Monday 8/24 — Desk Preview

Prepared Sunday 8/23/26. Landscape only, no recommendations.

The frame

Friday closed green across the board — S&P 500 +0.4% to 7,674.37, Dow +1.0% (+517.80) to 53,277.01, Nasdaq +0.4% to 26,180.45 — but all three indices still logged back-to-back weekly losses. The culprit was bond volatility: the 10-year backed up toward ~4.74% intraweek, capping the equity bounce even as VIX settled at 15.13 (-5.5% Friday) and WTI held ~$86.90. Retail earnings carried the tape's fundamental narrative last week: Home Depot beat top and bottom lines Tuesday, Target topped estimates with its first positive comps in over a year and TJX beat-and-raised Wednesday, Lowe's stayed under pressure on home-improvement softness, and Walmart capped it Thursday with adjusted EPS of $0.81 (vs. ~$0.73-0.74 consensus) on revenue of $187.9B (+5.9% YoY, beating the ~$186.3B estimate), e-commerce +23%, and a raised full-year outlook. Read together: the consumer at scale (WMT, TGT, TJX) is fine; anything tied to discretionary big-ticket/home spend (LOW) is not — the K-shaped economy thread persists, reinforced by July retail sales falling for the first time in nine months heading into the week.

The other tape-setter was Fabrinet (FN), which reported Monday 8/17 with record numbers — revenue $1.316B (+45% YoY), non-GAAP EPS $4.10, both beats — and still round-tripped: +5% intraday, -6.9% after hours, then down as much as ~20% the following session on weak margins, negative free cash flow, a softer near-term outlook, and a Thailand top-up-tax overhang, dragging peers like Marvell and Amphenol with it. That's the cautionary case study for the week: in AI-infrastructure-adjacent names, a clean beat is no longer sufficient if margins and forward commentary don't hold up — directly relevant with NVDA reporting Wednesday 8/26 after the close (consensus revenue ~$93-95B, EPS ~$2.07-2.08), the single print the summer's AI-capex debate has been building toward.

On policy: Kevin Warsh (sworn in as Fed chair May 2026) presided over a fractured July 28-29 FOMC that held at 3.50-3.75% on a 9-3 vote, with Cleveland's Hammack, Minneapolis's Kashkari and Dallas's Logan all dissenting in favor of a hike, and forward guidance stripped from the statement. The minutes, released Wednesday 8/19, showed the hawkish sentiment ran broader than the three dissenters ("policy tightening would likely be necessary if inflation did not decline") and revealed Warsh floating a cut in FOMC meeting frequency from eight to six per year. The market mostly looked through the minutes as backward-looking noise, rotating attention to this week: Jackson Hole runs Thursday-Saturday 8/27-29, with Warsh delivering his first keynote as chair Friday 8/28 at 10am ET (theme: "Financial Innovation: Implications for Payments and Policy," ~120 central bankers from 70+ countries). July PCE lands the same Friday. Markets are currently pricing roughly 1-in-3 odds of a September hike — not a cut — which makes Warsh's twenty minutes at the podium a live, two-sided risk event just 19 days ahead of the September 16 FOMC decision.

Against that backdrop, Monday itself is thin — NSSC and PDD before the open, no confirmed after-close names — but PDD is a genuine headliner: the last of China's "big three" platforms to report this cycle, arriving after JD.com (reported Aug 13) and Alibaba (reported Thursday 8/20) already put their numbers on the tape.

Tier 1 — before the open

PDD ($88.58) — the last of China's "big three" to report, and the stock that's been trading like the market already knows the answer. PDD Holdings reports Q2 2026 before the open with a 7:30am ET call; consensus sits around $2.72-2.77 EPS on $16.9-17.1B revenue. The prior quarter was ugly: Q1 2026 revenue of RMB106.2B (+11% YoY) missed the RMB109.82B estimate by 3.3%, and EPS of RMB9.51 missed RMB16.77 by roughly 43%, with basic EPS falling YoY (RMB10.59 to RMB8.94). The stock dropped as much as 8.4% to $88.51 on the print, erasing roughly $12.6B of market cap. The driver was a sharp deceleration in online marketing (ad) revenue growth — to roughly 2.5% YoY by some contemporaneous reads — masked at the headline level by transaction-services growth tied to Temu's third-party commission model, a mix shift the market read as margin-negative. That's the one metric that matters again Monday: ad-revenue growth rate versus transaction-services growth, and what it implies about take-rate and margin trajectory. PDD does not give formal quarterly guidance — the coded language to listen for is management's standard formulation about continuing to "invest for the long term" at the expense of near-term profit; the tell is whether that language gets more specific or more hedged than last quarter. Layered on top: the U.S. ended the de minimis exemption in 2025 (packages under $800 no longer duty-free), with tariffs on Chinese direct-mail parcels running as high as 54%, pushing Temu to shift from a China-warehouse "fully managed" model toward a "semi-managed" model using local U.S./Europe warehouses — watch for commentary on how far that transition has progressed and what it's doing to unit economics. Competitively, China's instant-retail subsidy war (Alibaba, Meituan, JD) burned an estimated RMB100B ($14B) in a single quarter last year before regulators fined seven platforms a combined RMB3.6B ($528M) in April for "ghost delivery" violations; the pattern into 2026 is a pivot from cash subsidies toward AI-integrated retail infrastructure (Alibaba folding an agent into its Qwen app, Meituan acquiring Dingdong Maicai) — the question for PDD is whether it keeps matching that spend or lets discipline show up in margin. On positioning: PDD's average earnings-day move over the last eight quarters has run about 7.6%, options activity has been elevated into the print, but short interest is modest at roughly 3.8% of float — this isn't a squeeze setup, more a repositioning-of-longs event. Forward P/E sits near 7.4x against a 52-week range of $71.94-$139.41, i.e., a stock priced well below its highs and at a real discount to global e-commerce peers. Bulls point to a valuation that already assumes stagnation, with transaction-services growth offsetting ad deceleration and the Temu U.S.-model reset largely lapped; bears point to unresolved margin compression, an unfinished subsidy war, and a tariff/regulatory overhang that JD's and Alibaba's own prints this cycle didn't resolve one way or the other.

NSSC ($42.63) — the first print under new CEO Kevin Buchel, and a valuation that leaves little room for a stumble. Napco Security reports fiscal Q4/full-year 2026 (quarter ended June 30) before the open with an 11am ET call — the first release since Buchel became CEO/President on July 8, so expect some "setting the table" language alongside the numbers. Consensus is $0.39 EPS on $53.2M revenue (+18.9% and +4.9% YoY respectively) — notably, EPS growth outrunning revenue growth implies further margin expansion is already baked into estimates. The prior quarter (fiscal Q3, reported May 4) beat: revenue $49.2M (+11.8% YoY), EPS $0.39 vs. $0.33 estimate, gross margin up to 60.0%, recurring service revenue (RSR) — the subscription line the stock's multiple depends on — up 15.4% YoY, adjusted EBITDA a record $15.8M (32.2% margin), non-GAAP net income +36.9%. GAAP results were distorted by a one-time $16M litigation settlement that drove a GAAP operating and net loss, so the underlying trend and the headline print diverged meaningfully — worth separating again Monday. The metric to watch is RSR growth and mix, since that recurring, high-margin base is what justifies the multiple. Balance sheet is clean: ~$100M net cash ($2.82/share), $0.15/quarter dividend. Valuation is rich for the category — forward P/E ~30.5x, trailing ~35x, EV/EBITDA ~28x — pricing this as a growth compounder rather than a legacy hardware security name. The 52-week range is $19.00-$48.12; the stock at $42.63 has already been substantially re-rated over the past year, with short interest a moderate 6.1% of float against a Street consensus that's overwhelmingly "Strong Buy" (average PT ~$48). Bulls see a small-cap converting equipment installs into durable recurring revenue with a fresh CEO able to reset the narrative; bears note that guidance implying flat sequential EPS against Q3's $0.39 actual, at 30x forward earnings, doesn't leave much cushion if the recurring-revenue growth rate decelerates.

The rest, by cluster

XPeng (XPEV) — not on the sheet, reports Monday before the open (8am ET call). Q2 deliveries of 103,295 vehicles (+65% QoQ) came in within the company's own 100,000-106,000 guided range, powered by the new GX flagship SUV. Company-guided revenue was RMB19.6-20.8B ($2.89-3.06B, +7.3-13.8% YoY); Street consensus sits near $3.02B revenue and a per-share loss of roughly $0.06. Watch gross-margin trajectory, overseas/Europe expansion commentary, and whether Q3 delivery guidance holds against China's ongoing EV price war.

EHang (EH) reports Tuesday 8/25, not Monday — some earnings-calendar aggregators misdated it into the Monday slate. Flagged here only because it extends the China-tech cluster around PDD/XPEV.

