Article published Jun 22, 2026. Prices below use latest available snapshots.
Weekly recap · companion to Full-Scan Market Brief
Data note: Trading week is the four sessions through Thursday 2026-06-18. Friday 06-19 was Juneteenth (US markets closed); today (Mon 06-22) has no published bars yet (providers ~a session behind). All price/RSI figures are code-computed from the 06-18 EOW close (or cited from the 06-20 scans). Sentiment percentages are flagged as social claims, not price truth.
1. TL;DR
- A mild index pullback masked a violent rotation down the AI stack. SPY −1.0% / QQQ −0.7% on the week with every index still strong-up and above its SMA20 — but underneath, capital fled app-software and mega-cap ad and piled into the physical layer: wafer-fab equipment, memory, and AI-power. The tape didn't break; it rotated, hard.
- The market is re-pricing "AI gets monetized in SaaS" and paying for silicon + electrons instead. Application software de-rated into capitulation (CRM RSI 30, ADBE 28, NOW −9.4% / RSI 41) while toolmakers and memory ripped (BE +19.3% 7d, GEV +13.0%, MU +6.5% / +51% 30d, AMAT RSI 76). This is the workspace's ai-power-bottleneck / picks-and-shovels thesis paying off.
- The leader is not NVDA and not "semis broadly." NVDA went sideways (−0.6%, RSI 50); AVGO/AMD consolidated. The bid was specific: WFE (AMAT/KLAC/LRCX) + memory (MU/MRVL) + AI-power (BE/GEV/VST).
- SpaceX repricing landed — and the workspace called it. SpaceX dropped ~16% on the day / ~30% on the week (social claim) after a post-IPO notes offering priced below expectation, bleeding into the satellite complex (LUNR −12.2% 7d, ASTS −9.9%, RKLB −3.8%). Anticipated by the spacex-ipo-liquidity-event perspective.
2. The Tape
Indices took a shallow, orderly step back — every major average is still strong-up and above its SMA20, with the down-day concentrated in software and comms, not breadth.
| Index | 7d | RSI | Trend | Read |
|---|---|---|---|---|
| 🟡 SPY | −1.0% | 54 | strong-up | Above SMA20, trend intact |
| 🟡 QQQ | −0.7% | 59 | strong-up | Above SMA20, no breadth break |
| 🟡 DIA | −0.5% | 58 | strong-up | Dow the relative haven |
| 🟡 RSP (equal-wt) | −1.5% | 56 | strong-up | Breadth holding |
Sectors — the rotation is visible at the sector line, not the index line:
| Sector | 7d | Trend | Read |
|---|---|---|---|
| 🟢 XLI Industrials | +1.5% | strong-up | The week's relative leader |
| 🟢 XLF Financials | +0.1% | up | Steady |
| 🟡 XLK Tech | −0.7% | mixed | Held up only by semis — software dragged |
| 🔴 XLC Comms/Ad | −2.7% | strong-down · RSI 33 | The broken sector — mega-cap ad de-rate |
| 🔴 XLE Energy | −2.4% (−9.6% 30d) | weak-down | Oversupply/de-escalation collapse |
| 🟡 XLP / XLRE / XLY / XLU | soft | — | Defensives offered no bid |
The Tape in one line: indices ticked down ~1%, but XLC (RSI 33) and XLE were the only true breakdowns — everything else is a rotation inside an intact uptrend, not a top.
