield reached 5.3% while equities made new highs. Breaking a single rising number into its three components, and why the term premium is the line worth watching. Educational content, not investment advice; no individual stock recommendations or price targets.
Does the Bigger Bankroll Always Win? A Man Who Bought Nokia Every Month Waited 25 Years to Break Even
A popular saying claims that with enough capital and patience you always win in the end. Roulette math and real dot-com stock prices show which half is true and which half hurts people. Educational and methodological, not stock advice.
AI Keeps Getting Smarter. Can the Hardware Keep Up? Notes on TechWave EP158 with Infineon Taiwan's Managing Director
On 9 October 2026, TechWave EP158 interviewed Chih-Hao Chen, Managing Director of Infineon Taiwan, on data-center power delivery, software-defined vehicles, and humanoid robots. These are my listening notes and personal reading — educational discussion only, no investment advice, no stock recommendations, no price targets.
TSMC in Texas? Turning "Anchor Customer" Into an Actual Number
Notes after listening to Asianometry's 2026-10-08 episode "TSMC is Coming to Texas?" — working back from Arizona's $265B to estimate how large a customer commitment a Texas fab would need, and where that chain of reasoning breaks. Educational, not investment advice; no price targets or stock recommendations.
An Old Path, Reopened: Notes on Caibaogou Episode 569
The 2026-10-08 episode of Caibaogou covers China's export curbs on gallium and quartz, a proposed US tariff on Chinese-made lasers, and a VCSEL redesign that stretches reach from 100 metres to two kilometres. Listening notes and supply-chain structure, for education only — no investment advice, no stock recommendations, no price targets.
When the Rubber Meets the Road: Jurrien Timmer on Rates, Earnings, and 60/20/20
Listening notes on the October 9, 2026 episode of The Compound and Friends (#263). Fidelity's director of global macro Jurrien Timmer on how a 5.3% 10-year yield and 2.9% real rates compress valuations, why a falling P/E doesn't rule out risk, and how to think about allocation once the 60/40 premise breaks. Educational notes only — not investment advice, no stock picks, no price targets.
After the Dollar Cleared 102, I Crossed "The Fed Is Out of Bullets" Off My List
Listening notes on MacroVoices Episode 553 with Brent Johnson: dollar strength, revaluing the gold on the books, and a Treasury that keeps crossing into the Fed's lane. Educational notes, not investment advice; no tickers, no price targets, and the conclusions need rerunning when market conditions change.
Should AI Agents Run Your Asset Allocation? Andrew Ang First Shows Taxes Take a Third
Listening notes on The Meb Faber Show #653 (2026-10-09): Columbia professor Andrew Ang calculates that taxes drag 160–170 bps a year on taxable US equity investors, compounding to over a third across thirty years, and discusses the AI agent swarm he is building for asset allocation. Educational commentary, not investment advice; no stock recommendations or price targets.