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.
Summer Insects and Ice: The Week the Market Sold While the Order Books Said Sold Out
An independent researcher I follow kept posting through the late-July AI deleveraging. Line up the public numbers he pointed at that week and a strange picture emerges: prices collapsing, while earnings, capex guidance and capacity allocation all point the same way — not enough to go around.
Gooaye EP683: You Cannot See the Mountain From Inside It
Notes on Gooaye EP683: backtesting whether you could have survived a leveraged drawdown badly overstates your resilience, because in a backtest you can see where it ended and in real time you cannot. Why this sell-off produced no capitulation volume. And one usable self-check on position size: if it is affecting your life, the position is too big. Educational notes, not investment advice.
MacroMicro EP209: The Same Capex Was Vision Last Quarter and a Crime This Quarter
Notes on MacroMicro After Meeting EP209: big-tech revenue and AI monetisation are still accelerating, yet the stocks are being punished for free cash flow turning negative. What changed is not the fundamentals but which line item the market prices. Plus the Microsoft capex 'cut' that was really a ten-year change in depreciation schedule, and why the death of forward guidance means watching monthly prints instead of officials' phrasing. Educational notes, not investment advice.
Gooaye EP684: Water Carries the Boat, and Water Capsizes It
Notes on Gooaye EP684: July's sharp drawdown had no matching fundamental trigger — it was a market-wide deleveraging. A highly levered AI-thematic fund was liquidated; its longs were the year's best-performing AI hardware and its shorts were software, so both sides broke at once. My extensions: why the diversification I thought I had may be fictional, why a 50% drawdown carries no information by itself, and why position size is a risk dimension of its own. Educational notes, not investment advice.
Gooaye EP682: How to Read a Sell-Off With No Bad News
The core observation in Gooaye EP682: this drawdown has no identifiable macro or industry trigger, which makes it look more like a valuation reset after an overheated spring. My notes on the episode, plus three extensions — why flat pricing is not the same as deteriorating profit, which side to believe when verifiable signals and price action disagree, and how the episode's advice to use moving averages as stops squares with my own backtest of moving averages. Educational notes, not investment advice.
MacroMicro EP208: Separate the Long Lead Items From the Short Ones, and You Can Read the AI Cycle
The main contribution of MacroMicro's After Meeting EP208: it splits the order-cut risk for the second half into two layers. Cuts to ordinary components only move the manufacturing cycle around; cuts to genuinely scarce parts are what a real turn looks like. Plus TSMC's inventory rising for two straight quarters and why that number has two meanings, Korea's leverage-on-leverage unwind, and why Kimi K3 did not trigger another DeepSeek moment. Educational notes, not investment advice.
Gooaye EP681: The Numbers Are Right, the Price Might Not Follow
The core tension in the latest Gooaye episode: nearly every AI supply-chain fundamental is on track, some even ahead of plan, yet prices are hard to call. The host's answer is to script multiple scenarios and hide in cheap, defensive names. My notes, plus an honest boundary: a mention is not a buy signal.
Statementdog (財報狗) EP539 Notes: While Everyone Stares at the AI Endpoint, I'm Watching Where the Shovel Gets Stuck
Personal notes from Statementdog (財報狗) Podcast EP539: Meta renting out compute is a business model taking shape, not a surplus of capacity; AI infrastructure hits hard physical bottlenecks in power and water; capital should look upstream from the endpoint toward mature nodes and silicon wafers; memory makers flip cash-flow structure with ultra-strict long-term contracts. Plus how the episode echoes two things I keep practicing — chokepoint-layer thinking, and reading that 'cheaper isn't peaking.' Educational notes, not investment advice; disclaimer at the end.
Dingmao EP14 Notes: The AI Server Battlefield Moves from 'Enough Compute?' to 'How Do You Run Multi-Node?'
Personal notes after Dingmao Industry Notes EP14 (Computex 2026 observations): Agentic AI forces servers toward multi-node architecture; the bottleneck shifts from 'compute' to management chips (BMC), power delivery (800V), and interconnect; plus how 'the future of devices' changes under the AI wave — from a single machine to a whole rack as one computer, power and cooling as the ceiling, the management layer as the nervous system, and endpoint devices redefined. Educational observation, not investment advice; disclaimer at the end.
Should the Computer Sit in the Corner of the Office, or at the Center?
Notes after listening to Odd Lots interview Boris Cherny, who runs Claude Code at Anthropic. The part that stayed with me wasn't how capable the tool is. It was a study from the 1990s he brought up: companies bought the same computers, and only some of them got any productivity out of it. The difference was never the purchase.
The Walls of Helm's Deep: In a Crash, Your Safest Refuge Is Discipline, Not TSMC
When markets crack, everyone wants to retreat into the steadiest stock — the way Rohan retreated into Helm's Deep. But the walls of Helm's Deep were blown open, and TSMC just posted a blowout quarter and fell 6% anyway. A note on why there is no absolute safe haven, and why real defense is an action, not a place.