Heidmar Maritime (HMR) surfaced in one earnings-calendar sweep as a Monday 8/24 name; sourcing was inconsistent and a confirmed August 24 report date could not be verified — see Flagged as unverified.

Macy's (M) also appeared adjacent to this date in some calendar sweeps; confirmed reporting date is September 10, not Monday.

Where the asymmetries sit (landscape, not advice)

PDD at ~7.4x forward earnings is priced for continued bad news on margins and competition — a "clean" print (stabilized ad growth, no fresh tariff shock, disciplined subsidy spend) has real room to re-rate the multiple, while another miss simply confirms the discount is earned. That cuts both ways and isn't a lean either direction. NSSC sits at the opposite end of the valuation spectrum — 30x forward for a sub-$1.6B security-hardware/subscription name — where the asymmetry runs toward downside: a soft print or a cautious new-CEO tone has more room to hurt than a clean beat has room to help. Zooming out, FN's post-beat crater is this week's live reminder that AI-infrastructure-adjacent names are being judged on margin quality and forward tone, not just headline beats, directly ahead of NVDA Wednesday. And macro sits genuinely two-sided: bond vol (10-year near 4.74%) is capping equity follow-through, Jackson Hole and July PCE both land Friday 8/28, and the market is pricing meaningful odds of a hawkish surprise from Warsh's debut speech — not the dovish-pivot assumption that's dominated recent cycles.

Dominoes

Monday's PDD/NSSC/XPEV prints open a stacked week: NVDA reports Wednesday 8/26 after the close, the print the summer's AI-capex debate has been building toward — set against a backdrop of hyperscaler capex now guided to a combined $720-745B for 2026 across the big four, financed increasingly with external capital (Alphabet's ~$84.75B equity-and-debt raise in June, buyback suspended, long-term debt roughly doubled to $98.2B; Meta's capex guide lifted to $125-145B from $115-135B; Oracle planning ~$40B of combined debt and equity for FY27) — the funding-strain thread that sits behind every AI-infrastructure print this quarter. Thursday-Saturday brings Jackson Hole, with Warsh's keynote and July PCE both landing Friday 8/28, feeding directly into the September 16 FOMC decision. On the China-consumer side, PDD's Monday print closes out the reporting cycle for the "big three" (JD Aug 13, Alibaba Aug 20, PDD Aug 24), with EHang's Tuesday 8/25 print extending the China-tech read-through into the back half of the week. Separately, the SMCI export-control indictment (co-founder Yih-Shyan "Wally" Liaw charged individually alongside two others; the entity itself was not charged) remains an unresolved legal overhang on the AI-hardware supply chain worth tracking as the case proceeds, and Micron's memory-cycle commentary (not reporting this week) remains the next scheduled DRAM/NAND cost-spike read-through for the AI-capex debate.

Flagged as unverified

• Heidmar Maritime (HMR) reporting on Monday 8/24 — sourcing was inconsistent (one source pointed to Aug 17, no confirmed Aug 24 filing found); could not verify.
• PDD USD consensus varies narrowly by source ($2.72-2.77 EPS; $16.94-17.13B revenue) — a midpoint was used for the calendar stats.
• No specific options-implied move percentage was found for PDD's or NSSC's Monday print; PDD's historical average earnings-day move (~7.6% over the last 8 quarters) is cited as context only, not a forecast.
• Alibaba's (8/20) and JD's (8/13) actual results and stock reactions were not independently re-verified in this pass — only their report dates are confirmed; they're referenced directionally as already-reported comps for PDD.
• The July PCE release date is reported inconsistently across this week's calendar sweeps (Friday 8/28 per most sources and the customary BEA last-Friday slot; one aggregator claimed Wednesday 8/26) — Friday is used here; verify against the BEA calendar.
• NVDA's data-center-segment growth figure (cited near >$25B in one aggregator) was not cross-checked against a second source.

Sources: company IR/press releases and SEC 6-K/8-K filings (PDD, NSSC, XPEV, EHang, Macy's, Walmart, Fabrinet, Alibaba, JD.com, SMCI, Federal Reserve); financial news and aggregators (CNBC, Yahoo Finance, Investing.com, TipRanks, Benzinga, StockTitan, Fortune, TradingKey, 24/7 Wall St., CnEVPost); data/statistics sites (stockanalysis.com, MacroTrends, Finviz, Nasdaq short-interest data, MarketChameleon, OptionsLam, TradingEconomics); Fed communications and Jackson Hole/Kansas City Fed coverage (MNI Markets, Federal Reserve press releases).

Tuesday 8/25 — Desk Preview

Prepared Sunday 8/23/26. Landscape only, no recommendations.

The frame

Market closed out the week of 8/21 firm and calm — S&P 500 7,674.37 (+0.43%), Nasdaq Composite 26,180.45 (+0.43%), 10-yr 4.74%, VIX 15.13 (-5.5% on the day, down from 16.01) — heading into what is genuinely one of the more loaded five-day stretches of the year. Kevin Warsh chairs a Fed that held 3.50–3.75% on a fractured 9-3 vote at the July 28-29 meeting, with Cleveland's Hammack, Minneapolis's Kashkari and Dallas's Logan all dissenting in favor of a 25bp hike — not a cut. Minutes released 8/19 showed the hawkish sentiment ran wider than the three dissenters, and Warsh himself floated cutting the FOMC calendar from eight meetings to six. One calendar note to nail down: sources conflict on whether July PCE and the second Q2 GDP estimate land Wednesday 8/26 or in the customary Thursday-GDP/Friday-PCE slots (8/27-28) — verify against the BEA calendar Monday. Either way, Wednesday is the most dangerous day of the week: NVDA's after-close print (Q2 FY27, guided to ~$91B revenue ±2%, consensus $91.7B / $2.08 non-GAAP EPS, roughly double last year) possibly stacked on top of the macro data. Friday 8/28 is Jackson Hole (theme: "Financial Innovation: Implications for Payments and Policy") and Warsh's first keynote as chair at 10am ET, three weeks ahead of a September meeting the market currently prices at roughly 1-in-3 odds of a hike.

Tuesday itself is a genuine two-tier day: BMO, BNS and DKS before the open; INTU, HEI, ZM, BOX, NCNO and SMTC after the close — all nine sheet names confirmed for 8/25 via company press releases and a cross-checked professional trading calendar. One correction worth flagging loudly: MongoDB does not report this week — MDB's Q2 FY27 print is confirmed for Monday, September 1, after the close, not Tuesday 8/25. Also worth knowing: the retail/mega-cap-tech pileup traders often mentally lump into "earnings week" — BBY, DG, DLTR, BURL, ULTA, MRVL, WDAY, CRM, CRWD, plus CIBC/RBC/TD — all land Wednesday or Thursday, not Tuesday. Tuesday's own non-sheet crowd is smaller and skews Chinese ADRs (detailed below). No other US large-cap is quietly missing from Tuesday.

Tier 1 — before the open

BMO ($238.80 CAD / ~$163 est. USD) — Canadian banking under a hawkish-Fed, tariff-anxious backdrop. BMO reports fiscal Q3 (May–July) with consensus soft after a strong Q2: adjusted net income was up 34% y/y in that quarter, credit provisions came in at $739M versus a $775M Street forecast (a beat), and global-markets/trading revenue was up 15% y/y to $1.32B. The crack in that quarter's release was BMO's own CRO flagging rising provisions on impaired credit-card loans and broader "economic headwinds in Canada" with consumer insolvencies climbing — even as mortgage-book credit quality was described as improving. Analyst consensus sits at Hold with a C$228.61 target (modest downside from spot) even after a run of price-target hikes through the summer (Scotiabank to C$239, TD Securities to C$239, CIBC to C$244) — a sign estimates have caught up to the stock rather than the other way around. What to watch: whether the unsecured-consumer provisioning trend from Q2 continued into Q3, US-segment performance (BMO has real US regional-bank exposure), and any explicit tariff commentary given Canada's trade exposure into a period where Warsh's Fed is leaning hawkish rather than toward cuts.

BNS ($119.36 CAD / ~$81.50 est. USD) — the Mexico/Latin America read inside Canadian banking. Bank of Nova Scotia hit an all-time high (C$127.63) on 8/17 before pulling back into the print. Prior quarter (Q2 FY26, reported May 27-28): adjusted EPS C$2.02 beat the C$1.94 Street estimate by ~4%, revenue C$9.84B was up 8% y/y, Canadian Banking pretax pre-provision earnings grew 13% y/y, and the productivity ratio improved 290bps to 52.5% — a genuinely clean beat, with a dividend hike attached. The caveat management itself flagged: "elevated energy costs, persistent trade uncertainty and higher unemployment" pressuring consumers and businesses, with retail-impaired performance in international markets — specifically Mexico — expected to stay elevated "due to trade uncertainty." BNS's Pacific Alliance footprint (Mexico, Peru, Chile, Colombia) makes it the one Tuesday name with the most direct read on how US tariff policy is bleeding into Latin American consumer credit. 2026 guidance calls for double-digit EPS growth and positive operating leverage; Tuesday tests whether that holds against the international-provisioning drag.

