Late-release note: This issue was delayed due to server maintenance; the content covers the 2026-09-29 (ET) pre-market perspective, for reference and archiving only.

US Pre-Market Brief | 2026-09-29 (ET) Tuesday

Coverage window: 2026-09-28 (Mon) 16:00 ET regular-session close → 2026-09-29 (Tue) 09:30 ET before the open; after-hours/pre-market quotes are timestamped. All times Eastern (ET).

0. Today's One-Line Summary

1. The strongest catalyst is NVIDIA: After the close on 9/28 it announced a new $150 billion share repurchase (total authorization $235 billion, running through January 2028 — the largest single buyback authorization in U.S. history, exceeding Apple's $110 billion in 2024), and released new AI safety software for AI agents plus the OpenShell open-source tool — directly propping up the AI leader under the weight of 10Y Treasury yields at 5.25%. NVDA rose +2.57% against the trend on Monday to close at $230.85, with funds continuing to flow back in pre-market.
2. The strongest theme remains the AI compute hardware chain: NVDA (buyback + products), TSM (reportedly planning a second U.S. manufacturing hub, unconfirmed), NVTS (+18% after hours on winning a U.S. Army 10kV SiC program), and MU (earnings Wednesday after hours, forward P/E of just 7×) all have catalysts. By contrast, software/cloud continues to bleed — ZS -9.9% and DUOL -6.6% after hours — compounded by a morning report from an unnamed consulting firm (source to be verified) claiming an annual revenue shortfall of roughly $4.2 trillion (about 2/3 of required revenue) for AI infrastructure; the AI valuation narrative has entered a split phase of "strong hardware, weak applications."
3. The biggest single-stock negative is FICO: FHFA Director Pulte announced that Fannie/Freddie will adopt a unified loan pricing grid, and Rocket is switching to VantageScore 4.0; the stock plunged 13%~20% pre-market (snapshots vary by quote source) — an institutional crack in the mortgage-scoring monopoly.
4. Macro pressure unresolved: 10Y yield at 5.21%~5.25% (20-year-high territory); the market prices roughly a 68% chance of a 25bp Fed hike on 10/28; gold plunged 4.07% on Monday and the VIX jumped 8% to 16.07. Geopolitically, Trump rejected Iran's proposal to reopen the Strait of Hormuz (WTI settled at $93.37), but overnight U.S. military escorts plus Saudi pipeline restoration of 3.5M bpd pushed crude lower pre-market, leaving energy stocks facing profit-taking pressure.
5. Pre-market status: Index futures are modestly rebounding, with wide disagreement across sources: CAPIS shows the S&P around +10, Nasdaq +85, Dow +60; FXDailyReport separately lists Dow +148 (+0.29%), S&P +19 (+0.25%), Nasdaq +132.5 (+0.43%); TradingStrategyGuides claims an early tape of "Nasdaq alone strong, Dow and S&P weak" — the three sources disagree, so take the actual open as authoritative. 10Y ~5.21%, oil lower, gold slightly rebounding, the dollar stronger against the euro and GBP and weaker against the yen, Bitcoin $84,192. Today at 10:00 the August JOLTS (customary release time 10:00; the source page lists only the date) and September consumer confidence (consensus ~90, prior 89.4); four Fed officials — Barr/Goolsbee/Musalem/Williams — speak in turn; MU earnings are Wednesday after hours.

1. News Overview

⚠️ Data-fetch failure note (ops): the local yfinance was rate-limited by Yahoo (YFRateLimitError, confirmed by a live test before generation); the curl fallback to stockanalysis.com returned a Cloudflare challenge page, and the stooq endpoint was down. Fundamental data was instead gathered by sub-agents via search/page scraping (mainly stockanalysis.com page data, company releases, MarketBeat, GuruFocus, etc.). Therefore: ① some tickers' exact 9/28 closing prices were not obtained (KMX last trade before earnings, DASH 52-week range, etc., marked "to be verified" in the text); ② valuation multiples use each source's TTM/snapshot basis, with minor differences across sources; ③ NVTS's stockanalysis 9/29 price data was anomalous (stuck at 0.00% unchanged), so the pre-market price uses TheStreet/Benzinga snapshots instead. Internal consistency check: the sub-agents' 9/29 intraday/closing prices were checked item by item (NVDA $227.21, MU $1,065.08, PANW $388.41, SMMT $16.39, FICO intraday low $609, etc.) and none are cited in the body; certain "distance from 52-week high" percentages were softened to range wording because exact 9/28 closes remain to be verified.