Who's Really Behind the 'Steady Index, Bleeding Stocks' Market? Notes on Gooaye EP680 — A 15% Drawdown Playbook, TSMC Mispriced, and a Misread AI 'Threat'
In the latest episode of Gooaye, host 'Chairman' lays out a read worth remembering: this wave of small-cap and high-beta carnage across Taiwan and US markets isn't driven by sector-specific bad news — it's spillover from a blow-up in highly leveraged Korean ETFs that has systematically compressed valuation multiples worldwide. TSMC's earnings were broadly excellent yet the stock got hammered — a flows problem, not a fundamentals one. And the Kimi K3 'AI threat' is a rerun of DeepSeek, best understood through Jevons Paradox. Full summary, stock-by-stock recap, plus our own take. A podcast listening note, educational, not investment advice.
Generative AI Certification — Complete Exam Cheat Sheet: 71 Key Terms + Chapter-by-Chapter Study Points
A last-hours review tool for Taiwan's III Generative AI certification: a full 71-term key-terms reference in five groups, plus the high-frequency points and 'see X, pick Y' distinctions across all four areas — fundamentals, prompting, applied skills, ethics and law. A companion to my prep-method post. Educational sharing, not a question bank.
Two AI Certs in Two Months: How I Used AI to Compress an Exam Down to What Actually Gets Tested
AI-900 in June, the III Generative AI certification in July. Here's my prep method: real past papers first, use AI to squeeze hundreds of practice questions into a single 'only-what-I-haven't-seen' review card, and judge on the spot which chapters to skip. Plus the topics this exam actually tests. A learning note.
The Last Short Seller on How AI Could 'Unwind' the S&P 500 — Notes on Meb Faber Show EP640 with Carson Block
Carson Block of Muddy Waters — the man Meb Faber calls 'the last short seller left in the world' — lays out an unsettling chain of logic on EP640: AI displaces high-paid knowledge workers, their 401(k) flows reverse, and the mega-caps that those flows inflated pull the S&P 500 down hard. Full summary, quotes, host and guest intros, and what we can take from it. A podcast listening note, educational.
How to Install Claude Code / Codex / Grok / Antigravity: A Beginner's Step-by-Step (Mac + Windows)
Never touched a terminal? You can still follow this. Step-by-step install for four desktop AI command-line tools (Claude Code, Codex, Grok Build, Antigravity agy) on Mac and Windows: how to open the terminal, which line to paste, how to log in and use it, and the mistakes beginners hit most. Pick the one you already pay for. Tutorial.
xAI Just Open-Sourced All of Grok Build: A Dissection Class on a Production AI Coding Agent
On 2026-07-15 xAI open-sourced Grok Build, its terminal AI coding agent — 840k lines of Rust, out in the open. What it is, why it's worth reading, how to install and use it, and the one thing we learned reading the source. A learning note; educational sharing.
Gooaye EP679: This Selloff Isn't Fundamentals — It's Margin Calls
Taiwan's biggest finance podcast nails the anatomy of the recent global shakeout: cascading deleveraging from over-margined Korean retail, not the Middle East, not inflation. Plus momentum's timeout, real shortages in substrates and OSAT, and the hallucination tax on AI tools.
A One-Person Engineering Department: Running Codex and Grok from Claude Code
I was paying for three AI subscriptions and using one. So I turned Claude Code into a chief engineer and made Codex and Grok my implementation lanes — five-part specs, fail-loud discipline, and the most expensive lesson of all: an AI's completion report is not completion.
AI Is Making Lawyers Busier: Jevons Paradox, Live on Odd Lots
Bloomberg's Odd Lots hosted Gary Wingens, chair of a major US law firm: due-diligence costs down 70%, yet more work than ever; billing rates up 10% while total client bills fall. A one-episode masterclass in whether AI destroys jobs — plus my own one-person-company confirmation.
How to Spot an Empty AI Money-Making Post — and Train Yourself to Think Deeper
The internet is full of 'office worker makes $2,000 a month with the latest AI' side-hustle marketing. This piece teaches you five fingerprints for spotting hollow content, a thirty-second wallet check, and six thinking habits you can practice daily — write down a dated prediction, explain it plainly, write your own answer first, ask how a claim could be falsified, think one step further, and cool off for 48 hours. Educational sharing, no accounts or platforms named, not investment advice, disclaimer at the end.
The Mirror Protocol: I Let an AI Read All 6,139 of My Commands. It Described Someone I Didn't Recognize.
A prompt circulating in the AI community: have an agent excavate every session log on your machine and answer 'who are you' using your own timestamps. Six phases, fully local, full prompt included for copy-paste. This is my run — including the part where my own records called my bluff.
Certainty for Sale: What My Colleague Is Really Buying from That Finance Channel
A colleague started following a large Taiwanese stock-tutorial channel, and I didn't read it as poor judgment on his part. What he's buying isn't financial knowledge—it's the feeling that the market can be understood, predicted, and controlled. This piece uses behavioral finance and media literacy to unpack the six psychological needs these channels target so precisely, the free-trust-to-paid-course marketing funnel, and three questions you can carry with you to test any of them. An educational reflection—no channel is named, not investment advice, disclaimer at the end.
Gooaye EP678 Notes: Meta Just Put Compute Pricing in the Sunlight — the Fog Behind Dream Multiples Is Lifting
Personal notes on Gooaye EP678: Meta repurposing old GPUs for inference and renting them out — refuting the 'AI server bad-debt' short thesis while making compute pricing transparent, which pressures dream-multiple valuations; rotation inside optical communications; the broad component price-hike wave. Educational notes, not investment advice; disclaimer at the end.