DKS ($183.23, +2.17% Friday) — Foot Locker integration, at 52-week-range lows. Consensus: EPS $3.80 (-13.2% y/y on FL-related costs), revenue $5.6-5.64B (+~55% y/y, almost entirely the Foot Locker consolidation), with legacy DKS same-store sales expected +4.1%. The prior quarter (Q1, reported late May) is the template to watch for repeat behavior: net sales +62.7% to $5.16B beat on the top line, but GAAP EPS of $3.54 (non-GAAP $2.90) missed on the bottom line, weighed by $96.5M in FL-related charges ($53.8M M&A/severance/store-closing, $42.7M inventory clearance). FL itself posted its first positive comp (+0.6%) since fiscal 2024 — a real turnaround signal — yet DKS simultaneously cut full-year consolidated operating income guidance to $1.69-1.81B (from $1.71-1.83B) and EPS guidance to $13.27-14.27 (from $13.70-14.70), even while nudging comp-sales guidance up. That combination — better sales, worse margin guide — is exactly the tension Tuesday's print needs to resolve. Stock is down 17.8% over 52 weeks and sits well off the $254.60 high near $166.37 lows; short interest is a notably elevated 8.36% of shares out (12.26% of float, 6.76 days to cover) against a 26-analyst Buy consensus with a $251.05 average target (+37% implied) — a real divergence between sell-side optimism and short positioning. Watch for tariff-driven footwear/apparel input-cost commentary alongside the FL margin trajectory.

Tier 1 — after the close

INTU ($367.00, +1.42% Friday) — the headliner, and the cleanest AI-agents-in-software test case on the calendar. This is Intuit's fiscal Q4 (May–July, seasonally its smallest quarter) plus the market's first look at FY2027 guidance, with a formal investor day set for September 17. Consensus: EPS $3.54-3.59 (adjusted) on revenue near $4.27B, in line with management's own $3.56-3.62 EPS / $4.25-4.28B guide (11-12% growth). The prior quarter is the whole story here: Q3 FY26 (reported May 20) actually beat — EPS $12.80 vs. $12.28 consensus, revenue $8.56B (+10% y/y, slowest pace since 2024) — but management admitted on the call that TurboTax "lost on price" against cheaper DIY competitors, and paired that with a 17%, 3,000+ person workforce cut. The stock fell 13% after hours that day and kept falling to a cumulative -20% by the next close ($383.93 → $307.07). Multiple securities-fraud class actions followed (lead plaintiff deadline September 8), and the past three weeks have brought a wave of price-target cuts — Truist downgraded to Hold ($350, from $410), TD Cowen to Hold ($304, from $504), while Deutsche Bank, Citi and Mizuho all cut targets ($425, $457, $430 respectively) while keeping Buy/Outperform ratings — sell-side trimming numbers without abandoning the thesis. The stock has since recovered to $367, roughly 20% above the May trough but still miles below both the pre-print $384 and the 52-week high of $705.08 (52-week range: $252.84-705.08, forward P/E now a de-rated 13.73 vs. its historical software-premium multiple). Options are pricing a ~5.2% implied move (some sources cite an 8-quarter median closer to 6.3%, i.e., the market may be underpricing volatility here), with pre-earnings volume running 3.8x normal and puts modestly leading calls 7:6. Short interest has nearly tripled off the crash, from ~1.7-2.0% in the spring to 5.09% of float as of the most recent settlement date — a real bet has built against this name since May. The core question for the desk: does Intuit's "Intuit Intelligence" agentic push across QuickBooks Online Advanced, the Enterprise Suite mid-market platform, and TurboTax actually let it raise its effective take (AI doing higher-value work customers used to pay accountants for) — or does it keep compressing pricing power the way the TurboTax DIY admission just demonstrated? Tuesday's Q4 print and the FY27 initial guide (Mizuho explicitly expects a "reset" lower) are the first hard data since the crash; the September 17 investor day is the second.

The rest, by cluster

Aerospace/defense compounder — HEI. HEICO reports Q3 FY26 (fiscal quarter ended July 31) after the close, call Wednesday 9am. Consensus EPS $1.51 (+19.8% y/y), revenue $1.34B (+17% y/y); Zacks' model leans toward a beat. The prior quarter was a genuine record — net income +49% to $233.8M, Flight Support Group sales +21% to a record $929.4M (26.2% margin), Electronic Technologies Group +34% to a record $459.5M (26.5% margin) — on the back of June bolt-on acquisitions (CalRamic ceramic capacitors, Cook Defense Systems armored-vehicle track systems). Stock is up ~30% in the past month alone, trading at a rich 38.76x forward earnings within a $256.11-376.86 52-week range — priced for continued perfection on commercial-aviation aftermarket demand and defense/space momentum.

AI-monetization software cluster — ZM, BOX, NCNO. All three report Tuesday after the close and all three are running some version of "does AI expand or cannibalize the seat-based model." Zoom ($107.43, mkt cap $31.5B, fwd P/E 17.85) guided Q2 FY27 revenue to $1.265-1.270B (+4.2% y/y) after a Q1 beat (EPS $1.55 vs. $1.32 est., AI Companion paid users +184% y/y, a fresh $1B buyback authorization) — short interest is a low 2.46%, not a crowded trade. Box ($32.91, mkt cap $4.56B, fwd P/E 20.31) is the more interesting setup: Q1 delivered its first double-digit revenue growth in 12+ quarters (+11% y/y) on Enterprise Advanced/Box AI attach, consensus for Q2 is EPS $0.37 on $307.4M revenue, but the company's own guide (~$319M) sits meaningfully above that consensus figure — worth verifying at print rather than assuming the gap is stale data. Short interest on Box is a hefty 11.07% of shares out. nCino ($20.62, mkt cap ~$2.3B, fwd P/E 15.90) is the most heavily shorted name on the entire Tuesday slate — 17.05% of shares out, 20.38% of float, 7.45 days to cover — despite a beat-and-raise Q1 (revenue +11% y/y, non-GAAP operating margin +1,100bps y/y, "Banking Advisor" AI usage up 38x since October) and a raised FY27 ACV guide. That combination — strong fundamentals, extreme short interest, -26% over 52 weeks — makes NCNO a genuine binary event.

China ADR cluster (non-sheet). Tuesday's before-open slate also includes BZ (Kanzhun/Boss Zhipin), EH (EHang), TOUR (Tuniu), VIPS (Vipshop) and GFI (Gold Fields), with JOYY, NOAH (Noah Holdings) and QFIN (Qifu Technology) after the close, alongside small-caps CTRN (Citi Trends), SLQT (SelectQuote), STRT (Strattec) and ELMD (Electromed). None carry the size or liquidity to move the tape, but the concentration of Chinese consumer/fintech ADRs reporting the same day is a minor read-through on China-consumer sentiment worth a glance, not a focus.

Where the asymmetries sit (landscape, not advice)

INTU: a name where sell-side has cut price targets aggressively but kept ratings, short interest has roughly tripled since May, and options are pricing a smaller move than the stock's own recent history — three signals pointing in different directions on how much is actually "priced in." DKS: a 37%-upside Buy consensus sitting against 8.36%/12.26% short interest — the market is split on whether the FL-driven guide cut was one-time integration noise or a margin-structure problem. NCNO: the extreme short base (17-20%) against two straight quarters of guide raises is the most binary setup on the sheet — a third beat-and-raise risks a real squeeze, a stumble validates the shorts. BOX: 11% short interest against a name that just posted its best revenue growth in three years. SMTC: not shorted heavily (7.85%) but re-rated +150%+ over the trailing year to a 41.75x forward multiple with no trailing GAAP earnings backstop — priced for the 35% sequential data-center growth guide to hold, which makes it more exposed to disappointment than the short-interest figure alone would suggest.

SMTC deserves its own line as the AI-datacenter-optics read. Semtech guided Q2 to ~35% sequential data-center revenue growth off a Q1 beat (EPS $0.51 vs. $0.46 est., revenue $291M +15.9% y/y) driven by 800G LPO design wins with hyperscalers and module makers in both the US and China, plus record LoRa sales. Consensus for Tuesday: $0.62 non-GAAP EPS (+51% y/y), $328.4M revenue (+27.5% y/y). This print sits at the intersection of two of the desk's live threads: "peak AI capex vs. backlog evidence" (does the 35% sequential DC growth guide actually hold, or crack first here before it shows up in NVDA) and the memory-cost/hardware-margin squeeze (Micron has locked in "historically high" DRAM/NAND pricing for five years amid AI-driven supply tightness — SMTC isn't directly memory-exposed, but it sits in the same AI-infrastructure cost chain that's now absorbing that squeeze elsewhere). A soft data-center number here, a day before NVDA, would land differently than it would any other week.

