Daily Brief Archive: July 18, 2026 — U.S. & Hong Kong Stock Analysis

Free preview

U.S. Markets

  • S&P -1.01% to 7,457; the SOX fell ~9% this week into a technical bear market
  • AI capex under fire: semis have shed over $1T of market cap since July began
  • The Kimi K3 chip demo rattled the compute narrative — SNPS -8%, CDNS -9% in a day
  • Earnings split the tape: NFLX -7% on a guide-down, TRV +9% to record highs on a blowout
  • US-Iran escalation drove WTI past $79 (+11% w/w) while a hawkish Warsh capped rate-cut hopes
  • 10Y steady at 4.55%, VIX at 18.0; defensives, energy and insurers took the baton

Hong Kong

  • HSI -1.78% below 25,000; Hang Seng Tech -4.37%, its worst day in 15 months
  • The HSI still gained 1.6% on the week — Friday alone gave it back as crowded AI positions unwound
  • SMIC -10%, Hua Hong -12%, Zhipu -28%: chips and AI-model names took the double blow
  • HS Biotech -6.33% with Akeso -13%, while power and high-yield names rallied against the tape
  • Southbound sold a net HK$1.46B Friday yet bottom-fished via broad ETFs; July inflows top HK$80B
  • H1 GDP +4.7% (Q2 +4.3%); with the Politburo meeting near, stimulus expectations are warming

Today's Watchlist

  • Book cleanup: three stopped out, four closed proactively, zero new positions
  • Monday 7/20: Starship relaunch window, China LPR fixing, US June LEI
  • Earnings night 7/22: Tesla (±7.6% implied), Alphabet, Texas Instruments
  • Intel on 7/23 and flash PMIs on 7/24 — back-to-back tests of the AI-capex story
  • The 7/27 pile-up: K3 weights release, CDNS and APLD earnings, CXMT's Shanghai debut
  • FOMC 7/28-29 under a hawkish Warsh; CAPR's adcomm 7/29 — our short is already banked

Deep Dives (4 names)

Neutral
Score7/10RSI(14) 30.7Rel volume 3.3xQ3 guide $128.6 亿 vs 预期 ~$130 亿To 52w low 5.6%

Technicals — A 3.3x-volume decline with the 52-week low at $65.08 just 5.6% away says downside momentum hasn't exhausted; RSI 30.7 kisses oversold, but yesterday's rule stands — only a three-day, shrinking-volume reclaim of $69-70 counts as first evidence of seller fatigue, and below $69 the map reads $65

Fundamentals — Q2 revenue of $12.56B (+13.4%) and $0.80 EPS edged past estimates, but growth decelerated from Q1's 16% and operating margin slipped YoY to 33.4%. The ad tier tops 250M MAUs with the $3B ad-revenue-doubling target reaffirmed, and the stock now trades near 24x normalized earnings — cheapest since 2022. Cheap is a fact; an inflection is not

News — The 7/16 print delivered a second consecutive guide-down (Q3 revenue $12.86B vs ~$13.0B, EPS $0.82 vs $0.84) with the full-year band narrowed to $51.0-51.4B; halving engagement disclosure to once a year read as a transparency retreat. At least 17 brokers cut targets the next day, with analysts warning the company is 'losing narrative control'

Short-term · 1-3 weeks
Neutral Sidelines

De-ratings through a growth downshift have no precise floor; 24x only matters alongside a stabilizing tape. The summer Upfront ad close is the next tradeable positive catalyst

Entry No catching the second knife: a three-day shrinking-volume reclaim of $70 is the first right-side evidence, while a high-volume break of the $65.08 52-week low voids every dip-buying fantasyStop —Target —
Long-term · months+
Neutral

Ad-revenue doubling and live sports (Women's World Cup / NFL / WWE) form a real second curve, but falling per-member watch time plus the disclosure retreat mean the growth story must re-prove itself