2. Strongest Themes, Ranked

3. Stock Strength Leaderboard

Bullish zone (sorted by total score, descending):

Bearish zone (sorted by impact):

* Bearish-list scores use the same scoring model (measuring "news intensity" rather than long value); the direction column already marks them — do not go long on the score.

4. Scoring Model (Total 100, by Category)

5. Top Stock Deep-Dives (Top 10)

1) NVDA (英伟达 / NVIDIA) | Priority deep-dive | S | 84 pts

• Related news: After the close on 9/28 (NVIDIA Newsroom², Reuters¹): the board approved a new $150 billion repurchase, lifting the remaining total authorization to $235 billion, running through January 2028 — the largest single buyback authorization in U.S. corporate history (exceeding Apple's $110 billion in 2024). The same evening it released new AI safety software for AI agents and the OpenShell open-source tool (Jensen Huang called OpenAI "one of the most influential companies in history," free to use any OpenShell components).
• Catalyst logic: This is a dual capital-return + product catalyst that does not depend on the current quarter. With the 10Y at 5.25% and the market questioning the AI capex narrative (the $4.2 trillion revenue-gap report), the $235 billion buyback is the strongest floor under the valuation; the AI safety software extends the narrative from "selling chips" to "AI software stack." No immediate revenue/EPS impact — the biggest impact is on sentiment and the valuation floor.
• Theme and phase: The AI compute theme is in a "crowded but still backed by real money" phase — hardware data (company Q3 guidance $108B±2%, data-center revenue +117% YoY) remains hard while the application side starts to be questioned.
• Fundamental check (stockanalysis et al., TTM to 2026-07-26): FY27Q2 revenue $96.22B (+105.9%), adjusted EPS $2.22 (beat 6%), gross margin 75.0%, data-center revenue $89B (+117%); TTM operating cash flow $134.4B, FCF $127.0B; cash and investments $62.5B, total debt $38.9B, net cash $23.6B ($23.6 billion; note: debt jumped from $11.4B within a year, judged to fund buybacks/infrastructure). P/E (TTM) 28.9×, forward 19.0×, P/S 18.1×, PEG 0.36. Consensus FY27 revenue $411.6B (+91%), FY28 $682.7B, 61 covering analysts at Strong Buy, average target $327.70. 52-week range $164.07–236.54; at the 9/28 close of $230.85 it is about -2.4% from the 52-week high.
• Moat: CUDA ecosystem switching costs, end-to-end AI training/inference stack, supply-chain lock-in, plus the net-cash + FCF machine.
• Pre-market and technicals: Monday +2.57% to $230.85 (largest market-cap gain in the entire market); keeps rising pre-market (confirmed by headlines across sources; exact pre-market gain: no reliable data). Position: 52-week high zone; not a post-earnings setup, so no IV crush issue.
• Final call: The highest-certainty name today. A buyback cannot create demand directly, but it changes the answer to "who holds the valuation when rates are 5.25%." Risks: macro beta (beta 2.2+), the AI-revenue-disproof chain ($4.2 trillion gap report, unnamed, to be verified; OpenAI pausing model releases), China restrictions. Verification points: whether pre-market/opening volume can hold above $230; semi ETF (SMH) linkage.