Dominoes

Tuesday's prints (especially INTU on AI-agent monetization and SMTC on AI-datacenter optics) set the tone heading directly into Wednesday 8/26 — NVDA's after-close report, possibly stacked on top of July PCE/Core PCE and the second Q2 GDP estimate the same morning (release dates conflicting across calendars — verify). That is arguably the single most important 30-hour window of the month. Thursday 8/27 brings the retail cluster the market often mentally attaches to Tuesday — BBY, DG, DLTR, BURL, ULTA — plus CM/RY/TD and MRVL/WDAY, making DKS's Tuesday print an early, imperfect tell for that Thursday retail read. Jackson Hole runs Thursday through Saturday, with Warsh's first keynote as Fed chair Friday 8/28 at 10am ET, three weeks ahead of a September FOMC where the market is pricing meaningful odds of a hike, not a cut — a dynamic this whole earnings week is unfolding underneath.

Flagged as unverified

• The July PCE / Q2 GDP second-estimate release dates conflict across sources this week (Wednesday 8/26 per one aggregator vs. the customary Thursday-GDP/Friday-PCE 8/27-28 slots per most others) — verify against the BEA calendar before trading around them.
• BMO and BNS USD price/market-cap figures in the calendar stats are estimated conversions from CAD quotes using an approximate FX ratio derived from a stale cross-reference (not independently confirmed for 8/21/26); treat as directional, not exact.
• BMO's forward P/E and price-target figures were sourced from a stockanalysis.com snapshot last fully refreshed June 26, 2026 — likely stale relative to the Q2 print and subsequent target hikes referenced in the same note.
• HEI short interest could not be sourced this session.
• VIX close for 8/21 appeared in search results as both 16.01 and 15.13; 15.13 is used here as the more specific closing print, with 16.01 read as the prior day's close.
• INTU implied-move figures came from two sources that didn't fully reconcile (5.2%/$36.39 vs. a cited "6.3% median move" comparison) — reported as found rather than resolved to one options-desk source.
• Box's cited consensus revenue ($307.4M) sits below the company's own Q2 guide (~$319M); flagging the gap rather than assuming which is current.
• NCNO consensus EPS for Q2 FY27 could not be sourced with confidence; only the company's revenue guide range is used in the calendar stats.

Sources: Company press releases and 8-Ks (Intuit, DKS, HEICO, Box, nCino, Semtech, Zoom, BMO, BNS); Federal Reserve/Kansas City Fed; CNBC, Reuters, WSJ, Yahoo Finance, TipRanks, StockAnalysis.com, Benzinga, Zacks/TradingView, SeekingAlpha earnings-call transcripts, Investrade weekly event calendar, Earnings Whispers.

Wednesday 8/26 — Desk Preview

Prepared Sunday 8/23/26. Landscape only, no recommendations.

The frame

Wednesday sits in a vice. On one side, a Fed under new chair Kevin Warsh that just held at 3.50–3.75% with a 9-3 vote and three hawkish dissents (minutes out 8/19); on the other, Warsh's first Jackson Hole keynote Friday 8/28 at 10am ET (symposium runs 8/27–29, theme "Financial Innovation: Implications for Payments and Policy"), with July PCE also due Friday. Markets are pricing roughly 1-in-3 odds of a September hike — an unusual place to be two days after the single biggest earnings print of the cycle. Friday 8/21 closes: S&P 500 7,674.37 (+0.43%), Dow 53,277.01 (+0.98%), Nasdaq 26,180.46 (+0.43%), Russell 2000 3,017.87 (+0.85%), VIX 15.13 (-5.50%), gold $4,680.60 (+2.39%), Bitcoin $77,274 (best week in two years), WTI $87.06. Note the divergence: equities bounced Friday but every major index still closed the week lower after a bond selloff — Treasury Secretary Bessent expanded bond buybacks past $4B/issue mid-week trying to cap long yields, but the relief was brief and 10s/30s snapped back toward the week's highs (exact Friday close not independently confirmed — flagged below). Gold and bitcoin both ripping while stocks limp into the week is a tell about how much duration risk and macro hedging is running underneath the AI trade. Layer on Monday's scheduled Bessent press conference detailing an "economic warfare" plan against Iran (China is the pressure point, as the largest buyer of Iranian-adjacent Gulf crude), and NVDA's print Wednesday night has to clear a market that's already jumpy about rates, oil, and China all at once — then hand the baton straight to Warsh 36 hours later. This is a "position for the week, not just the print" setup.

Tier 1 — before the open

ANF ($~105-109) — decelerating off a hard compare, tariff drag now explicit. Consensus: EPS $1.90 (-18.1% y/y), revenue $1.24B (+2.8% y/y). Company guidance called for net sales +2-4% y/y and a 10% operating margin including a specified $20M tariff hit — the first time the company has sized the tariff line that cleanly, which is itself the tell to watch (is it holding or widening). Forward P/E ~7.3-8.0x, short interest ~10%, reflecting a stock that ran hard on the turnaround story and is now facing its toughest comp of the cycle. Both sides: bulls point to AUR discipline and Americas/APAC strength persisting; bears note a mid-teens EPS decline consensus already embeds real deceleration, and any miss on margin (not just sales) reprices a name still carrying a "premium retailer" multiple relative to peers.

BBWI ($20.07) — leadership transition meets a soft prior-quarter revenue print. Reports BMO with new CEO Daniel Heaf and interim CFO Tom Javitch on the call. Consensus: EPS $0.36, revenue $1.552B. Last quarter beat on EPS ($0.49 vs. $0.47) but missed revenue ($1.40B vs. $1.42B) — the pattern to watch is whether that gap (margin engineering over top-line) repeats. Forward P/E ~7-8x, short interest a heavy ~10.3% of float. This is a battleground name: cheap on paper, but the market is pricing real skepticism about whether a new CEO can reaccelerate comps in a promotional mall-retail category.

KSS ($18.42) — the most mechanically loaded name on the whole slate. Short interest is 27.2% of shares out — genuinely extreme, and the setup that most resembles a pure positioning trade independent of the print itself. Consensus: revenue $3.32-3.52B (roughly flat to -0.9% y/y), EPS $0.56 (flat y/y), but the EPS estimate has been revised up 40% over the past 30 days — a meaningful directional signal from the sell side. Q1 delivered the best comp performance in four years, and the trailing four-quarter average surprise is +69%. Both sides: a squeeze is mechanically live on any in-line-or-better print given the short base; the bear case is that value/low-income consumer spending has been the softest cohort all year, and the turnaround narrative still needs multiple consecutive clean quarters, not one, to be believed.

WSM ($242.31) — the quality retailer, priced for continued execution. Consensus EPS $2.03 (+1.5% y/y). FY26 guidance stands at $8.02-8.33B revenue with 17.5-18.1% operating margin, and the company has beaten in each of the last four quarters. Forward P/E ~19.8x — a real premium to the rest of the retail cluster, which means the bar is "beat and hold guidance," not just beat. The read on tariff/input-cost management here doubles as a read on the higher-income consumer generally, in contrast to KSS/BBWI's lower-income exposure.

DY ($437.76, stale — see flags) — the literal fiber/grid-buildout read. Consensus: EPS $4.62 (+38.7% y/y), revenue $1.97B (+43.2% y/y) — this is a company still growing revenue faster than 40% on a base already in the billions, driven by fiber-to-the-home and data-center-adjacent grid work. Last quarter beat EPS estimates by 61.9% and revenue by 18%; the trailing four-quarter average surprise is +25%. Forward P/E ~26x, short interest 5.4%. This is the name to watch as a tell on whether the physical infrastructure side of the AI/power buildout (not the chips, the actual trenching and interconnection work) is still accelerating or starting to normalize off an unsustainable growth rate.

LI ($~17-20, low-confidence — see flags) — China EV price war, unfiltered. Guided Q2 deliveries of 95,000-100,000 vehicles (down 10-14.5% y/y) and revenue of RMB 24.1-25.4B (down 16-20% y/y). Consensus has swung to a per-share loss (~-$0.01) after Q1 posted a RMB 2.3B net loss (vs. RMB 647M profit a year ago) as gross margin collapsed to 7.9% from 20.5%. Short interest 4.5%. This is the cleanest read on Chinese EV pricing dynamics on the whole sheet — the question isn't whether growth is negative (guided already), it's whether margin has stabilized at all sequentially from Q1's cliff, or is still bleeding.

