Daily Brief Archive: July 4, 2026 — U.S. & Hong Kong Stock Analysis

Free preview

U.S. Markets

  • Pre-holiday Thursday: S&P -0.1%, Nasdaq -1.7% as chip stocks saw broad profit-taking
  • Memory bloodbath: Micron -21% and SanDisk -22% on the week — glut fears plus an antitrust suit
  • Tesla Q2 deliveries hit 480K (+25% YoY), crushing estimates — stock fell 7.5% anyway; sell the news
  • Oracle -42% on the month, near 52w lows: $70B FY27 capex plus a planned $40B raise
  • Markets closed Friday for July 4th; first post-holiday focus is SpaceX joining the Nasdaq-100

Hong Kong

  • Hang Seng +1.3% Friday, HS Tech +1.0%; southbound flows bought a net HK$4.5B
  • Robotics ripped: UBTECH +17.6%, Estun +33%, Horizon Robotics +9.1%
  • BYD +7.4%: June sales at 403K units with overseas volume up 95% YoY — the new growth engine
  • Alibaba loiters near 52-week lows around HK$94, sitting out the market bounce
  • High-yield SOEs stay oversold: China Mobile RSI 19.8, CCB RSI 23.4, both yielding 5%+

Today's Watchlist

  • Mon 7/7: SpaceX officially joins the Nasdaq-100 — index funds become forced buyers
  • Tesla earnings 7/22: deliveries are in the books; margins and ASPs are the only verdict left
  • Memory bull-bear duel: BofA lifts SanDisk target to $2,500 vs. Samsung/Hynix supply adds
  • Oracle's $134.6 52-week low is the AI-faith litmus test — a break spreads the fear
  • HK robotics is running hot: Estun +33% in a day at RSI 71 — chase only with a stop

Deep Dives (5 names)

Neutral
Score9/10RSI(14) 48.5Off 52w high -22.3%P/E (TTM) 22.1Revenue YoY +167%

Technicals — Broke the $1,000 handle to close at $976, down 20.9% on the week; the 20-day ($1,043) now caps rallies and the next real support is the 50-day at $852. RSI at 48.5 is merely neutral — at this volatility, no oversold reading means no bottom yet. Daily ATR is a wild $97 (~10%); stress-test any position for 10% daily swings.

Fundamentals — Fundamentals are intact: $41.5B revenue last quarter (4x YoY), management guiding memory tightness through 2028, and just 22x earnings against +167% revenue growth. What changed is the market pricing 'cycle peak' — Samsung and SK Hynix announced supply additions, and the 'AI capex peaks in 2026' thesis is gaining the upper hand.

News — Institutional profit-taking after an 80% first-half run is the main driver — no single fundamental break. Piling on: a California class action alleging Samsung, SK Hynix and Micron colluded to restrict DRAM supply (prices up 700% in four years). The selling spread to Korea, with Samsung and Hynix down 7–9% in a day.

Short-term · 1–3 weeks
Neutral Sidelines

Momentum is broken and the 20-day caps at $1,043. Wait for volume dry-up plus a reclaim of the 5-day before entering; a straight slide toward the 50-day ($852) means systematic exit from the memory trade — don't catch that knife.

Entry Scale in at $900–950 after volume dries upStop $850Target $1150
Long-term · months+
Accumulate

The AI memory supply gap hasn't gone anywhere: HBM capacity is locked under long-term agreements and management guides tightness into 2028. 22x earnings against triple-digit growth makes pullbacks a gift for the long book — provided you scale in and keep dry powder.

  • Samsung/Hynix supply lands and memory pricing peaks early
  • The DRAM antitrust suit turns into real fines or pricing constraints
  • If AI capex truly peaks in 2026, cyclicals get de-rated hard
Community Voices
  • WSB · Still #1 on the board, but mentions collapsed from 1,179 to 227 in 24h — retail gunfire stopped before the decline didOriginal ↗
  • StockTwits · Dip-buyers posting fills, top-sellers posting gains — each side convinced the other is the bagholderOriginal ↗
Institutional Views
  • 华尔街共识 · 52 analysts, average target $1,574 (+61% vs. spot), rating 1.13 — near strong buy
  • 加州集体诉讼 · Alleges Samsung/Hynix/Micron colluded to restrict DRAM supply and inflate prices 700% over four years — a legal overhang on the whole groupSource ↗
Neutral
Score8/10RSI(14) 46.8Off 52w high -25.9%P/E (TTM) 60.7This week -22.0%

Technicals — Down 14.1% on the day to $1,745 and 22% on the week, breaking the 20-day ($1,952) with the 50-day at $1,610 below. Daily ATR of $211 (~12%) makes it the most volatile mega-cap on the board. The 52-week range is $40–2,354 — on this chart 'support' is a psychological concept, not a technical one.