2) KMX (CarMax) | Watch closely | A | 71 pts

• Related news: 9/29 06:30 ET (company press release⁷): Q2 FY27 (ended 8/31) revenue $7.88B, +19.5% YoY, beat; retail same-store +13%, total volume 387,735 units +14.7%; diluted EPS $1.16 vs consensus ~$0.66–0.70; the restructuring (725 layoffs) contributed cost savings; announced the Q3 buyback resumption.
• Catalyst logic: Own-earnings driven. The big revenue beat is a genuine volume-plus-price beat; but note GAAP net income up only +2.1% YoY — the EPS jump partly comes from a lower share count via buybacks, the tax rate, and restructuring savings, so quality needs a discount. Impact ranking: sentiment > revenue expectations > EPS.
• Theme and phase: Consumer recovery theme; used-car demand and CAF credit quality improve as rates peak.
• Fundamental check (stockanalysis, 2026-09-25 snapshot): TTM P/E 37.2× (distorted at cycle bottom), forward 20.1×, P/S 0.29×, P/FCF 8.2×; TTM net income only $220 million (net margin 0.79%); cash $132 million vs total debt $18.83 billion (including auto-loan ABS), interest coverage 5.74×; share count -6.3% over the year. Consensus 3-year revenue CAGR only +1%; 19 analysts, average target $56.54 (Hold). Exact 9/28 close to be verified (closed $57.21 on 9/25).
• Pre-market and technicals: +3.4% (CAPIS pre-market). Earnings day — expect a volatile open.
• Final call: The "genuine part" of the beat is revenue, not profit. Suitable for watching whether the revenue trend reverses the mediocre consensus; avoid chasing the opening gap. Risks: consumer confidence (today 10:00), high leverage. Verification points: whether the opening 30-minute gap fills; management's tone on second-half credit losses.

3) PANW (Palo Alto Networks) | Watch closely | A | 68 pts

• Related news: After the close on 9/28 (Yahoo¹³): BTIG target $404→$425, Buy maintained (after management meetings, positive on Chronosphere/CyberArk/Prisma AIRS growth); intraday (Timothy Sykes¹⁴): Unit 42 launched Continuous Frontier AI Defense, wiring Anthropic/OpenAI models directly into the offensive security toolchain.
• Catalyst logic: Dual rating + product driver; Monday +4.63% to $392.09 led the S&P 500, and CAPIS recorded another +4.5% after hours (another quote source shows after hours roughly flat; the discrepancy awaits the open).
• Theme and phase: Within cybersecurity, a "platforms take all" phase — money concentrates from point products (ZS, S) into platforms (PANW, CRWD).
• Fundamental check (stockanalysis, 9/28 close basis): FQ4 FY26 (released 9/1) revenue $3.41B +34% acceleration, FY26 full year $11.5B (+24%); gross margin 70.4%, FCF margin 35.8%, net cash $504 million; but GAAP net margin only 2.7% (high SBC), share count +7.8% over the year. Forward P/E ~92.7×, P/S 27.7×; 55 analysts, target $396.12 — only +2% upside. 52-week high $398.88 (8/13); price right at record high.
• Final call: Strong momentum, accelerating fundamentals, but no target headroom + high valuation — poor risk-reward chasing. Risks: 82-week high position, dilution, rates. Verification points: whether it can break the $399 prior high on volume today; whether ZS selling keeps dragging sector sentiment.

4) SMMT (Summit Therapeutics) | Priority deep-dive | A+ | 64 pts

• Related news: After the close on 9/28 (company release¹⁰, WSJ¹¹): AstraZeneca's $2 billion strategic investment (subscribing convertible preferred, equivalent $18.36/share, an 18.6% premium to the 9/28 close), plus a clinical collaboration on ivonescimab + AZN's ADC pipeline; Citi the same day raised its target $40→$43.
• Catalyst logic: Dual equity-financing + BD catalyst. The $2 billion directly removes the market's biggest bear argument (cash exhaustion), and AZN's endorsement sharply raises the probability of the Phase III assets being realized; for AZN it is a strategic option, for SMMT a valuation re-rating.
• Theme and phase: Big-pharma capital injection into biotech — a BD theme; event-driven.
• Fundamental check (stockanalysis, 9/28 close basis): zero commercial revenue, TTM net loss $856M, FCF -$460M; pre-deal cash $690.7M, debt $20.4M; pro forma net cash ~$2.6–2.7 billion after the injection (computed value; completion timing per the announcement); short interest 26.26% of share count, insider ownership 76.9% (small actual float); 17 analysts, PT $28.55, Buy.
• Pre-market and technicals: Closed $15.48 on 9/28; after hours +15% to $17.85 at one point; pre-market +21.6% (CAPIS) — already above AZN's entry price of $18.36, more than half the premium realized.
• Final call: The event is real, the money is real, but with a 21% pre-market gain, chasing the open is risky; the approach is to wait for a pullback toward $18.36 (the AZN-endorsement price) and watch for support. Risks: convertible-preferred conversion terms (await the full 8-K), single-asset clinical risk, two-way high volatility of a short squeeze and collapse.

