A note before the first note.
Read the era. Trade the thesis.
Two seats, two markets, one publication.
Two seats, one letter
Our combined experience spans global investment banks, Fortune 500 energy companies, and government — three vantage points from which the structural shifts of this era look most distinct, and which most independent research has access to only one of.
That blend puts the work at three intersections most independent research doesn’t bridge:
— Industry and capital markets. Operating reality from one side, capital-markets discipline from the other. Each perspective is incomplete on its own; most independent research carries just one. We carry both.
— East and West. We operate in both contexts every day. That is also why every flagship publishes in English and Chinese within 24 hours, with each version edited for the reader in front of it — not translated.
— Traditional energy, renewables, and AI. The three biggest macro narratives of the next decade, held in one frame. Each pulls demand and capital expenditure in different directions; reading them together is what makes the bigger picture cohere.
Neither of us will publish anything we’d be embarrassed for the other to read. That is most of what editing means here.
What’s broken
Three failures, in order of how often they show up.
Stories without mechanisms. “AI infrastructure is bigger than people think” is an observation, not an argument. The work of explaining why, through what channel, and with what second-order effects is what separates analysis from punditry. We will do that work on the page, every time.
Claims without horizons. A statement that can never be wrong can never be right. When we say something is about to happen, we will say when — and what data would tell us we were wrong.
Conviction without record. A public scoreboard isn’t a marketing exercise. It is the discipline that keeps an analyst honest. Every substantive call lives in a single timestamped page, with the outcome and the lesson. Including the ones we’d rather not remember.
Analysis that skips all three is, operationally, indistinguishable from punditry. We are not punditry.
What we owe you
Every flagship piece will carry:
One thesis you could explain in a paragraph: what’s changing, why, and through what channel.
The mechanism, not just the conclusion — the specific causal chain we believe is underway, broken into the stages where we expect to see it.
A horizon and an invalidation point. When this should play out. What data would tell us we were wrong.
Three scenarios — how it plays out if consensus is right, if our view is right, and the tail most people aren’t pricing. Magnitudes, not portfolio math.
The indicators we’ll watch — the two or three data points that will move first if the thesis is correct. Operating experience helps here: usually these aren’t the indicators the sell-side tracks.
A bilingual edition. English first, Chinese within 24 hours. Not a translation — a second telling.
A public scoreboard, updated quarterly. Every substantive call, what happened, and what we got right or wrong about the reasoning. No selective memory.
What we won’t do
No news-cycle chasing. If a piece would only be interesting this week, it shouldn’t be written.
No territory we can’t write honestly about. We will not cover companies, supply chains, or themes where either of us has a direct conflict. The list of things we can’t write about is part of the cost of doing this work honestly.
No pretending we’re the only voice worth reading. Where another analyst has done the work better, we’ll cite them and move on to where we can add something.
When East meets West, one language isn’t enough
There is a real gap between the two largest capital pools on earth. Chinese readers who want serious global thematic work read it through translation layers that flatten the texture. English readers who want a clear-eyed view of Chinese industry get consultants reading off charts, or expat columnists writing from Hong Kong. Neither catches the moment an early move on one side reshapes the other.
We can do both because between us, we operate in both contexts every day. Bilingual isn’t a feature — it’s the point.
What’s coming
Vega covers four lanes — and, more importantly, how they bend each other.
— AI. The technology layer (silicon, inference, agentic deployment), the labor consequences (Salesforce hiring zero, entry-level employment falling), and the macro feedback loop. Our first flagship piece is on the circulation ceiling — where the AI build-out runs out of customers.
— Energy. Traditional (oil, gas, grid power) and renewable (battery, solar, storage, transmission) read against each other, not as separate worlds. The AI lane runs on watts; the energy lane sets the speed limit on the AI lane.
— Macro. Labor markets, capital allocation, monetary policy, demographics. Big enough to set the cycle, specific enough that we can name the breaking points.
— Intersections. Every flagship piece lives at one. AI demand reshapes power and battery economics; the cost of the energy transition changes the trajectory of inflation; macro liquidity decides which AI capex actually survives. Most research treats these as three separate stories. We treat the connections between them as the actual story.
The AI Ceiling series
The first flagship — AI eats its own customers — laid out the math: AI’s growth has a built-in ceiling because automation shifts income from people who spend to people who don’t. Silicon Valley hits it first. The next five pieces extend that argument along five vectors:
1. The Macro vector. How the ceiling reshapes monetary policy and fiscal response. Why a Fed easing into a structural demand gap doesn’t fix it, and which kinds of redistribution actually push the ceiling out.
2. The Hardware vector. Why AI capex bends to the ceiling before the demand wave breaks. Training-to-inference shift, chip cycle compression, data-center power economics — and the indicators that mark when the wave has crested.
3. The Social vector. Labor markets, education, urban geography after the entry-level ladder collapses. What the canary city looks like when it stops singing, and which institutions reorganize fastest.
4. The U.S. AI outlook. How the ceiling plays out in the United States. Political coalitions, the redistribution window, regional asymmetries — what the next eighteen months actually look like for the place that hits the wall first.
5. The China AI outlook. How the ceiling plays out in China: lower starting labor share, state-as-spender cushion, the toC-toB penetration inversion, and where the internal cycle stage sits. Whether China hits its own version of the wall earlier or later than the U.S.
Everything is free for the first six months. After that, the deeper analytical work moves behind a paid tier; the free letter continues. We’d rather have a smaller paid base of readers who genuinely use the work than a larger one we have to keep convincing.
The compact
We will be wrong. The question is whether we are wrong in interesting ways we own publicly and learn from, or wrong in lazy ways we hide. Hold us to the first standard.
The first piece publishes next week.
—
Read the era.
Vega Thmatic
2026/6/7