Tier 1 — after the close

NVDA ($214.72, $5.20T mcap) — the referendum, and it gets the deepest treatment of the week. Consensus sits at EPS ~$2.08-2.09 and revenue ~$91.7-92.1B, essentially right on top of the company's own Q1 guide of $91.0B ± 2% (ex-China Data Center compute). That's the first thing to internalize: unlike prior quarters where Street numbers ran meaningfully above guide, this time consensus and guidance are converged — which raises the bar for what counts as a "real" beat. Context on the pattern that matters most: Q1 FY27 (May) delivered revenue of $81.6B (+20% q/q, +85% y/y) and non-GAAP EPS of $1.87 vs. $1.77 consensus (a 5.4% beat, Data Center alone at $75.25B, +92% y/y) — and the stock still fell 3.1% the next day against a flat-to-up tape. That's four straight quarters of beating and fading; NVDA has beaten in 18 of the last 20 quarters and still dropped 5% after the February print. The options market is pricing a ~6-7% move (~$310-320B of market cap), which would be the largest single-day dollar swing on record for any stock post-earnings — the market is explicitly pricing a fat tail, not a shrug. Short interest is only 1.21-1.26% of float, so this isn't a squeeze setup in either direction; it's a pure positioning/expectations problem. China is its own sub-thread: H20 is now "green-zone" and flowing under case-by-case license review since a January 2026 rule change, but H200 shipments remain "trivial" despite roughly $10B in approved licenses (25% tariff, 50% volume cap, KYC requirements attached); Blackwell (B200, GB200 NVL72/36) still requires licenses that as of the last filing had not been granted, a loophole that let some Blackwell flow into China was closed in May, and Rubin looks locked out of China for 18-24 months post-launch regardless. The other side of the print is hyperscaler capex: sell-side 2026 capex estimates across five hyperscalers now sit near $697B (up $173B YTD), with Amazon guiding $200B, Alphabet raised to $180-190B, Meta raised to $125-145B (that raise alone knocked Meta down 9.25% on capex-fatigue fears), and Microsoft tracking above $120B for its fiscal year — roughly 75% of that capex is AI infrastructure. The strain: AI capex has gone from ~33% of hyperscaler operating cash flow in 2023 to an estimated ~93% in 2026, FY26 free cash flow is projected near zero or negative for everyone except Alphabet and Microsoft, and Wall Street estimates put $1.5T of tech-sector debt issuance over the next three years to fund it. That's the AI-ROI debate in one paragraph: NVDA's backlog language and Blackwell/next-gen ramp commentary Wednesday night is the only real-time evidence the market gets on whether that spending is translating into demand that justifies itself, or whether the funding math starts to bite first. Valuation context: forward P/E 21.42, PEG 0.48, stock ~9% below its 52-week high of $236.54 (low $164.07), trading just above its 200-day moving average (~$195) — not stretched on paper, but every quarter this year has proven that "not stretched" hasn't been sufficient to prevent a post-print selloff when the beat isn't emphatic enough.

CRWD ($~188, post 4:1 split, $209B mcap) — the agentic-pricing test case, already de-risked into the print. Consensus: revenue $1.44B (subscription revenue $1.37B, +24.3% y/y), non-GAAP EPS $0.29 — a setup where the Street is sitting almost exactly on the company's own numbers rather than above them. Last quarter beat both lines but the stock sank 13% after-hours on "modest guidance," and the stock has already fallen 16% in the five sessions into this print — some of the de-risking that usually happens after the fact has happened before it this time. Short interest is thin at 0.76%. The valuation bar is real: one preview pegged the multiple near 175x, "demanding a blowout, not just a beat." The specific thing to watch is credit-based/agentic pricing (Charlotte Agentic SOAR, Continuous Identity for AI Agents) — this is the cleanest per-agent-monetization data point in the whole cybersecurity complex, and it's the direct test of whether AI agents expand CrowdStrike's TAM (more things to secure) or whether agentic workflows compress the seat-based ARR base it was built on.

SNPS ($397.87, $76.2B mcap) — a name that's already been through its own reckoning. No independently confirmed consensus EPS/revenue figures surfaced in this research pass (flagged below). What's clear: the stock is down 33.81% over the past 52 weeks even as shares outstanding grew 17.34% y/y — consistent with dilution from the Ansys deal weighing on the name through integration. Forward P/E 25.07, short interest 2.74%. China EDA export exposure remains a known overhang for the semiconductor design-software complex generally; watch commentary on China revenue trajectory and Ansys integration cost/synergy tracking as the two swing factors, separate from whatever NVDA does five hours earlier the same evening.

HPQ ($29.47, $24.4B mcap) — the direct Micron read-through. Consensus: EPS roughly $0.66-0.72 (Zacks pegs $0.66, up against a guide of $0.61-0.71 with a $0.66 midpoint that was already slightly above the old consensus), revenue $14.2-14.6B. AI PCs were ~44% of shipments last quarter, up from ~35% the quarter before — the adoption curve is real. But this print lands directly on top of a DRAM/NAND cost spike that's arguably the least-discussed thread of the week: Mizuho flagged potential DRAM contract pricing up 355% and NAND up 510% in 2026, TrendForce logged 90-95% q/q DRAM contract price increases in a single quarter, and Samsung/SK Hynix/Micron have collectively shifted ~93% of production toward HBM for AI data centers, starving general-purpose RAM. Micron itself says it has no line of sight to when supply catches up. HPQ is a hardware margin business sitting directly downstream of that input-cost spike — this print is the first real look at whether PC/server OEMs can pass memory costs through or whether they eat it. Forward P/E ~10x, short interest a notably high 9.84% — a value/skeptic setup either way.

OKTA ($135.16, $23.3B mcap) — identity for AI agents, live debate. Consensus: EPS $0.96 (+5.5% y/y), revenue $792.1M (+8.8% y/y), cRPO estimate $2.51B vs. $2.27B a year ago. The stock is up nearly 60% year-to-date but just gave back 5.2% on competitive-AI concerns — the market is actively repricing whether AI-native entrants disintermediate identity management or whether "identity for agents" (a stated Okta strategic push) becomes a new growth vector layered on top of the legacy seat business. Forward P/E ~39-40x, short interest 3.85%. This print is as close as the sheet gets to a pure-play answer on the agents-vs-seats debate, from the security-infrastructure side rather than the CRM/software side.

CRM ($209.17, $171.3B mcap) — confirmed for 8/26, and it's the poster child for agentic-vs-seat compression. The instinct to double-check the date was right to have — Salesforce has drifted into the first week of September in some prior years — but this cycle the company explicitly set August 26 after the close as the Q2 FY2027 date, confirmed directly via investor relations. Consensus: EPS $3.09, revenue $11.31B. Forward P/E 15.01 (well below the software-industry median), stock down 15.13% over the past 52 weeks even as the S&P made highs — the market has already partially repriced Salesforce lower on Agentforce-versus-seat-compression fears ahead of the print. Short interest 3.59%, with short interest having roughly halved from the prior month's 50.64M shares — some of the bear thesis has already been covered into the report. The single metric that matters: Agentforce consumption/ARR disclosure and current remaining performance obligations, as the market's real-time gauge of whether usage-based agentic revenue is additive to or cannibalizing the core Sales/Service Cloud seat base.

NTNX ($55.04, $14.9-17.4B mcap) — a quieter infrastructure name riding the AI-adjacent hyperconverged wave. Consensus: EPS $0.48-0.49 (+29.7% y/y), revenue $737-738M (+12.9% y/y). Stock up 26% in the month heading into the print. Forward P/E 24.99, short interest an elevated 7.51%. Less of a direct AI-capex read than NVDA/HPQ, more a read on enterprise IT modernization budget health broadly — worth watching as a sentiment cross-check against the bigger infrastructure names reporting the same night.

The rest, by cluster

Quiet industrials/staples (no AI narrative, portfolio diversifiers): DCI (Donaldson) reports BMO — consensus EPS $1.13 (+9.7% y/y), revenue $1.04B (+7.2% y/y), forward multiple ~23-25x, short interest a low 1.48%; steady filtration compounder, dividend just declared at $0.32/share. SJM (J.M. Smucker) reports BMO — consensus EPS $2.21 (+16.3% y/y) on revenue roughly flat at $2.1B, earnings growth driven by mid-single-digit green coffee cost deflation offsetting low-single-digit inflation elsewhere in the portfolio; forward P/E 11.87, short interest 6.12%. Neither name touches the AI/rates thread directly — they're the low-beta ballast on an otherwise loaded tape.