5) NFLX (奈飞 / Netflix) | Watch closely | B+ | 61 pts

• Related news: Pre-market 9/29 (TSG¹⁸): Deutsche Bank upgrade to Buy (from Hold, advising buy-the-dip after the sharp September pullback).
• Catalyst logic: Pure rating catalyst, indirect; the significance is "the first major-bank bottom-fishing signal after September's de-rating."
• Fundamental check (stockanalysis, TTM): revenue $48.37B (+13–16% slowing quarter by quarter), gross margin 49.1%, net margin 28.2%, ROE 49.5% (highest in the group); OCF $11.97 billion, FCF $11.15 billion; net debt $7.53 billion; P/E (TTM) 22.2×, forward 20.3×, PEG 0.96; 51 analysts, target $92.82 (+32%). Completed a 10:1 split in 2025/11; 52-week range $65.08–124.86,9/2026's deep drop having pulled the price back near the 52-week low zone (exact current price to be verified). Note: Q1 and Q2 both "revenue or EPS missed in at least one."
• Final call: Cheap + cash cow + major-bank endorsement, but the growth story's certainty has declined; suitable for watching whether institutional inflow persists after the upgrade (opening 5-minute volume), not for chasing on a single rating. Risks: pre-10/20-earnings wait-and-see, YouTube competition.

6) TSM (台积电 ADR / TSMC ADR) | Watch closely | B+ | 60 pts

• Related news: Pre-market 9/29 (TSG): reportedly planning a second U.S. manufacturing hub; the morning AI revenue-gap report and "hardware capex keeps expanding" form a bull-bear standoff.
• Catalyst logic: Capacity news is an indirect catalyst, confirming the political premium of U.S. advanced capacity; the substantive increment awaits company confirmation.
• Fundamental check (stockanalysis/company quarterly reports): Q2 2026 revenue $40.2B (+33.7%), gross margin 67.7% record high, EPS per ADR $4.31 far above expectations; full-year guidance raised to USD revenue +40%+, capex $60–64B; TTM FCF $35.89 billion, net cash $77 billion; forward P/E 19.9×; 21 analysts, target $552.26 (Strong Buy). 52-week high $479, price running in the 52-week high zone (exact 9/28 close to be verified, distance from high modest).
• Final call: One of the most solid fundamentals in the market, but today's catalyst is "media report" grade; as a core AI-hardware-chain allocation, watch whether it can reclaim the prior day's high (tape guide: TSG suggests watching whether TSM's opening order flow breaks through the prior day's supply zone).

7) NVTS (Navitas) | Watch only | B+ | 53 pts

• Related news: 9/28 16:41 ET (GlobeNewswire⁵): won the U.S. Army ALATTIS (Accelerated, Large-Area, 10kV SiC IGBT) program in a competitive selection (amount undisclosed); will develop 10kV SiC IGBT and PiN diodes for military systems.
• Catalyst logic: A government order endorsing the U.S.-domestic power-semiconductor supply-chain narrative; but the contract amount is undisclosed, so the elasticity against an annual revenue base of ~$42 million cannot be quantified — sentiment value exceeds near-term financial value.
• Fundamental check (stockanalysis): Q2 2026 revenue $10.5M (-28% YoY, +22% QoQ), gross margin 39.5% (rising), non-GAAP net loss $9M (improved 5% YoY); Q3 guidance $13.5M±0.5M (+28% QoQ, back to growth); net cash $552 million (~18% of market cap); P/S TTM 83.8×; share count +20.3% over the year (ongoing dilution); completed a $500 million ATM in July; consensus 3-year revenue CAGR +39%. Analyst PT $14.08 (Hold).
• Pre-market and technicals: +18.5% after hours to ~$13.98; pre-market 6:36 at $13.13 (+11.94%), back to $12.74 at 8:20 — the gap is narrowing, a classic gap-fill risk.
• Final call: A textbook "news-driven + loss-making fundamentals" small cap ($3.06 billion market cap, beta 3.9). Don't chase; watch $12 gap support. Risks: tiny contract amount, dilution, pace of AI-power revenue delivery.