3. The Rotation — the week's whole story
The single insight: the market is de-rating the idea that AI gets monetized in application software, and rotating down the stack into the silicon and electrons that the buildout physically requires. App-software and mega-cap ad got sold; wafer-fab equipment, memory, and AI-power got bought. NVDA — the name everyone watches — sat it out in the middle.
| 🟢 IN — down-stack (silicon + electrons) | 7d | 30d | RSI | 🔴 OUT — app-software + mega-cap ad | 7d | RSI | |
|---|---|---|---|---|---|---|---|
| BE (AI-power) | +19.3% | +8.8% | 65 | NOW | −9.4% | 41 | |
| GEV (AI-power) | +13.0% | +3.7% | 66 | CRM | −8.2% | 30 | |
| MU (memory) | +6.5% | +51% | 66 | ADBE | −5.9% | 28 | |
| AMAT (WFE) | +5.4% | +36% | 76 | ORCL | −5.4% | 43 | |
| VST (AI-power) | +6.4% | −0.5% | 61 | MSFT | −5.1% | 35 | |
| KLAC (WFE) | +1.3% | +29% | 70 | PLTR | −5.1% | 41 | |
| LRCX (WFE) | 0.0% | +21% | 69 | IGV (software ETF) | −4.4% | 42 | |
| MRVL (memory/silicon) | +1.8% | +49% | 65 | META | −3.1% | 43 | |
| SMH / SOXX (semi ETFs) | +1.5 / +1.3% | +9.6 / +12% | 64 | XLC (comms/ad) | −2.7% | 33 | |
| GOOGL (−10% 30d) | −0.3% | 39 |
Consolidating, NOT leading — the leaders everyone defaults to went quiet: NVDA −0.6% / RSI 50, AVGO +4.3% but −2.5% 30d (weak-down) / RSI 51, AMD −2.5% / RSI 61. The bid stepped around the center of the chip complex, not through it.
The nuance that matters: this is not "semis broadly," and it is decidedly not NVDA. The buyer is paying for three precise things — wafer-fab equipment (AMAT/KLAC/LRCX), memory (MU/MRVL), and AI-power generation (BE/GEV/VST). The 06-20 scans corroborate the read independently:
- wfe-test-metrology — the entire WFE/metrology complex in a synchronized strong uptrend near 52wk highs; AMAT parabolic (RSI 76). The hottest sleeve on the desk.
- nvda-ecosystem — NVDA's supply chain parabolic while NVDA itself (RSI 50) and the platform layer (AMZN/PLTR) lag — the money is in the picks-and-shovels.
- monster-scan — breadth re-expanded 187 → 225, memory/storage + the semi capital-equipment / OSAT / test complex re-accelerating into the parabolic zone.
- cloud-etfs — the mirror image: all four cloud ETFs below SMA20, IGV (RSI 42) the laggard, software-heavy exposure reflecting the single-name SaaS wreck.
Tie to the perspective: ai-power-bottleneck (active · confirming) fired on three names this week — BE near its all-time high (+19.3% 7d), GEV breakout, VST/TLN firm. The thesis was always that AI capex routes to generation and silicon, not to the application layer that captures the headlines. The down-stack rotation is that thesis printing in price.
4. SpaceX & the space complex
The week's loudest single move was SpaceX, and the workspace had it framed in advance.
- The move (social claim): SpaceX fell ~16% on the day / ~30% on the week, with roughly $1T of market cap wiped, after a post-IPO notes offering (~$20B raise) priced below the ~$2T mark the market had penciled in.
- It was anticipated. The spacex-ipo-liquidity-event perspective (active) is built on exactly this mechanic — post-IPO repricing and lockup/liquidity drain. The prior assessment flagged the liquidity drain would land in the space halo, not the AI cohort; this week the halo took the hit on schedule, while AI-infra kept ripping (SMH +9.6% 30d).
- The satellite bleed. Damage radiated through the launch/satellite complex — LUNR −12.2% 7d (−34.5% 30d), ASTS −9.9% (−32.6% 30d), RKLB −3.8% (−25.1% 30d) — the epicenter of weekly damage outside software. SPCX dragged with it.
Read: contained, mechanical repricing of a single private-market event flowing into its public proxies — not a risk-off signal for the broad tape. The AI-infra overweight remains under no liquidity stress.
5. Word on the Street
Two crowds, one rotation — read from opposite ends. All figures in this section are social claims, not price truth.
X (29 live tweets · fast money): consensus is explicit — "rotation, not crash."