Small-cap chip-adjacent and shipping: PLAB (Photronics) reports BMO — a photomask maker that already pre-guided Q3 revenue/EPS with its Q2 release; forward P/E 17.92, short interest 7.9% of float (up 41.6% y/y — meaningfully more contested than it was). A tangential, small read on semiconductor capacity utilization one rung down the supply chain from NVDA. SFL Corporation reports "preliminary" Q2 results (call at 10am ET/4pm CET) — consensus revenue ~$175.3M; a ship-leasing/income name with essentially zero correlation to the AI complex, more a global trade-flow and rate-environment read.

Life sciences/dev tools, after the close: Agilent (A) — consensus EPS $1.48 (+8% y/y), revenue $1.84B (+5.9% y/y), forward P/E ~20-22x, 52-week range $108.35-$160.27, short interest 1.85%; a life-sciences tools recovery story, unrelated to the AI thread but a read on biotech/pharma capex health. Veeva Systems (VEEV) — consensus EPS $2.22, revenue $904.1M, trailing P/E 44.69 (no forward figure surfaced), 52-week range $148.05-$310.50, short interest 1.58%; the stock is up 39.5% in the past month alone heading into the print, an aggressive run that raises the bar regardless of fundamentals.

The "P" resolution — Pure Storage, now Everpure Inc.: the bare ticker "P" on the sheet is not a data artifact. Pure Storage formally rebranded to Everpure, Inc. in February 2026 and retained its NYSE ticker "P" — it reports Q2 FY2026 results Wednesday 8/26 after the close. Price $108.55, market cap $36.08B, 52-week range $56.78-$119.10 (up nearly 95% over the past year, +46% in the last month alone — an aggressive AI-storage-demand run), trailing P/E 159.63x. No reliable forward EPS/revenue consensus surfaced in this pass. As an AI-data-infrastructure hardware name it sits in the same memory/component-cost crosscurrent as HPQ, and its stock momentum makes it one of the more extended names reporting all week.

Where the asymmetries sit (landscape, not advice)

NVDA's setup is unusual in that guidance and consensus have converged rather than consensus running ahead of guide — which mechanically raises the bar for what counts as a beat large enough to break the four-quarter pattern of "beat, then fade," while the options market is pricing the largest single-day dollar move in the stock's history either way. CRWD walks in already down 16% over five sessions with thin short interest — some of the typical pre-print de-risking has already happened, which cuts against a repeat of the post-Q1 -13% shock but doesn't eliminate valuation risk at a reported ~175x multiple. KSS carries 27.2% short interest into a report where the EPS estimate has already been revised up 40% in a month — a setup where positioning mechanics can dominate the actual print. HPQ and Pure Storage/Everpure sit on opposite sides of the same memory-cost spike: one is a margin-compression risk (HPQ, PC/server OEM eating higher DRAM/NAND costs), the other has run up on AI-storage-demand enthusiasm even as its own input costs rise. OKTA and CRM are both mid-repricing on the same underlying question — whether agentic AI is additive or substitutive to their existing revenue base — with the market having already moved CRM's multiple down and OKTA's up over the past several months ahead of getting real data Wednesday night.

Dominoes

NVDA's after-hours reaction is the read-through event for the whole complex Thursday morning: Marvell (MRVL) reports the very next evening, Thursday 8/27 AMC, not Wednesday — it will trade heavily off whatever tone NVDA sets on hyperscaler demand and custom-silicon competition. Super Micro (SMCI) is not on the sheet but trades on NVDA sympathy by habit; the company's internal investigation cleared CEO Charles Liang as of 8/20, but co-founder Wally Liaw remains under SDNY indictment (unsealed March 2026) over an alleged $2.5B scheme routing NVDA-chipped servers to China via a Southeast Asian front company — Liaw has resigned from the board, and the company itself was not charged, but the legal overhang is still live and will get relitigated in commentary around any China-related remarks NVDA makes Wednesday night. The broader memory-cost story (Micron, Samsung, SK Hynix) doesn't report this week but will be read through both HPQ's margin commentary and Pure Storage/Everpure's cost base. Power and grid capacity — the physical bottleneck behind the whole AI buildout — gets its most direct read from DY's fiber/infrastructure numbers Wednesday morning, hours before NVDA speaks to the compute side of the same buildout that night. And whatever tone NVDA sets carries directly into Friday's Jackson Hole session, where Warsh's rate-path signal will be read by a market that just spent a week absorbing an AI-capex verdict on top of a hawkish FOMC dissent.

Flagged as unverified

The exact Friday 8/21 close for the 10-year Treasury yield could not be independently confirmed in this pass — directionally it "snapped back" higher after a brief dip on Bessent's expanded bond-buyback announcement, but no precise print was verified; treat any specific 10-year level cited elsewhere with caution. July PCE is treated as landing Friday 8/28 per the majority of calendar sources and the customary BEA slot, but at least one aggregator this week claimed Wednesday 8/26 — worth an independent check of the exact BEA release date/time. Some earnings-calendar summaries grouped Workday and Intuit into "this week's tech earnings" framing without pinning an exact date; INTU is confirmed Tuesday 8/25 AMC and WDAY Thursday 8/27 AMC per the sheet and company releases. SJM's quoted spot price ($114.91, described as "+12.9%, +$13.14") appears in source material tied to a prior (June) post-earnings rally rather than a current spot quote — treat as approximate. WSM's cited "52-week" range ($240.41–$249.77) reads more like a recent multi-day trading band than a true 52-week high/low — treat cautiously. DY's quoted stock price ($437.76) is dated to July 2, 2026 and is likely stale by several weeks. Li Auto's cited market-cap figure (~$16.8B) conflicts with expected share-count math and should be treated as low-confidence. No consensus EPS/revenue figures for SNPS, PLAB, or Pure Storage/Everpure's Wednesday print surfaced with confidence in this research pass.

Sources: CNBC, Yahoo Finance, Reuters/Bloomberg via search aggregation, Nvidia/Salesforce/HP/CrowdStrike/Synopsys/Okta/Nutanix/Agilent/Veeva/Urban Outfitters/Abercrombie/Bath & Body Works/Kohl's/Williams-Sonoma/Dycom/Donaldson/J.M. Smucker/Photronics/SFL/Li Auto investor relations releases, StockAnalysis.com, WallStreetZen, TipRanks, MarketBeat, GuruFocus, Zacks/TradingView aggregation, CNBC/Fortune/CNN/Bloomberg reporting on the SMCI indictment, Mizuho/TrendForce/TechInsights on the memory market, Goldman Sachs/CreditSights/FactSet on hyperscaler capex.

Thursday 8/27 — Desk Preview

Prepared Sunday 8/23/26. Landscape only, no recommendations.

The frame

Friday 8/21 close: S&P 500 7,674.37 (+0.43%), Nasdaq Composite 26,180.45 (+0.43%), 10-yr yield 4.74%, VIX 15.13. A quiet, grinding-higher tape into a loaded week.

One macro-sequencing item is genuinely unsettled across calendars: some aggregators place the Q2 GDP second estimate and July PCE on Wednesday 8/26, while the customary BEA slots would put GDP Thursday 8/27 morning and PCE Friday 8/28 — verify against the BEA release schedule before trading around either print. Whichever slot they land in, Thursday itself is the NVDA-hangover session: the whole tape trades off Wednesday night's print (consensus $2.09 EPS, ~$92B revenue per Street estimates cited by Susquehanna), with Wednesday's monster software/hardware slate (CRM, CRWD, HPQ, SNPS, NTNX, OKTA, VEEV, Agilent) all digesting before Thursday's open. Layered under it: Jackson Hole's opening day. The Kansas City Fed's symposium runs Thursday 8/27–Saturday 8/29 under the theme "Financial Innovation: Implications for Payments and Policy," but Chair Kevin Warsh's actual keynote — his first as chair, ~3 weeks before the September FOMC — doesn't hit until Friday 8/28 at 10am ET. Thursday is positioning day, not verdict day, on Fed messaging.

Behind it: the July FOMC held 3.50–3.75% on a 9-3 vote, with Cleveland's Hammack, Minneapolis's Kashkari and Dallas's Logan all dissenting for a hike. Minutes released 8/19 showed the hawkish sentiment ran broader than the three dissenters — "policy tightening would likely be necessary if inflation did not decline" — and disclosed Warsh floated cutting the FOMC calendar from eight meetings to six. That's the backdrop for every rate-sensitive name reporting Thursday (AFRM's funding costs, TD's net interest margin, MBUU/mortgage-adjacent discretionary spend).

Thursday's slate is also, distinctly, two separate trades stapled together: a before-the-open discount/dollar-store and off-price cluster reading the trade-down consumer a week after Walmart/Target, and an after-the-close AI-capex-and-software cluster headlined by Marvell landing 24 hours after Nvidia — the market's cleanest read yet on whether custom silicon (Amazon Trainium, Microsoft Maia) is taking share from merchant GPUs or simply riding the same capex wave higher.