8) MU (美光 / Micron) | Watch closely (event) | B+ | 51 pts

• Related news: FY2026Q4 earnings Wednesday after hours (radar/weekly calendar says 9/30; the stockmarketwatch list page separately shows 9/29 — sources conflict, date per company IR, flagged to be verified); consensus adjusted EPS $31.73 (source TipRanks; $3.03 in the same quarter last year), options-implied move roughly ±10%.
• Catalyst logic: Pre-event watch, not today's trade. The HBM/DRAM supercycle narrative + "cheap" forward P/E of 7.2× vs the widest consensus dispersion in the market (lowest target -66%, implying a cycle-top call).
• Fundamental check (stockanalysis): FY26Q3 revenue $41.46B (+345.7%), adjusted EPS $25.11 far above, GAAP gross margin 84.6%; TTM FCF $26.17 billion, net cash $23.75 billion (net debt just a year ago); P/E (TTM) 24×, forward 7.2×; 49 analysts Strong Buy, average target $1,520.76. 52-week range $154.65–1,255; the stock has already pulled back double digits from the 52-week high (exact current price to be verified).
• Final call: Wednesday's earnings are the next two-way re-pricing event for the AI hardware chain; today just watch, don't act — IV is elevated pre-earnings and IV crush is feared in either direction. Verification points: actual Wednesday-after-hours EPS vs $31.73; HBM sell-out pace and FY27 supply commitments.

9) FICO (Fair Isaac) | Avoid | Bearish S | (bearish model 65 pts, do not go long on the score)

• Related news: Night of 9/28 into the morning of the 29th (MarketWatch¹⁵, Investing.com¹⁶): FHFA Director Pulte announced Fannie/Freddie will adopt a unified loan pricing grid; the nation's largest mortgage originator Rocket announced VantageScore 4.0 as its preferred score; TransUnion simultaneously extended VantageScore.
• Catalyst logic: A regulatory + major-customer double hit, squarely on FICO's most profitable mortgage-scoring monopoly. Distinguish carefully: the FHFA rule changes the pricing grid (fee structure), while VantageScore replaces the score entry point — both lines squeeze at once.
• Fundamental check (stockanalysis, pre-event basis): Q3 FY26 revenue $674.2M (+25.7%, slightly below expectations), GAAP EPS $10.45 (+41%) beat; Scores segment +41%, but Software non-platform revenue -25%; gross margin 85.1%, FCF margin 41.6%, net debt $5.35 billion (long-term buybacks drove shareholders' equity negative). Pre-event P/E (TTM) 19.5×.
• Pre-market and technicals: Closed $840.89 on 9/28; pre-market -13.4% (CAPIS) to -17%~-20% (Investing.com headline); 52-week high $1,998.01 (2025/10, CNBC basis), already down about 58% from the peak — a "second leg down after the guillotine."
• Final call: An institutional bearish driver, not sentimental mispricing; the low multiple is the numerator collapsing, not the denominator improving. Avoid; if anything, only an event-driven short watch, but guard against an oversold bounce and regulatory-detail reversals. Verification points: whether other mortgage originators follow Rocket; FHFA detail hearings; VantageScore pricing.