- Selling: mega-cap software and ad — GOOGL (−5/6%, cited "lost two top AI people"), AMZN (−4%), PLTR / CRM / ORCL weak.
- Buying: semis / AI-infra — NVDA green, MU +4–5%, MRVL +7%, SMCI +15%, AMD +2%.
- Structure tell: Dow green / Nasdaq red. The loud headline was SpaceX (~−16% day / ~−30% week / ~$1T wiped) bleeding into ASTS/SPCX. Bear tell cited: breadth — an SPX 7500 rejection.
Discord "Dumb Money" (value / social-arb crowd, Camillo-style): buying the rotation's victims.
- Conviction docs on beaten-down consumer-defensives / HALO — KMB, AMCR (~6.8% yield claimed), HRL — plus SaaS-as-value: INTU ("AI fears slashed SaaS = value"), FVRR ("priced like a dead company").
- Framing: "defensives abandoned while capital chases AI = the dislocation IS the opportunity."
- Still core-long AI/semis/power (top-3 longs claimed: BE, AMZN, SMH). Anthropic-IPO proxy interest in SKM (holds an Anthropic stake).
The synthesis: two sides of the same rotation. Fast money is chasing the down-stack momentum (buy silicon, sell software); the value crowd is fading the dislocation it creates (buy the abandoned defensives and the SaaS thrown out with the bathwater). One trades the rotation; the other trades its overshoot. Both agree it's a rotation, not a crash.
6. What we're watching next week
| Watch | Why it matters | Trigger |
|---|---|---|
| 🔍 Breadth / SPX 7500 | The lone bear tell from the X crowd; a rejection turns "rotation" into "distribution" | SPX rejects vs. reclaims 7500 |
| 🔍 Software RSI-sub-30 names | CRM (30), ADBE (28), NOW (41) — do they base or keep bleeding? | First higher-low / RSI turn up off the lows |
| 🔍 Leadership: broaden or hand back? | Does WFE/memory/power leadership broaden, or does NVDA reclaim it? | NVDA breaks > SMA20 or WFE digests and the bid widens |
| 🔍 SpaceX / satellite contagion | Whether the ~T+50 (early-Aug) insider unlock keeps the space halo heavy | ASTS/RKLB/LUNR stabilize vs. extend the slide |
| ⏳ 06-22 EOD repost | Today's bars aren't published yet — confirm the intraday social reads against the close | Re-pull 06-22 summaries once published |
7. Cited evidence
Scans (06-18 bar, dated 06-20):
- market-pulse — indices strong-up + vol collapsing; rotation into semis, out of energy/comms/metals
- monster-scan — breadth re-expanded 187→225; memory + semi-equipment own the top of the board
- nvda-ecosystem — picks-and-shovels skew extreme; equipment/power rip, NVDA flat at center
- wfe-test-metrology — the hottest sleeve; whole WFE complex synchronized melt-up, AMAT RSI 76
- cloud-etfs — the OUT-side mirror; IGV laggard (RSI 42), no oversold capitulation yet
Companion brief + SpaceX assessment:
- Full-Scan Market Brief — 2026-06-20 — the full-desk read this recap summarizes
- SpaceX IPO post-event assessment — 2026-06-20
Perspectives:
- ai-power-bottleneck (active · confirming) — the down-stack thesis; BE/GEV/VST/TLN firing
- spacex-ipo-liquidity-event (active) — anticipated the SpaceX post-IPO repricing + space-halo drain
- memory-supercycle (active · accelerating) — MU +51% / MRVL +49% 30d
Sentiment (social, not price truth): X live-tweet sweep (29 tweets, fast money) + Discord "Dumb Money" (value / social-arb crowd) capture, week ending 06-18.
Desk note, not hype. Index pullback shallow; the rotation underneath is the story. Hold the down-stack quality (extended — don't chase), watch the software floors and the breadth line, and confirm against the 06-22 close.
Sources
- Tape data (price/RSI/trend figures): desk pre-computed watchlist summaries as of the artifact date — *.