Fundamentals — The NAND shortage thesis holds: revenue +83% YoY, and BofA argues the supply-demand imbalance and firm pricing persist through 2027, lifting its target from $2,100 to $2,500. But at 60.7x earnings the 'shortage' is priced well past 2027 — the margin for error is razor thin.

News — Led the memory decline two days running (-10%, then -14%), on the same triggers as Micron: profit-taking, Samsung/Hynix supply-add chatter, and 'AI storage glut ahead' notes. BofA raised its target to $2,500 into the selloff — a rare head-on bull-bear collision.

Short-term · 1–3 weeks
Neutral Sidelines

A 12% daily ATR means any stop can get swept by noise — size it like a wager, not an investment; chips no bigger than dinner money. Lose the 50-day ($1,610) and the cycle story goes into triage.

Entry Small size at $1,600–1,700 (upper band of the 50-day) after stabilizationStop $1500Target $2000
Long-term · months+
Neutral

The purest NAND torque on the board, but a single product line at 61x earnings after a 38x one-year run leaves long-term odds against the buyer. For memory supercycle exposure, Micron's HBM story offers better value per unit of risk.

  • If NAND pricing peaks before 2027, it's a double de-rate
  • Samsung/Hynix/Kioxia add supply faster than expected
  • The volatility itself is a risk: a 12% daily ATR amplifies every mistake
Community Voices
  • WSB · #2 on the board with mentions down from 262 to 71: half shouting 'NAND tight through 2027,' half posting -22% loss pornOriginal ↗
Institutional Views
  • 华尔街共识 · 29 analysts, average target $2,174 (+25% vs. spot), rating 1.33 — buy-to-strong-buy
  • 美银 · Raised its target to $2,500 (from $2,100) into the selloff, keeping a Buy: NAND imbalance and firm pricing persist through at least 2027Source ↗
Long
Score8/10RSI(14) 41.2Off 52w high -17.6%P/E (TTM) 29.8Revenue YoY +70.7%

Technicals — Down just 1.4% to $194.83 in a chip-sector bloodbath week — notably resilient. Price is hugging the 200-day at $191, the first genuine test of the long-term trendline in 2026. RSI 41 is soft but not oversold; the 20/50-day ($203/$210) cap upside. Its relative strength is the best signal in the sector this week.

Fundamentals — Revenue +70.7%, EPS +110%, yet the P/E sits at just 29.8 — sideways price action plus earnings catch-up has pushed the multiple to its cheapest percentile in a year. The worry is the second derivative: if hyperscaler capex peaks in 2026, 30x isn't cheap either. For now, order visibility still runs into 2027.

News — The week's key incremental positive: the strategic partnership with Palantir, putting its AI models into U.S. government and critical-infrastructure deployments — a revenue narrative that doesn't lean on hyperscaler capex. Sector-wide profit-taking pressure lingers, with Broadcom's guidance scare still echoing.

Short-term · 1–3 weeks
Long Lean bullish

First 200-day retest, sector-leading relative strength, and the cheapest multiple in a year — the best odds on the board amid the panic. A break of $178 (one ATR below the 200-day) upgrades panic to trend reversal; take the loss and walk.

Entry Scale in at $188–195, around the 200-day at $191Stop $178Target $220
Long-term · months+
Accumulate

Pricing power in AI compute plus the CUDA moat remain unshaken, and 29.8x for 70% growth is a rare 'growth stock at a value multiple' window this cycle. Government and sovereign-AI orders are opening a second growth curve.

