As of Jul 30, 2026 (ET), after the latest data pull · views from the news, not from live prices
A real multi-year memory shortage, now confirmed by Apple from the buyer’s side — and the Korea leverage crash that hit the stock has already reversed.
High rates and the AI-spending worry are both pushing down and neither is resolving; slower growth removes the cushion.
Changes if: the 30-year yield drops back under ~5%, or a big spender shows the AI spend paying off.
Down (yields up). The Fed is comfortable with rates high.
Read in order, the Korea selling was leverage unwinding and it is about 90% done — Korean tech has since reversed hard (Samsung +18%). On top of that, the multi-year story is a real shortage: Samsung locking 70% of capacity into five-year contracts through 2028, and Apple now raising prices and cutting margins because memory costs are spiking. The near-term bounce and the structural shortage point the same way: up.
Note: The model’s own weekly read on MU is stale — it still counts the heavily-covered Korea crash, not the newer reversal. Read the sequence, not the tally.
Demand is clearly real (Azure, AWS, $565B of capex prove it), but the stock can fall anyway: lenders are nervous — record cost to insure its debt on the circular OpenAI financing — and China threatens the long-run demand case.
Microsoft gets paid because its AI spending already shows up as cloud revenue; Meta and Google are being punished for spending without the payoff yet.
It beat but guided down, and its margins are getting squeezed by memory costs. The one silver lining: that confirms the memory shortage — which helps Micron.
OPEC+ paused output hikes and Middle East supply keeps getting disrupted.
The $53.86B Patriot award and live missile demand aren’t tied to the AI selloff.
The $1.8 trillion buildout needs electricity. These win as long as the data centers get built, even if the AI-model economics are questioned.
Momentum views on fresh catalysts — the edge is recency, not discovery. Treat as good while fresh.
The Hormuz disruption is still on (oil exports down about 80% from the June peak). War-risk premiums jumped and cargoes are rerouting to Suez and around Africa — rerouting and war premiums raise tanker day-rates.
Valero beat with its highest quarterly profit since 2022 as Hormuz cut supply; diesel futures topped $176. The disruption widens refining margins — a cleaner long than crude itself.
American Airlines cut its 2026 earnings guide to zero as fuel costs rise. Fuel is the direct hit; the marginal carrier is the sharpest loser.
A repeated safe-haven bid across both the war and the tariff stories. Holds as long as rates, war, and trade risk are all live at once.
Down on import costs from the 10%–12.5% tariffs on 60 trading partners.
Down on tariff pass-through to consumer prices.
The tariffed markets themselves — Canada, Brazil, China — down.
The largest-ever PAC-3 order lifts the whole missile and prime complex, not just Lockheed. Same clean demand as the LMT view, unrelated to the AI selloff.
The FDA advisory panel vote to recommend broader pharmacy compounding helps the telehealth/compounding sellers.
As of: 2026-07-30 (ET), after the latest data pull. Built from the news and the market-impact model. Numbers are what the source stories reported; these are views from the news, not from live prices.
Two things are driving the market at once: interest rates are high and staying high, and investors are now judging big tech by whether its huge AI spending already makes money. The result is a broad selloff plus a split. Companies whose AI spending shows up as revenue are going up; companies that are only spending are going down.
Read this section in time order. The structural story is a real memory shortage. On top of it, Korea had a leverage-driven crash earlier in the week, and that crash has since reversed. The crash got most of the coverage because it played out over three days; the reversal is newer and thinly covered, so don't let the volume of crash stories stand in for where things are now.
So the current read on chips is a hard bounce off a leverage-driven bottom, not an ongoing crash, sitting on top of a multi-year shortage that is only getting more confirmed. Whether the rally continues from here is open, but the direction of the last two days is up. (The market-impact model still reads this week's chip stories as down because it ranks by how much coverage a story got, which favors the older, heavily-covered crash over the newer bounce. Read the sequence, not the tally.)
These are momentum views on fresh, specific catalysts from the last week to ten days. The market has largely digested each one, so the edge is that the move is recent and the driver is still live, not that it's undiscovered. If the catalyst ages out or reverses (as the oil price briefly did on a US-Iran framework before Hormuz stayed choked), the view weakens — treat these as good while fresh.
A note on what's already in the price. Where a view rests on a story that itself reports the move ("shares fell N% on weak guidance", "stock down on the miss"), the initial drop is already digested and in the price. That kind of view is still useful, but only as a bet that the driver keeps pushing (continuation), not on the fall you already read about. It's a weaker, second-hand read than a fresh catalyst that hasn't fully played out, and different again from a milestone story (a price level with no driver, which is worth nothing forward). Below, the big-tech earnings reactions (Apple, Meta, Google already fell on their prints) and the airline call are digested-move views; the memory shortage, the Korea reversal, and the oil chain are still developing.
Three calls: best structural long is Micron (real multi-year memory shortage, now confirmed by Apple, and the Korea leverage crash that hit it has already reversed); best short is long bonds/TLT (Fed wants rates high); best risk/reward is the AI-power basket (you get the buildout without betting on whether the models pay off).
Method: figures are the numbers stated in the source stories; the stock views draw on the market-impact model's per-week reads (currently through the week of Jul 27–30), corrected for sequence where the model's coverage-weighted ranking lags a newer update. Views are labeled as such and are not a price forecast. A live price feed exists but returns real-world quotes that don't map to this dataset, so it isn't used here.