Because the news tells you what happened. It rarely tells you what's holding.
AI is the largest build-out of our time, and it arrives as noise. Every week brings more filings, raises, benchmarks, and headlines than anyone can read against each other. Each one demands to be the thing that matters most. Most of it won't matter in a month.
The standard response is to chase the flood — to react to each release, each raise, each model that tops a leaderboard, and to call that coverage. It produces a great deal of motion and very little bearing. Read enough of it and you know everything that happened and almost nothing about where any of it is going.
The Standing Wave does the opposite. We hold a fixed set of hypotheses about how the build-out resolves — durable structural questions we track across years, not weeks. How does the cost of compute actually fall, and who captures the saving? Where does the physical layer bind? Which capabilities convert to durable demand, and which don't? These are the questions whose answers will define the decade, and they don't change with the news cycle.
Each week, we read the evidence against them. A filing, an earnings call, a capability jump — we ask not "is this exciting?" but "does this move one of our hypotheses, and in which direction?" The hypotheses give us a fixed vantage point. The evidence moves through; our reading updates; but the questions hold their shape.
The hypotheses are durable. The frames are temporary. We map the signals; the hypotheses don't follow them.
That last line is the whole discipline. A frame — the particular lens we use to connect the hypotheses in a given week — is allowed to be temporary, even disposable. But the underlying questions are chosen precisely because they outlast any single week's framing. When a new frame serves the evidence better, we change the frame. We don't change the questions to fit the news.
A hypothesis is not a prediction we defend. It's a structural question we hold open, with a current reading of where the evidence points — a reading that can strengthen, weaken, or cut both ways as the build-out unfolds. We track each one's status over time, so the publication accumulates a record: not just what we think today, but how our thinking has moved, and why.
When a hypothesis resolves — when the structural question is genuinely answered — we say so, and we say it plainly, including the times the evidence resolved against the reading we'd been carrying. A tracked record only has value if it's honest about where it was wrong. That accounting lives in the quarterly review.
We organise every hypothesis across four layers of the build-out, from the capital and compute at the base to where it meets the market. The layers are the spine: a reader following one thread — the economics of compute, say — can trace it across every issue.
Fourteen hypotheses are active across these four layers — the set runs to H15, with H13 never activated — each with a dated horizon and a countable trigger. See the full set →
The reading is only as good as what it's read against. We draw from across the full record — company filings and earnings disclosures, analyst coverage, primary reporting from the major financial and technology press, and the serious end of commentary — rather than reacting to whatever a single outlet ran that morning. Where a source materially affects a call, we attribute it in-text.
This is deliberate. Leaning on any one feed would mean importing its bias and its blind spots. Reading widely, and saying where the evidence came from, is what lets the hypotheses be tested rather than merely asserted.
Every issue carries two distinct instruments, and it's worth being clear that they are not the same thing. One names this week. The other tracks over time. Neither carries the other.
Our call on the single development that mattered most this week — declared clearly, with the rest of the issue earning the right to make it. It changes every week. It's a judgment, not a hedge.
The tracked record beneath the issues: where each durable question sits today, and how our reading has moved. It changes slowly, by design. It's the through-line, not the headline.
The Standing Wave is independent. It tracks the commercial and structural trajectory of AI through durable hypotheses, and it answers to its readers rather than to anyone in the space it covers.
It makes structural calls, not investment recommendations. The author invests in the AI sector, including companies the publication covers — and says so, openly, because the aim is to inform your own thinking, not to steer it. Knowing where the author has skin in the game is part of what lets you weigh the reasoning honestly and reach your own view. Nothing here is financial advice; the work is educational, and readers act on their own judgement.
The Tuesday Signal is, and will remain, free to read.
The method in practice — this week's read across the four layers.
Start with this week's Signal →