  • Hyperscaler capex peaks or shifts to in-house ASICs
  • Tighter export controls cutting off China and parts of Middle East demand
  • AI monetization disappoints and the whole complex de-rates
Signal BacktestCumulative +0.00%price itself -1.39%
  • 07-04Long$194.83 → open → $194.83+0.00%
Community Voices
  • WSB · Mentions fell from 168 to 39; the faithful stayed eerily quiet through the bloodbath — no panic is what real conviction looks likeOriginal ↗
Institutional Views
  • 华尔街共识 · 66 analysts, average target $313.4 (+61% vs. spot), rating 1.13 — firmly in strong-buy territory
  • Palantir 合作 · Teaming with Palantir to deploy AI across U.S. government and critical infrastructure opens a non-hyperscaler revenue lineSource ↗
Neutral
Score7/10RSI(14) 52.3Off 52w high -11.4%P/E (TTM) 1703-month gain +154%

Technicals — Down 4.3% to $517.82, sitting exactly on the 20-day ($516.8), with the 50-day at $460 and the 200-day way down at $278 — the uptrend is fully intact, just too steep. RSI 52 is neutral, ATR $36 (7%). The $500 round number plus the 20-day form the bulls' first must-hold line.

Fundamentals — The MI lineup keeps taking AI inference share, with revenue +35% and EPS +123% — but 170x trailing earnings means the market is paying on 2027's delivery. The Street's $511.8 average target sits below spot — not bearishness, just target-raising that can't keep pace with the tape.

News — No stock-specific negative — this is sector profit-taking. In early June's Broadcom-triggered AI chip squeeze, AMD dropped 10.9% in a day and fully recovered — money still treats it as the top offensive play on the AI inference era.

Short-term · 1–3 weeks
Neutral Sidelines

A trend this steep isn't for fresh heavy positions — 7% daily ATR on a 170x multiple amplifies any sector panic. Holders should trail stops to protect gains; those flat should wait for the 50-day squat.

Entry Light momentum size if $500 holds; the better entry is a deep retest of $460 (50-day)Stop $460Target $585
Long-term · months+
Accumulate

Inference is a bigger, longer-lasting market than training, and AMD's price-performance play is structurally favored there. With datacenter and client both firing, 2027 earnings can grow into today's multiple. Pullbacks toward the 50-day are long-term add zones.

  • 170x leaves zero tolerance — one guidance stumble is -20%
  • NVIDIA price cuts or CUDA lock-in fight back and stall the share-gain story
  • If the AI capex peak thesis lands, its high beta amplifies the downside
Community Voices
  • WSB · Mentions at 27, a monthly low: too high to pound the table, too strong to bottom-fish — it's always quiet near the summitOriginal ↗
Institutional Views
  • 华尔街共识 · 59 analysts, average target $511.8 (-1.2% vs. spot), rating 1.26 — the tape has lapped the targets; a wave of raises is pending
  • 板块观察 · If MI300X/MI400 share keeps expanding it directly compresses NVIDIA's premium — AMD is the biggest beneficiary of the AI narrative rebalancingSource ↗
Long
Score7/10Off post-IPO high -28.2%IPO price $135Starlink subs 10.3MNasdaq-100 add 7/7

Technicals — Up 2.8% to $162, within the post-IPO range of $147–226 and in its lower half. Three weeks post-listing, moving averages haven't formed yet — trading anchors to events and psychology: $147 (the post-IPO low) is the line of life and death, with trapped supply thick above $170.

Fundamentals — History's biggest IPO (raising over $85B), with Starlink at 10.3M subscribers across 164 countries — but ARPU slid from $86 to $66, it lost $4.9B last year and another $4.3B in Q1, and Starship has burned $15B. The $2.1T market cap buys the 'space monopoly' narrative, not the income statement.

News — Officially joins the Nasdaq-100 on 7/7 with passive buying imminent; consumer Starlink prices were raised $5–10/month in May to counter ARPU erosion; and Starship flight 13 targets late July — three catalysts stacked within two months.

Short-term · 1–3 weeks
Long Lean bullish

Index inclusion is guaranteed passive demand, and the Starship flight is a lottery option — clean event-driven logic. But this is a swing, not a hold: the inclusion premium historically unwinds within a month. The stop below the $147 post-IPO low is non-negotiable.

