Selloff Toppicks Sector IC Memo

Executive Summary & Action Plan

Verdict

BUY (selective) — This week's crash re-ranked the map without breaking a single thesis. We re-priced all 49 covered names against inherited IWANNAVY fair values at Monday's US close and Tuesday's Seoul close: 23 now trade below fair value (13 of 24 on the 7/9 screen). The margin of safety still lives in de-rated applications software (ADBE +31%, CRM +28.5%) and contracted power — but the oil-shock-plus-hawkish-Fed regime downgrades every pre-revenue gap (OKLO, LUNR) to option-sized. Top Pick: Vistra (VST), BUY, FV $189 (+19.5%) — the one wide gap backed by signed contracts rather than narrative, with same-day catalysts (PJM 2028/29 capacity auction results post today after 4pm ET). Memory stays SELL even at crash prices; Tuesday's Seoul bounce was flows, not policy.

Loading chart

Price $158.12 | Market cap $53.3B | Target $189 | Upside +19.5% IWANNAVY Fair Value (log-median DCF · 10.5x FY27 organic EBITDA · EV/FCF, as-of 2026-07-10) · price as of 2026-07-13 US close · Street consensus $223 (reference) — we sit ~15% below Street because Street already capitalizes the unclosed Cogentrix acquisition and Meta nuclear-PPA upside we hold as bull-case optionality

Loading chart

Abstract

Three shocks collided in the July 7–14 window: Samsung's record-but-not-enough preliminary print set off a global memory de-rating that ended in Seoul's −8.95% circuit-breaker Monday (SK Hynix −15.4%, its worst session on record); Washington's reinstated blockade of Iranian shipping repriced Brent from ~$76 to ~$84 in a day and revived the oil–CPI–Fed-hike loop; and inside US tech, crowded AI-hardware positioning rotated violently into 2026's software laggards (memory ETF −9%, Oracle −6.5% versus Salesforce +4.8%). Re-pricing all 49 covered names against inherited IWANNAVY fair values shows 23 below fair value, concentrated in two quality pockets: applications software (Adobe +31%, Salesforce +28.5%) and power (Vistra +19.5%), with space/defense contract compounders next (Kratos +10.7%). The "higher-for-longer-plus" rate regime staleness-flags every pre-revenue gap — we would haircut OKLO/LUNR-type fair values 10–20% and size them as options. Top Pick: Vistra (VST) — contracted PJM capacity revenue, hedged exposure to the very gas-price shock hurting everything else, and the lowest thesis risk in coverage (p_thesis_wrong 21%). Memory remains SELL; AI semis are back at fair value, not yet cheap.

Action Plan

Own Vistra (VST) as the regime-aligned core position; pair with the app-layer value pair Adobe (ADBE)/Salesforce (CRM); add Kratos (KTOS) on ignored contract momentum; keep memory, data, and pre-revenue duration (OKLO, LUNR) on the do-not-chase list until the 7/14–7/30 verdict window clears.

Loading chart

Risk/Reward (VST) ~2.0:1 (bull $255 vs bear $110) | Prob-weighted 12M return +24.0% (Bull 31% × +61.3% + Base 52% × +19.5% + Bear 17% × −30.4%) | Confidence: High (VST) · Medium-High (ADBE/CRM) · Medium (KTOS)

The full post-crash ranking below is the report's spine: every covered name's upside to its inherited IWANNAVY fair value at the fresh closes, with as-of dates per fair value in the appendix.