Tier 1 — before the open

BBY ($~90) — memory-cost inflation meets a consumer that's already tapped out. Consensus for Best Buy's fiscal Q2 sits near $1.37 EPS on $9.58B revenue. The stock trades at 13.3x forward earnings (16.2x trailing), sitting near the top of its 52-week range ($55.10–$91.27) after a big run — yet the Street's average 12-month target ($82.35) sits below the current price, an unusual setup where the tape has outrun the analyst community. The metric that matters this quarter isn't comps, it's pricing commentary on PCs, phones and TVs as DRAM/NAND spot prices spike on AI-driven memory demand — the same memory cycle that made Micron the barometer everyone's watching. If Best Buy flags device cost inflation getting passed to consumers into the holiday season, that's a read-through for the entire electronics/durables complex, not just BBY.

DG ($~mid-$90s-$100s, unconfirmed) — a beat that got sold before. Consensus $2.01 EPS / $11.19B revenue for Q2. Worth noting: last quarter (June 2, 2026) Dollar General beat estimates ($2.00 vs. $1.89 consensus) and the stock still sold off 3.3% the next session — a sign the market is trading guidance and margin trajectory, not the headline beat, on the "Back on Track" turnaround. Watch same-store traffic vs. ticket, and whether tariff-driven cost pressure shows up in gross margin guidance.

DLTR ($~$4.85B revenue consensus, $1.11–1.13 EPS) — the post-Family-Dollar story. Street sees revenue up 6.3% YoY and EPS up ~44% YoY, a growth rate that's largely a base-effect artifact of last year's Family Dollar divestiture reshaping the P&L. This is now effectively a pure-play Dollar Tree story: multi-price rollout, tariff mitigation via sourcing diversification, and whether the trade-down consumer that's showing up at Walmart and Costco is showing up here too.

BURL ($2.19 EPS consensus) — off-price's tariff hedge. Burlington's chase-inventory, closeout-sourcing model is structurally less exposed to tariff line-item risk than full-price retail, which is the bull case discretionary long/short desks lean on into every off-price print. Watch merchandise margin and inventory-to-sales trends for signs the trade-down tailwind (a direct read on BBY/DG/DLTR's same consumer) is still accelerating.

TD ($120.95, $199.8B mcap) — the AML-remediation arc, one quarter closer to resolution. TD trades at 16x trailing / 16.0x forward earnings, near the top of a 72.21–124.87 52-week range after a ~61% one-year return — the market has been re-rating the stock on hopes that the U.S. asset cap imposed after TD's 2024 guilty plea on Bank Secrecy Act failures (and the $3B+ in fines) is closer to lifting. Consensus sits near $1.77 EPS for the quarter. Notably, this week also brought a fresh AML headline: former TD employees were sentenced to prison for money-laundering/fraud involvement tied to the same case — a reminder the legal tail isn't fully closed even as the stock has re-rated hard. Watch commentary on the remediation program's cost and timeline, and any update on the U.S. retail banking growth constraint.

HRL ($0.35 EPS / $3.03B revenue consensus) — protein-input-cost story. Hormel's Q3 sits at the intersection of commodity meat costs (pork, turkey) and branded volume trends (Planters, SPAM). A low-growth, low-multiple name that mostly matters for cost-inflation read-through to the broader packaged-food complex.

HQY ($105.41, $8.81B mcap) — HSA custodian riding AI-as-efficiency, not AI-as-threat. HealthEquity trades at 21.6x forward earnings (39.5x trailing) near the top of its 52-week range (72.76–107.62), rated Strong Buy with a $118.53 average target (+12.4% implied). TTM EPS is up 94.8% YoY. The company has raised full-year guidance in each of the last two quarters and announced an "Agentic AI" member-experience initiative last November — a useful counter-example in the AI-agents debate: here agentic AI is framed as a cost-takeout/service-quality tool for HealthEquity itself, not a displacement threat to its business model the way it's framed for seat-based software. Watch whether the guidance-raise streak continues and HSA asset growth (last cited near $34B+) holds pace.

Tier 1 — after the close

MRVL ($0.93 EPS / ~$2.70–2.71B revenue consensus) — the AI-capex derivative print, 24 hours after Nvidia. This is the header name of the week for anyone trading the AI-capex debate. Marvell is up roughly 180% year-to-date through 8/20 versus Nvidia's ~17% — a gap that reflects the market pricing Marvell as the cleanest way to own the "custom ASIC is taking incremental share from merchant GPU" thesis (Amazon Trainium, Microsoft's custom silicon programs, XPU-Attach expansion, electro-optics/interconnect). But Marvell's own reaction history is a cautionary tale on guidance risk: last quarter (May 27, 2026) it delivered an in-line beat and the stock rose a modest 3.1%; the quarter before that (March 5, 2026) a beat drove an 18.4% pop; but a year ago (Aug 28, 2025) an in-line print on a disappointing guide sent the stock down 18.6% in a single session. That asymmetry — modest upside on a beat, violent downside on any guidance wobble — is the single most important thing to understand about how this stock trades on earnings. Stifel models Q2 revenue above the Street's $2.70B (implying 11.7% sequential growth) and looks for a fiscal Q3 guide above their own $3.0B estimate; they carry a Buy and $350 target. The read here lands directly against Wednesday night's Nvidia numbers — if Marvell's data-center/custom-silicon commentary corroborates hyperscaler capex intentions Nvidia laid out 24 hours earlier, that's confirmation; any divergence gets picked apart immediately.

WDAY ($2.61 EPS consensus) — the core test case for AI-agents-vs-seat-based software. Workday is the cleanest read on whether agentic AI is a net threat or a net opportunity to per-seat HR/finance SaaS pricing. The bear case: if AI agents can execute HR/finance workflows directly, the seat-based licensing model that underwrites Workday's economics gets structurally pressured over time. The bull case: Workday's own "Illuminate" AI-agent monetization layer becomes incremental ACV on top of the existing seat base rather than a substitute for it. Subscription revenue growth has been decelerating from the 30%+ era to a more mature low-double-digit pace, so current remaining performance obligations (cRPO) — the forward bookings proxy — is the number that actually moves the stock, more than the headline EPS beat/miss. Guidance for the balance of FY27 is the other half of the setup.

ADSK ($3.12 EPS consensus) — subscription model, lower agentic-displacement risk. Autodesk sits adjacent to the WDAY debate but with a different risk profile: CAD/engineering design workflows are less commoditized by chat-style AI agents than HR/finance data entry, so the "AI eats the seat" bear case is less immediate here. The bigger swing factor is the ongoing transition to Autodesk's new transaction/licensing model and whether billings and net revenue retention hold up.

ULTA ($6.19 EPS consensus) — prestige beauty's resilience test. Ulta reports in the same discount/consumer-trade-down window as BBY/DG/DLTR/BURL/GAP, but sits at the opposite end of the spending spectrum — prestige beauty has proven more resilient than discretionary hardlines in this cycle. Watch comps, loyalty member growth, and competitive commentary on Sephora-at-Kohl's and Amazon's beauty push.

AFRM ($0.35 EPS consensus, Q4 FY26) — consumer credit and GMV under a still-hawkish Fed. Last quarter (May 7, 2026) Affirm blew through estimates ($0.30 actual vs. $0.17 consensus) — yet the stock still fell nearly 5% the next session, evidence the market is trading GMV growth, delinquency trends and funding costs ahead of EPS. With three FOMC members on record wanting a hike in July, Affirm's securitization-funding cost structure and 0%-APR loan mix are the numbers that matter more than the bottom line. Rated Strong Buy by 18 covering analysts heading in.

S ($0.07 EPS consensus) — security agents, part one of two. SentinelOne reports alongside Rubrik in the "security agents" pairing the desk is watching: agentic SOC/AI-driven detection-and-response as the next leg of cybersecurity spend, layered on top of legacy EDR. ARR growth deceleration and competitive positioning against CrowdStrike (which reports the night before, Wednesday 8/26) are the framing questions.

RBRK ($0.04 EPS consensus) — security agents, part two. Rubrik's data-security/cyber-resilience pitch leans on AI-driven ransomware recovery as a differentiator. Still early-profitability (EPS barely above breakeven), so ARR growth and net new logo commentary carry more weight than the EPS line itself.

The rest, by cluster

Discount/off-price/consumer trade-down (the week-after-WMT/TGT read): GAP ($0.48 EPS / $3.7B revenue consensus, implying a ~0.6% revenue decline YoY) rounds out the retail cluster — watch Old Navy/Gap brand comps and tariff-cost commentary. CSIQ (Canadian Solar, -$0.47 EPS / $1.17B revenue consensus) is the odd one out geographically but reports in the same window — the story here is entirely solar policy: tariff exposure on Chinese-linked manufacturing, IRA subsidy trajectory, and module pricing, not consumer trade-down.