Entry $155–165 before the add date, scaling out within a week after inclusionStop $147Target $190
Long-term · months+
Neutral

A de facto launch monopoly plus Starlink's cash-flow layer is a real moat — but $2.1T of market cap on $5B annual losses prices a decade of flawless execution. The ARPU slide hints at Starlink's growth ceiling. Long-term, wait for a fatter pitch.

  • A Starship failure hits the valuation narrative directly
  • Continued ARPU erosion plus Amazon's LEO competition undercut Starlink economics
  • Any capital raise during the loss-making phase means dilution
Signal BacktestCumulative +2.83%price itself +2.83%profitable since 07-04
  • 07-02Long$157.54 → open → $162+2.83%
Community Voices
  • WSB · The Nasdaq-100 countdown is everywhere; everyone wants to front-run the index funds — the trade everyone wants is the hardest one to get paid onOriginal ↗
Institutional Views
  • 华尔街共识 · 13 analysts, average target $229.2 (+41% vs. spot), rating 1.62
  • The Motley Fool · Historically, Nasdaq-100 inclusion brings positive excess returns from passive buying — but the front-running tends to unwind within a month of the add dateSource ↗

Rapid Scan (18 names)

TickerCloseChangeScoreDirectionOne-line take
MU logoMUUS$975.56-5.49%9NeutralMemory leader down 21% on the week: fundamentals intact, positioning broken — wait for the flush to dry up
SNDK logoSNDKUS$1,745-14.13%8NeutralPure NAND torque, -14% in a day with a 12% ATR; BofA raised its target to $2,500 into the teeth of it
NVDA logoNVDAUS$194.83-1.39%8LongOff just 1.4% in the sector bloodbath; first 200-day test at $191 with the multiple at a one-year low
AMD logoAMDUS$517.82-4.26%7NeutralSitting on the 20-day after a 154% quarter; the tape has lapped analyst targets — wait for the squat
SPCX logoSPCXUS$162+2.83%7LongNasdaq-100 inclusion 7/7 brings forced passive buying, with Starship flight 13 late July — event window open
MSFT logoMSFTUS$390.49+1.62%7LongA +7.6% week and counting: the Mag-7 laggard at 23x — a rare drift into value territory
META logoMETAUS$582.9-4.90%6NeutralAnnounced its AI-cloud entry and fell 4.9% — the capex fear contagion jumped from Oracle to its own tape
B logoBUS$38.21+4.83%6LongLed gold miners +4.8%; 10.6x with EPS up 180% — it moves first when gold steadies
MSTR logoMSTRUS$100.77+7.90%6NeutralReclaimed $100 with a 7.9% pop, still -35% YTD — the double-edged beta of a leveraged BTC proxy
DTE logoDTEUS$154.06+2.32%5LongUtility at fresh 52-week highs — datacenter power demand plus a 3% yield; defense playing offense
GH logoGHUS$167.98-1.63%6LongShield's expanded coverage powered a 92% three-month run; RSI 77.5 — wait for the pullback to board
LMND logoLMNDUS$71.52+3.16%5NeutralA 23% week on an upgrade and AV-insurance dreams — but the $62 average target is already underfoot
GEO logoGEOUS$30.25-0.26%5NeutralImmigration-enforcement play up 30% on the month near 52-week highs — pure policy beta into Q2 earnings
RGC logoRGCUS$6.37-20.67%2Short-20.7% in a day, -74% on the month: the squeeze stock's curtain call — don't go treasure-hunting in a graveyard
WEN logoWENUS$8.6-3.80%4NeutralA 6.5%-yielding fast-food turnaround giving back an 11.7% month — retail's favorite cigar butt
DRTS logoDRTSUS$13.23+4.67%5NeutralRadiotherapy small-cap closing at all-time highs, +320% in a year — fierce trend, thin liquidity; speed bumps come without warning
BETR logoBETRUS$28.6-5.27%4NeutralSqueeze narrative and insider buying vs. a -5.3% tape — a sentiment stock's bull-bear meat grinder
EU logoEUUS$1.29-5.15%3NeutralUranium junior grinding along 52-week lows, down 54% in a year — the uranium story has run out of narrators
Stock Brief 2026-07-04: Daily U.S. & HK Market Analysis Archive · Quant Brief