  • With engagement disclosure reduced, bad news gets confirmed later and at a higher price
  • YouTube's screen-time squeeze and media-M&A reshaping the competitive map
  • Lagged churn from the second US price hike
Community Voices
  • WSB · Mentions halved from 380 to 123 yet aggregate sentiment stays firmly bullish — retail is buying the dip, and the dip-buyers are waiting for the next wave of dip-buyersOriginal ↗
Institutional Views
  • 华尔街共识 · 57 analysts average $102.02, +48% from here, still Buy-skewed with zero Sells — but targets are being slashed in real time, making consensus a moving target
  • JPMorgan · Kept Overweight but slashed the target from $118 to $85Source ↗
  • Barclays · Equal Weight, target $85 to $80, saying Netflix is losing narrative controlSource ↗
  • UBS · Buy maintained, target $130 to $115Source ↗
Neutral
Score7/10Day -14.15% / 量比 3.6xQ2 EPS $2.80 vs 预期 $2.51FY procedures guide 13.5%-15.5%（维持）RSI(14) 30.2

Technicals — A 3.6x-volume marubozu closed at $345.42, a quarter-percent above the $344.55 52-week low; below that there is air to the $320 shelf. The 50-day sits 17% overhead at $417, and while RSI touched 30.2, oversold the day after a crash carries no statistical weight

Fundamentals — Q2 beat on both lines — $2.89B revenue (+19%), $2.80 non-GAAP EPS — with 468 da Vinci placements (246 of them da Vinci 5), Ion procedures +36%, and $8.63B in cash. The trouble is all in growth mix: US procedure growth slowed to ~12% (weakest in years), bariatric volumes fell high-single-digits, and management flagged China competition and pricing pressure by name

News — The 7/16 print beat, but holding the FY da Vinci procedure guide at 13.5%-15.5% 'near the midpoint' translates to a cut in growth-stock pricing language. ACA subsidy expiry deferring insured elective surgeries and GLP-1 pressure on bariatrics are the two structural headwinds, with April's Class II E-brake recall (454 systems) still weighing

Short-term · 1-3 weeks
Neutral Sidelines

De-ratings born of 'beat but no raise' typically need 2-3 quarters of data to appeal the verdict; the only short-term trade here is an oversold bounce, sized small with an iron stop

Entry The $344.55 52-week low is the last near support — it must hold with no new low for three sessions before any bounce qualifies; below a reclaim of the earnings gap there is no right-side signalStop —Target —
Long-term · months+
Neutral

The robotic-surgery penetration thesis stands (installed base +12%, Ion +36%, ongoing buybacks), but the US growth downshift and GLP-1/insurance headwinds need digesting — the phase where estimates get cut after the multiple has only begun

  • If 12% US procedure growth is structural rather than one-off, guidance has further to fall
  • China tender-style price cuts and domestic robot competition
  • Recall follow-through and FDA posture
Community Voices
  • StockTwits · Post-crash the split widened: one camp runs PEG math against $8.6B of cash and calls valuation finally fair, the other asks if 12% US growth is page one of a structural slowdown — nobody is sure, which is the honest part of this levelOriginal ↗
Institutional Views
  • 华尔街共识 · 35 analysts average $508, +47% from here — but that map predates the crash; JPM, Wells and Mizuho have already redrawn targets to the $400-487 range
  • Wells Fargo · Target slashed from $654 to $487Source ↗
  • JPMorgan · Target $550 to $450Source ↗
  • Baird · Outperform kept, target trimmed just $525 to $500 — the gentlest cut on the StreetSource ↗
Neutral
Score7/10RSI(14) 23.9Peer CDNS -9.47% 收 $330.11NVDA entry cost $414.79（已跌破）Watershed 7/27 K3 权重开源

Technicals — RSI 23.9 is the most oversold of today's deep-dive names, with the close under Nvidia's $414.79 entry cost; at 59% of its 52-week range with the 50-day ($471) and 200-day ($454) both overhead, the first bounce resistance is the $414-420 zone. Event-driven breaks obey technicals for bounces, not reversals

