Yesterday's tape was broad and red. Today's was narrow and green. The inflation print is the hinge between them.

No pre-market or mid-day edition was published today, so this close is graded against last night's close — the Aug 11 edition. Three of its four calls resolved in its favor. This desk will be equally specific about the one that did not.

1 · "CPI is the whole overnight" · heldLast night's edition argued the selling was concentrated at the long-duration end of the board specifically because a September meeting reported as a coin flip sat eight hours behind an inflation print — and it named the cohort: AMZN, AVGO, CSCO, NET, CRWD. The print came in line and that exact cohort turned. CSCO −1.75% → +2.86%, NET −1.20% → +1.47%, CRWD −1.45% → −0.05%, AVGO −1.50% → −0.01%. Four of five turned or stopped bleeding. AMZN −2.09% → −1.83% did not, and that is carried forward below as its own question rather than folded into a cohort win.

2 · The on-deck call · direct hitLast night's board flagged SMCI in the after-close slot, reporting $1.70 against a $0.92 consensus. Today SMCI closed +19.02%, the single best name on the board. GuruFocus tied the move to a $72 billion outlook. Putting the beat on the board the night before is precisely what the on-deck section exists for.

3 · The NVDA positioning tell · reversed inside 24 hoursLast night this desk flagged NVDA closing −0.02% on overwhelmingly bullish news flow and wrote that when a name stops rewarding good news, the news is in the price. Today NVDA closed +3.03%. One in-line inflation print was enough to un-stick it. The observation was accurate as a description of Tuesday and it was not durable for a day. Positioning tells describe a state, not a schedule — and this publication records that as written.

4 · "Watch the decoupling" · resolved, but not as framedLast night's watch-item was that airlines had rallied into rising crude, and it asked whether that would hold or snap back. Neither happened cleanly. The entire crude pairing simply went quiet on both sides: UAL −0.94% and DAL −0.53% gave back the bounce while SLB −1.99%, CVX −0.05% and XOM −0.03% stalled. The honest read is that we posed a two-sided question when the actual answer was that the input stopped mattering. Crude did not resolve the trade. CPI and AI earnings took the wheel away from it.

Withheld from this editionOur screener again returned +9.14% on a high-momentum name at an unchanged price — byte-identical to the value it has served across multiple prior sessions, including yesterday's, and still unsupported by any dated news flow on the name. This is at least the fourth consecutive appearance of the same figure. It is omitted from the leaderboard again. Every leader shown above survived that check.

One scheduled event dominated the session and one earnings theme paid for it. Every item below is attributed to a named outlet.

Central banks & rates · pressure released, not removedCNBC reported July CPI rose 0.1% on the month for an annual rate of 3.4%, core up 0.2%, all in line with Dow Jones consensus and each down 0.1 percentage point from June (CNBC · Aug 12). NBC News characterized the report as inflation remaining stubborn (NBC News · Aug 12), while Quartz reported inflation cooled to 3.4% (Quartz · Aug 12). The underlying release is published by the Bureau of Labor Statistics (BLS). Reporting put September rate-hike pricing near 44%, down from the even split cited a day earlier, with Treasury yields pulling back broadly. The thread: note the two framings side by side — "cooled" and "remained stubborn" describe the same 3.4%, still well above the Federal Reserve's 2% target. What the tape responded to was not disinflation; it was the removal of an upside surprise. That is enough to un-freeze positioning without settling anything, which is why the long-duration cohort recovered while the index moved only 0.3%. Those are reported figures and market-implied probabilities, not this desk's projections.

Corporate · the AI earnings supercycleThe Wall Street Journal reported Aug 12 that Cisco posted a sharp jump in profit as AI orders roll in, and Zacks reported the company beat Q4 earnings and revenue estimates. MarketWatch reported that Cisco saw record results from an AI "supercycle" but that its stock pulled back. GuruFocus reported Super Micro jumped on a $72 billion outlook, and CNBC carried commentary describing the AI data-center trade as back (The Wall Street Journal · Zacks · MarketWatch · GuruFocus · CNBC · Aug 12). The thread: this is what actually paid today. SMCI led the board outright and CSCO closed +2.86% going into its own after-close print. Read the Cisco pair carefully, because it is the more instructive one — record results and a reported after-hours pullback in the same name. A beat the stock does not reward is the same shape this desk flagged in NVDA on Tuesday, arriving one session later in a different ticker.

Supply chains · memory has its own clock24/7 Wall St. reported Aug 12 that memory stocks rallied with SK Hynix, SanDisk and Micron all jumping, and separately that SK Hynix and SanDisk climbed 8% and Western Digital gained 4% as the memory shortage deepens. Invezz reported the same day on Micron and SK Hynix rallying following CoreWeave and Super Micro earnings, and Zacks reported record margins and upward revisions in its Q2 earnings scorecard (24/7 Wall St. · Invezz · Zacks · Aug 12). The thread: MU +4.92% and WOLF +5.21% did not move on the inflation print. They moved on a supply constraint, and a constraint has a longer half-life than a data release. This is the second consecutive session in which reporting has pointed at a physical bottleneck rather than at demand — Tuesday it was power infrastructure, today it is memory supply. Where the bottleneck sits is where the pricing power sits.

Defense & rearmament · still running underneathZacks reported Aug 12 on defense ETFs as the Pentagon pushes to boost missile stockpiles, and ETF Trends reported the same day on United States missile shortages driving defense ETF flows — following The Motley Fool's Aug 11 report on the Army's $54 billion Patriot purchase from Lockheed Martin (Zacks · ETF Trends · Aug 12; The Motley Fool · Aug 11). Our sentiment read on LMT is 14 positive items against 1 negative. The thread: the same thread flagged last night, one day further along, now showing up as a stated procurement shortage rather than a single contract. It did not lead today. It is the slow clock running under a session dominated by a fast one.

The calendar risk is behind youThat changes what you are exposed to, not whether you are exposed. With CPI printed, there is no comparable scheduled macro event before tomorrow's close. That removes the stated reason for the de-risking that dominated Tuesday and hands the tape back to single-name earnings. Positioning that was defensive into the print is now positioning without a stated reason.

Cisco is live after hoursThe name closed +2.86% on the session and then reported. Per the Wall Street Journal and Zacks it beat; per MarketWatch the stock pulled back on record results. Whatever gap it opens is information about how much of the AI order book was already priced — the same question NVDA raised on Tuesday and did not answer.

AMZN is the standing anomalyTwo consecutive sessions of underperformance, −2.09% then −1.83%, through both a red tape and a green one, without joining the cohort recovery the rest of its group received. One name diverging from its own cohort across opposite market days is worth a look rather than an assumption.

Estimates shown are consensus figures from our earnings calendar, not this desk's projections. AMAT is the one that touches today's leadership directly — semiconductor capital equipment is the upstream read on whether the AI-infrastructure orders that paid SMCI and CSCO are converting into actual tool spend.

Overnight risk vetoA 66%-advancing board that closed higher than a 72%-advancing board did the day before is a description of concentration, not of breadth. One cohort carried this session, and the scheduled event that had been suppressing risk appetite is now spent. Neither of those is a forecast. Both describe an exposure that is narrower tonight than the index print suggests. Define the level that says you are wrong before the open, not after.