10) ZS (Zscaler) | Avoid | Bearish A | (bearish model 42 pts)

• Related news: -9.9% after hours on 9/28 (CAPIS record); no new company release in the after-hours session — a market-level sector drain (software/cloud broadly lower); the recent CRO change (it gapped down on 9/25 for that) and the 10/6 investor day uncertainty loom.
• Catalyst logic: A decline with no own news is the weakest and hardest to stop; it forms an intra-sector mirror image with PANW's "rating upgrade + product launch."
• Fundamental check (stockanalysis): FQ4 FY26 (9/3) revenue $898.2M (+25%, growth stepping down from 30%+), ARR $3,771M (+25%), non-GAAP EPS $1.19 beat; but GAAP operating loss $133.3M; net cash $1.61 billion; forward P/E 40.4×; 46 analysts, target $208.86 (only +5.3%); -41% from the 52-week high.
• Final call: The valuation-vs-growth rebalance is incomplete and target headroom is zero; avoid — no visible catalyst before the 10/6 investor day. Risk (short's perspective): after -41%, any M&A rumor or broad market bounce could trigger a sharp snapback.

6. Bearish / Avoid List

7. Intra-Theme Ranking

AI compute / semiconductors (strength ★★★★★):

Cybersecurity (strength ★★★★, mixed):

Credit scoring / mortgage data (strength ★★★★):

Biotech BD (strength ★★★):

Consumer (strength ★★★, mixed):

8. Opening Verification Signals

• Pre-market: ① Can NVDA hold above $230 with volume — the master switch for today's sentiment; ② Is SMMT's +21.6% gap a gap-and-go or a gap-fill (watch the $18.36 AZN pricing line); ③ NVTS's gap has already narrowed (+12%→+8%), watch $12 support; ④ After FICO's plunge into the $670–730 zone, whether buyers step in.
• Index divergence: Early tape shows "Nasdaq alone strong, Dow weak" (TSG record); after 9:30 watch whether the QQQ vs DIA gap converges — if the Dow stabilizes, the rebound is high-quality; if tech alone can't hold, beware a broad drop. S&P short-term support 7600 / resistance 7700; QQQ support 730 / resistance 740 (CAPIS technical levels).
• Intraday: ① 10:00 consumer confidence (consensus ~90) and JOLTS — a big miss cools hike pricing (good for growth), but the "bad data = bad economy" read could also sell first and bounce later; ② four Fed speakers (Williams 14:00 is the most important; he said last week "possibly one more hike this year"); ③ sector linkage: whether SMH and QQQ move together, whether energy ETFs profit-take as oil falls (TSG suggests trimming upstream into strength); ④ whether ZS selling spills over to PANW/CRWD.
• Options sentiment: MU pre-earnings IV ~±10%, IV keeps rising into Wednesday; FICO skew turned negative post-event; NVDA not in an earnings window, IV relatively mild.
• Risk list: ① gap-up open then broad reversal (10Y back above 5.25%); ② lone-rally without sector follow-through (e.g., only NVDA up while SMH down); ③ futures diverging from the 10Y; ④ wait-and-see mood before macro data; ⑤ geopolitical second shock (Iran's reaction to the escort operation; now week 32 of the conflict).

9. Final Conclusions

① Five most noteworthy stocks today

② Three strongest themes today

1. AI compute hardware (NVDA buyback / TSM expansion / NVTS Army order / MU earnings preview) — hard catalysts, long durability, but needs to digest the valuation pressure from the $4.2 trillion revenue-gap report; representatives: NVDA, TSM.
2. Cybersecurity split (rating money floods into platform PANW, draining ZS/S) — medium durability; representative: PANW.
3. Big-pharma BD injection (AZN→SMMT $2 billion) — a single strong event catalyst; representative: SMMT.

③ Avoid today: FICO (institutionally damaged by regulation); ARM / high-valuation long-duration software (DUOL, S, etc., valuations keep compressing under 5.25% rates); chasing energy upstream (oil lower pre-market); China EV makers (weak without catalyst).

④ One-liner: Today is a day of "the AI leader fighting rates with real money" — NVDA's $235 billion buyback put a floor under the market, but the 10Y at 5.25%, the 68% October-hike odds and the Iran situation cap the rebound's height; structurally long hardware, avoid valuation, and watch the 10:00 consumer confidence and the Fed speakers.

⚠️ Risk warning: This list is only a pre-market information scan and watchlist, not investment advice. U.S. stocks carry high volatility and pre-market gap risk; post-earnings IV crush and guidance reversals can occur; automatically generated content may contain information lag or factual errors. Refer to company disclosures/SEC filings, and do not use this directly as a basis for trading.