Fundamentals — May's raised FY2026 guide and design automation at 80% of Q2 revenue underline how deeply the duopoly is embedded at 2/3nm; the quieter subplot is the 7/7 decision to sunset manufacturing process-control software (EES/FDC) and pivot to high-margin AI design — a transition now colliding with a narrative shock

News — On 7/17 Moonshot disclosed that Kimi K3 ran a full design-to-verification flow on a 45nm, 4mm², 1.46M-cell chip in 48 hours using only open-source EDA — instantly compared to the January 2025 DeepSeek shock. The caveats are equally loud: no tape-out, no yield, no advanced-node signoff, weights and methodology not out until 7/27 — the same day Cadence reports

Short-term · 1-2 weeks
Neutral Sidelines

An RSI-23.9 technical bounce can fire any day, but bounces in an event vacuum are rented, not owned; let 7/27 adjudicate the narrative before picking a direction

Entry No position before the 7/27 double event (K3 weights + CDNS earnings) — until then every position is a bet against a black box. If replication fails and the narrative lifts, reclaiming $414 (Nvidia's cost basis) is the right-side signalStop —Target —
Long-term · months+
Neutral

The advanced-node duopoly and Ansys integration didn't evaporate overnight, but 'AI eats EDA pricing power' just graduated from sci-fi to a short thesis that must be rebutted quarterly

  • If third parties replicate even part of the demo after 7/27, the valuation anchor drops again
  • Execution risk in sunsetting the manufacturing-software line
  • An August guide-down would resonate with the narrative shock
Community Voices
  • SemiWiki · Industry forums are buzzing but engineers are broadly skeptical: between a 45nm open-source library and 2nm signoff lies a decade of engineering debt — engineer calm and trader panic never share a time zoneOriginal ↗
Institutional Views
  • 华尔街共识 · 26 analysts average $569.89, +48% upside on a Buy consensus — all struck pre-crash, so downgrade risk and dip-value coexist
  • BNP Paribas · Calls to buy the dip: K3 shows capability, but top customers still default to the duopoly for complex designsSource ↗
  • Bernstein · More guarded: K3 proves Chinese labs can match the US frontier, and the market's instant reaction is broadly rationalSource ↗
  • Benchmark · The worst-timed desk on the Street: initiated CDNS at Buy ($450) and set a $570 SNPS target the day before the crash, citing duopoly barriersSource ↗
Neutral
Score6/10Q2 core EPS $10.04 vs 预期 $5.34Combined ratio 83.6（改善 6.7pts）Core ROE 24.9%RSI(14) 75.5

Technicals — A gap-and-go breakout to record highs on 2.7x volume put RSI at 75.5 in overbought territory at 99.6% of the 52-week range; the earnings gap back to the $338 pre-print close is the first pullback checkpoint with a $9.31 ATR. Event-driven breakout days are for watching, not chasing — the entry logic lives at the gap retest

Fundamentals — Q2 net income of $2.21B (+46%) on ~$1.7B of underwriting profit with commercial underwriting income doubling, an 83.6 combined ratio and 24.9% core ROE; yet net written premiums of $11.53B were flat YoY — the CEO refuses to buy growth with price ('a fool's errand'), which earns marks for discipline and costs marks for growth

News — The 7/17 pre-open print crushed: cat losses nearly halved to $518M from $927M, $578M of favorable prior-year reserve development, investment income +14%, and $1.58B returned to holders ($1.31B in buybacks). As the first P&C major to report light cats, it was read as a positive preview for Chubb and Allstate, dragging insurers higher on a crash day

Short-term · 1-3 weeks
Neutral Sidelines

Upside from above every Street target relies on forced upgrades and rotation flows, neither of which is forecastable; waiting for the retest is cheaper than betting on inertia

Entry No chasing at RSI 75.5; a low-volume hold of the earnings gap (the $338 area) on a pullback is the second-chance entry for the defensive-rotation tradeStop —Target —
Long-term · months+
Accumulate

24.9% core ROE, pricing discipline and heavy buybacks make a quality compounder for the high-rate plateau — the center of the defensive rotation