Loading chart

What Happened / State of the Sector

The selloff ran in three acts. Act 1 (7/7): Samsung Electronics (005930.KS) printed the largest quarterly operating profit in its history (~₩89.4T preliminary) and fell ~7% — revenue missed, a ~₩400T capex plan spooked supply discipline, and the memory "peak-out" debate went global (CNBC, 7/7). Act 2 (7/11–13): tanker attacks resumed in the Strait of Hormuz, CENTCOM ran weekend strikes, and Monday Trump reinstated the Iranian-port blockade with a 20% transit fee (enforcement 7/14 4pm ET) — Brent jumped ~10% to the $83–84 area (Bloomberg, 7/13; the IMO rejected the fee's legality, so markets priced risk premium, not a durable toll). Act 3 (7/13): Korea Investment & Securities cut SK Hynix's (000660.KS) Q2 estimate ~8% below consensus and pushed full-scale HBM4 sales to Q3'26; Seoul crashed −8.95% through a circuit breaker (SK Hynix −15.4%, ~$200B erased) and the contagion split the US tape — memory ETF −9%, SOXX −2%, while funds rotated into 2026's software laggards (CRM +4.8%, IGV −12.5% YTD entering the day). JPMorgan called the semis damage "crowded positions, not the beginning of a reversal of the AI upstream cycle" (CNBC market wrap, 7/13). Tuesday's Seoul deep-V rebound (KOSPI +0.73%, Samsung +3.3%, SK Hynix +3.7%) was institutions (~₩2T) and foreigners (~₩0.8T) absorbing ~₩2.8T of retail and margin-call selling before any stabilization package — the F4 policy meeting was still pending (7/16) at the close (Seoul Economic Daily, 7/14) — so we treat it as a dead-cat bounce, not a re-entry signal.

GAP verification (±5% single-session moves, all news-checked): ORCL −6.5% and CRWV −6.3% were OpenAI-counterparty stress — Apple's trade-secret lawsuit against OpenAI (filed 7/10) compounded S&P's BBB− cut and its warning that ~half of Oracle's RPO rests on OpenAI's ability to pay (Motley Fool, 7/13). OKLO −6.2%, LUNR −6.1%, RDW −5.8%, RKLB −5.3% carried no company news — pure de-risking of space/SMR beta amplified by China's Long March booster-recovery milestone, which prompted Bernstein to call China "SpaceX's top rival" (24/7 Wall St, 7/13). FIG +12.0% was a short squeeze (~42% of float short) ignited by Buy initiations arguing Figma wins from AI (Motley Fool, 7/13). Kratos (KTOS) fell −2.6% despite announcing a ~$100M sole-source space-domain-awareness prime award the same day (GlobeNewswire, 7/13) — fundamental news the tape ignored, which is what a positioning purge looks like.

The sector-by-sector verdicts that fall out of the re-rank, with the pick (or the honest absence of one) per sector:

Sector	Stance	Pick	Why (post-crash)
Power / Energy	OW	Vistra (VST) — BUY	Contracted, not narrative: 10,566 MW cleared in the record 27/28 PJM auction; 28/29 results post today. Oil shock is margin-neutral-to-positive — US power prices key off Henry Hub (~$3, barely moved), and a hedged net-long generator monetizes strength
Applications	OW	Adobe (ADBE) — BUY (Salesforce (CRM) co-pick)	Monday's rotation confirms the de-rated app layer; ADBE takes the slot on receipts — 6/11 beat-and-raise, Firefly ARR ~$300M (+~50% QoQ), Topaz Labs acquisition — vs CRM's still-unproven Agentforce conversion (Guggenheim upgrade 7/1 marks the sentiment floor)
Space / Defense	Neutral-OW	Kratos (KTOS) — BUY (starter)	Two July sole-source awards (~$36M 7/2, ~$100M 7/13) into a falling tape; Hormuz escalation raises unmanned/air-defense funding urgency; LUNR (+32%) only as an option-sized satellite position
AI Semis	Neutral	none — NVIDIA (NVDA) HOLD	Back at fair value (+3.2%) after −3.5% Monday; demand intact (SemiAnalysis sees H2 DC revenue ~20% above consensus) but no cushion against contagion
Security	Neutral	none — Zscaler (ZS) HOLD-acc	+5.1% gap into a confirmed FY27 growth reset (16–17% vs 24%) is dead-money risk; PANW/CRWD remain 40%+ above FV
Cloud / Hyperscale	Neutral	none	MSFT/AMZN at fair value; ORCL and CRWV got cheaper but are OpenAI-counterparty risk, not value
Data	UW	none	SNOW/DDOG/PLTR still 19–25% above FV — and they rallied Monday
Memory	UW	none — SELL stands	Crash removed froth, not overvaluation: KR complex still 43–71% above peak-margin-normalized FV; MU −9% vs FV
Quantum	UW (watch)	none	Washed to fair value for the first time (IONQ +18% vs FV after −9.3% Monday); option value only