  • The Aug-Sep Atlantic hurricane season is Q3's biggest unknowable
  • Half the quarter's excess profit came from unrepeatable cat/reserve tailwinds
  • Flat premium growth, if extended, erodes the multiple
Community Voices
  • StockTwits · The only victory lap in an AI-unwind week: from 'boring insurer' to 'chaos money-printer' in one earnings report — the sentiment pendulum swings far faster than the underwriting cycleOriginal ↗
Institutional Views
  • 华尔街共识 · 34 analysts average $334.17 — the close now sits ~10% above the consensus target, and the 1.9 rating mark is the weakest on our board; either the sell side gets dragged into upgrades or the price mean-reverts, one of the two
  • Truist · Initiated at Buy with a $395 target on 7/10 — the Street high pre-print, now the only target the price hasn't caughtSource ↗
  • Cantor Fitzgerald · Target $335 to $360 on 7/9, OverweightSource ↗

Rapid Scan (11 names)

TickerCloseChangeScoreDirectionOne-line take
CDNS logoCDNSUS$330.11-9.47%6NeutralDown 9.47% to $330.11 as the K3 shock hammers EDA. RSI 28.6 is oversold but the knife is still falling — Jul 27 earnings collide with the K3 open-weights release. Wait for the turn.
ASTS logoASTSUS$57.8+5.07%6NeutralUp 5.07% to $57.80, a dead-cat bounce after a 22% weekly drubbing. Convert deal, launch slip to 2027, rocket-company M&A chatter — three grenades still smoking. A bounce is not a reversal. Not touching it.
UNH logoUNHUS$426.09+0.64%6NeutralUp 0.64% to $426.09. The three-month +33% repair rally still looks healthy at RSI 57.9, but with no fresh catalyst there is nothing to chase. On watch.
KO logoKOUS$81.56-3.96%5NeutralDown 3.96% to $81.56 on double volume — even the defensives took a beating. With earnings season closing in, no entries until this red candle explains itself.
MRK logoMRKUS$127.5-0.10%6NeutralFlat at $127.50 (-0.1%). The +11% monthly pharma repair trade is on the right side of the tape, RSI 57.9 not stretched — but no edge to enter here. Just watching.
STX logoSTXUS$787.66+5.66%5NeutralUp 5.66% to $787.66, the least-damaged of the storage trio at -8.6% on the week — HDD is less levered to NAND/DRAM pricing. Sector beta, no stock-specific tape bomb. Watching the bounce.
WDC logoWDCUS$477.22+2.23%5NeutralUp 2.23% to $477.22 after a 15.4% weekly storage beatdown: DRAM price-hike forecasts trimmed as customers balk at 30% increases, Samsung/Hynix capacity adds looming. The profit-taking stampede after a +159% quarter is not done — not buying the bounce.
FIX logoFIXUS$1,674.06-0.39%6NeutralDown 0.39% to $1,674.06. The data-center MEP contractor is collateral damage in the AI capex scare, off 14% on the month on 2.8x volume — real disagreement. Let the panic wash out first.
ALAB logoALABUS$303.62-5.04%5NeutralDown 5.04% to $303.62, a fifth straight loss for -23.4% cumulative and $17B of market cap gone. A 204x PE is first against the wall in the Hynix contagion, insider selling on top. Pure valuation strangulation, no fundamental break — still not catching it.
CRWV logoCRWVUS$73.21+0.41%6NeutralUp 0.41% to $73.21, the eye of the neocloud storm. Meta flipping from anchor customer to competitor guts the $99B backlog's credibility, with $35B of debt and $536M quarterly interest on top. The financing structure is the whole debate — watching, not playing.
MTZ logoMTZUS$329.59-3.29%6NeutralDown 3.29% to $329.59 on 4x volume — the data-center construction panic has reached the infrastructure contractors. Off 13.5% on the week with the tape still deteriorating. Steering clear.
Stock Brief 2026-07-18: Daily U.S. & HK Market Analysis Archive · Quant Brief