What to Watch Now

Today (7/14): June CPI 8:30am ET (consensus −0.2% m/m, +3.8% y/y — pre-blockade data; July re-accelerates mechanically) + JPM/BAC/C/WFC/GS Q2 prints + Fed Chair Kevin Warsh's first congressional testimony + blockade enforcement from 4pm ET + PJM 2028/29 capacity auction results after 4pm — the direct VST/CEG repricing event (PJM Inside Lines).
7/16 — TSMC (TSM) Q2: CoWoS capacity, 2nm ramp, any capex raise — the first credible arbiter of whether the memory scare is an AI-cycle signal or Korea-specific.
7/22–7/30 — the verdict gauntlet: GOOGL (7/22), FOMC (7/28–29, ~33% hike odds priced), MSFT+META (7/29), AMZN (7/30); SK Hynix Q2 (7/22 E, some calendars 7/29) tests the HBM4-slip narrative against a washed-out tape.
7/23 — First Solar (FSLR) Q2 (guide: 3.4–4.0 GW sold, adj EBITDA $400–500M), then the early-August (E) Commerce Section 232 polysilicon ruling — Wells Fargo's $320 case rests on it (Stocktwits, 7/6).
August float events: SPCX first lockup tranche after Q2 (8/6 E); FIG extended lockup Aug 17–31 with ~42% of float short — squeeze mechanics cut both ways.

What to Watch Next (Technology)

HBM4 ramp → 2027 LTA repricing: executed long-term agreements are what separate a 2027 memory plateau from a classic rollover; qualification news flows through H2'26 (7/22 and 7/30 prints are the first tests).
Custom silicon vs merchant GPUs: Meta's Broadcom-designed Iris accelerator enters TSMC production in September — the share of accelerator capex bypassing NVIDIA sets 2027 fleet economics.
Agent monetization receipts: Agentforce ARR ($1.2B, +205% at the 5/27 print; next disclosure 8/26) and Adobe's Firefly ARR (~$300M) decide whether the app-layer re-rating has a second leg beyond positioning.
Power regulation as the VST/CEG swing: FERC and state backlash against record PJM capacity prices and data-center co-location is the one variable that clips the contracted-windfall thesis.
The SMR criticality race: Oklo missed DOE's July 4 reactor-pilot criticality deadline while four peers hit it (Energies Media, July 2026) — milestone pace, not narrative, now prices the pre-revenue nuclear cohort.

Fair Value Notes

All fair values inherited from same-framework IWANNAVY reports (log-median of DCF / EV/EBITDA / EV/FCF for cash generators; probability-weighted forward multiples or rNPV for pre-profit names), as-of dates disclosed per name; full audit trails in the source reports. One line each for the picks and contested names:

Vistra (VST) $189 (07-10) — 10.5x FY27 organic adj EBITDA $7.6B, log-median vs FCFE-DCF $170 and 6% FCF-yield $217; excludes Cogentrix close and Meta ~2.6GW nuclear PPA (both upside). Thesis breaker: FERC/state caps on PJM capacity prices or co-location. Scenario probs 17/52/31 (bear/base/bull) — bull-skewed because the demand shortfall is contracted, not forecast.
Adobe (ADBE) $302 (07-12) — cheapest quality multiples in coverage (FCF yield ~10%), bear DCF $167 assumes terminal decline. Thesis breaker: net-new ARR miss on 9/10 or Firefly stalling. Probs 24/49/27. Street violently split (BofA Sell $190 vs HSBC Buy $308) — we are between, above both spot anchors.
Salesforce (CRM) $220 (07-12) — reverse-DCF at $171 still implies sub-inflation perpetual growth; co-pick, weaker near-term receipts than ADBE (probs 20/45/35, p_wrong 36 vs ADBE 33).
Kratos (KTOS) $52 (07-10) — 5.5x forward EV/Sales on a record ~$2.0B backlog (~72% funded); July's ~$136M of sole-source awards are upside to the revenue path. Thesis breaker: fixed-price margin slippage or a continuing resolution.
Pre-revenue cohort (OKLO $60, LUNR $20, RDW $12.10, VOYG $34.70, all 07-10) — gaps of +17% to +32% are stated at face value but the hawkish-Fed oil shock stales these discount-rate assumptions; risk-adjusted, haircut 10–20% and size as options. OKLO is the single most fragile name in coverage if the selloff extends (~$1B ATM into a falling price = reflexive dilution).
Memory (000660.KS ₩1.06M, 005930.KS ₩150K, 042700.KS ₩61K, all 07-13; MU $850, 07-04) — peak-margin normalization: even post-crash, the complex prices 71.5% operating margins as permanent. SELL stands; re-entry belongs at normalized fair value, not at flow-driven bounces.
No-pick zones: NVDA $210 (07-10) at fair value; MSFT $399 / AMZN $246 at fair value; ORCL $113 — the −6.5% Monday move validates the counterparty-risk case, still 14% above FV; security ex-ZS and data cohorts remain far above fair value.

Catalysts & Risks

Catalyst	Timing	Impact	·	Risk	Prob	Impact
PJM 2028/29 auction results (third record clear?)	7/14 post-4pm	High (VST, CEG)	·	July FOMC hike (~33% priced) de-rates all duration	Med	High
June CPI + Warsh testimony + bank prints	7/14	High	·	Hormuz military escalation; oil through $90	Med	High
TSMC Q2: CoWoS/capex read on AI cycle	7/16	High	·	Hyperscaler capex hedge at 7/22–30 prints ripples down-chain	Low-Med	High
Hyperscaler gauntlet + SK Hynix Q2 + FOMC	7/22–7/30	High	·	Second Seoul margin-call cascade (~₩38T credit unwinding)	Med	Med
KTOS Q2 (8/6 E) + defense award cadence	Aug	Med	·	OpenAI counterparty stress (Apple suit) hits ORCL/CRWV marks	Med	Med
FSLR Section 232 polysilicon ruling	early Aug (E)	Med (binary)	·	Pre-revenue dilution reflexivity (OKLO ATM, LUNR mission risk)	High	Med

References

Samsung preliminary Q2 (CNBC, 2026-07-07) · SK Hynix worst session on record (TechTimes, 2026-07-13) · Seoul deep-V rebound (Seoul Economic Daily, 2026-07-14)
Trump reinstates Hormuz blockade, 20% fee (CNBC, 2026-07-13) · Oil holds jump (Bloomberg, 2026-07-14) · Are oil stocks worth buying now? (Motley Fool, 2026-07-13) · US-Iran strikes and oil (Al Jazeera, 2026-07-13)
Monday US market wrap (CNBC, 2026-07-13) · Tuesday setup: CPI + banks + Warsh (Yahoo Finance, 2026-07-14)
Why Oracle slumped (Motley Fool, 2026-07-13) · Why Figma jumped 11.9% (Motley Fool, 2026-07-13) · Space complex selloff on China booster milestone (24/7 Wall St, 2026-07-13)
Kratos $100M sole-source SDA award (GlobeNewswire, 2026-07-13) · PJM 2028/29 auction (PJM Inside Lines) · Guggenheim upgrades Salesforce (Yahoo Finance, 2026-07-01) · Oklo and the July 4 criticality deadline (Energies Media, July 2026)
Companion IWANNAVY reports (fair-value sources): AI Value Chain sector map (2026-07-12) · AI/Quantum/Energy/Space FV screen (2026-07-10) · Frontier gap screen (2026-07-09) · Korea AI Memory re-underwrite (2026-07-13) · Cerebras tearsheet (2026-07-11) · Specialist notes: selloff scout notes