Smaller software/SaaS: ESTC (Elastic, $0.58 EPS consensus, FQ1) — last quarter's beat (0.61 vs. 0.56 est) drove a 12.3% pop, so the bar carries real weight. PD (PagerDuty, ~$0.31 EPS consensus) — smaller-cap incident-response name, thinly covered relative to the mega-cap names reporting the same night.

China/international: BILI (Bilibili, $1.51 EPS consensus) — Chinese video/gaming platform, a read on Chinese consumer ad spend and gaming monetization independent of the U.S. macro story.

Discretionary/marine: MBUU (Malibu Boats, $0.78 EPS consensus, Q4/full-year) — small-cap, rate-sensitive discretionary recreation name.

Thin-coverage/low-visibility: FPS (Forgent Power Solutions, $0.24 EPS consensus) — limited analyst coverage and sparse public data; treat estimates here as low-confidence.

Where the asymmetries sit (landscape, not advice)

• MRVL's reaction history is genuinely bimodal — in-line beats have produced anywhere from +3% to +18%, but a disappointing guide a year ago produced -18.6% in a single session. The options market has historically priced double-digit implied moves into MRVL prints (an ~11% swing was priced ahead of the March 2026 report per trader commentary); expect similarly elevated implied vol into this one given it follows NVDA by one session.
• DG's pattern of beating and still selling off (most recently -3.3% on a beat in June) suggests the market has already priced the EPS line and is trading guidance/margin quality instead — worth knowing going in given consensus already sits at $2.01.
• AFRM did the same thing in reverse — a blowout beat (0.30 vs 0.17) that still fell ~5% — reinforcing that GMV/credit-quality/funding-cost commentary is driving these prints more than the reported number.
• TD's re-rating (52-week range 72.21–124.87, +61% one-year return) has run well ahead of the AML remediation actually being resolved — this week's prison sentencing of former employees on the underlying case is a reminder the legal overhang isn't fully closed even as the multiple has expanded.
• BBY's average analyst price target ($82.35) sits below the current price (~$90) — an unusual divergence between where the stock trades and where the sell-side has it modeled, worth noting heading into a print that's also a bellwether for memory-cost pass-through across consumer electronics.

Dominoes

• Wednesday 8/26, before Thursday's open: NVDA earnings after close ($2.09 EPS / ~$92B revenue consensus), plus CRM, CRWD, HPQ, SNPS, NTNX, OKTA, VEEV, Agilent, ANF, KSS, WSM, SJM, MCFT all report Wednesday — Thursday's entire tape opens already reacting to all of it (possibly plus GDP/PCE if the Wednesday-release calendars are right; verify).
• Thursday 8/27: Jackson Hole symposium opens (runs through Saturday); weekly jobless claims, and possibly the Q2 GDP second estimate if the customary Thursday slot holds.
• Friday 8/28: Warsh's first Jackson Hole keynote as Fed chair, 10am ET — the actual Fed-messaging catalyst for the week, landing the morning after Thursday's earnings deluge digests. July PCE also lands Friday per most calendars.
• Following Thursday, 9/3: DOO (BRP) and lululemon (LULU) both report — see flagged items below on why they're not this week's story despite appearing on some calendars as Thursday 8/27 names.

Flagged as unverified

• GDP second estimate / July PCE dates: Sources genuinely conflict — one aggregator pass pointed to Wednesday 8/26 for both, while most calendars and the customary BEA slots put GDP Thursday 8/27 and PCE Friday 8/28. Verify against the BEA calendar directly before trading around these.
• DOO (BRP): The sheet lists BRP for Thursday 8/27, and one earnings-calendar aggregator (Kiplinger, sourced from Briefing.com) still shows it there — but BRP's own investor-relations press release and a second independent tracker (WallStreetZen) both confirm Q2 FY27 results land Thursday, September 3, 2026, a full week later. Treat DOO as not reporting this Thursday.
• MCFT: Listed as TBD on the sheet; resolves to Wednesday 8/26 before the open, not Thursday.
• FPS (Forgent Power Solutions): Confirmed for Thursday 8/27, but this is a thinly-covered, low-visibility name — consensus figures should be treated as low-confidence.
• lululemon (LULU): Appeared in at least one search result as an Aug 27 reporter, but the company's own press release confirms September 3, 2026 — it is genuinely not on this Thursday's slate; flagged in case it shows up on other calendars as a Thursday name.
• TD's consensus EPS ($1.77): Sourced from a single aggregator; Toronto-Dominion reports in CAD and some estimate feeds mix currency bases — confirm the currency convention before using this figure directly.
• Price/market-cap/short-interest/52-week-range gaps: Full stats were confirmed for BBY, TD, HQY, MRVL and NVDA (context only); for the remaining sheet names, consensus EPS/revenue figures are reasonably well sourced (Kiplinger/Refinitiv, TipRanks, company press releases) but price, market cap, forward P/E, short interest and 52-week range could not all be independently verified this cycle — those fields are left blank in the calendar tables rather than estimated.
• Options-implied move and short interest %: Not independently verified for any name this cycle beyond the qualitative MRVL history cited above; treat all implied-move commentary as directional, not quoted.

Sources: Kansas City Fed, Federal Reserve press releases (FOMC minutes 8/19), BEA release schedule, CNBC/AP market wraps (8/21 close), Kiplinger earnings calendar (Briefing.com/Refinitiv/S&P Global sourced), TipRanks, StockAnalysis.com, WallStreetZen, Yahoo Finance, company investor-relations press releases (Marvell, Workday, Autodesk, Ulta, Affirm, SentinelOne, Elastic, Gap, PagerDuty, Rubrik, Best Buy, Dollar General, Dollar Tree, Burlington, TD, Hormel, HealthEquity, Malibu Boats, Canadian Solar, Bilibili, BRP, lululemon, MasterCraft).

Friday 8/28 — Desk Preview

Prepared Sunday 8/23/26. Landscape only, no recommendations.

The frame

Friday is a macro day wearing an earnings-calendar costume. The Tier 1 slate is thin — Frontline is the only confirmed pre-open reporter, and a sweep of eWhispers/eco calendars found no other Friday 8/28 additions worth flagging (Alibaba, which reported late August the last two years, already printed Thursday 8/20 this cycle and doesn't repeat — its next report isn't until December 1). So the tape's center of gravity Friday is entirely macro: Kevin Warsh's first Jackson Hole keynote as Fed chair at 10:00am ET, July PCE at 8:30am ET ninety minutes earlier (the customary last-Friday BEA slot; one aggregator this week claimed Wednesday 8/26 — verify), and whatever verdict the market has already reached on AI capex durability from NVDA (Wed night, 8/26) and MRVL (Thu night, 8/27).

The setup into it is more contested than the "coin-flip cut odds" framing from mid-August suggests. The July FOMC held 3.50–3.75% but split 9-3, with Cleveland's Hammack, Minneapolis's Kashkari and Dallas's Logan all dissenting in favor of a 25bp hike — the most divided vote since September 2016 — and the August 19 minutes showed the broader committee saw a case for hiking if inflation doesn't cool. That dissent was forced by the Strait of Hormuz standoff: U.S./Iran tensions have cut transit volumes to a fraction of pre-conflict levels, pushed WTI into the high-$80s and Brent toward $90, and driven July CPI energy prices up 14.7% y/y even as headline CPI ticked down to 3.4% and core to 2.5%, and PPI came in flat m/m (below the 0.2% consensus). That's the tension Warsh has to thread Friday: energy-driven headline pressure against a core trend that's still decelerating. September 16 odds have accordingly whipsawed hard — from roughly 82% hike-priced in mid-July down toward the 30-40% range by August 21-22 in some feeds, while other trackers show cut odds re-firming toward the mid-80s over the same window. The dispersion across CME FedWatch, Polymarket and Kalshi as of this weekend is wide enough that it's a genuine three-way live debate (hike / hold / cut), not the two-way coin flip the desk may still be carrying from two weeks ago.

Warsh's own pattern since being sworn in in May — shortened post-meeting statements, evasive pressers, explicit skepticism toward forward guidance as a policy tool — argues he uses the "Financial Innovation: Implications for Payments and Policy" theme to talk framework rather than preview September. That itself is a live risk: a market braced for signal that gets none is its own kind of repricing event. Layer on that NVDA and MRVL will have reported by the time Warsh speaks, and Friday becomes a stack of catalysts rather than a single one — whatever verdict AI capex delivers Wednesday/Thursday night is still being digested when the PCE print and the Fed chair speech hit. Month-end rebalancing flows land the following Monday (8/31), arriving right on top of however the week resolves. Coming into this, the tape had posted back-to-back weekly losses on bond-market volatility despite Friday 8/21's bounce (S&P +0.4%, Dow +1.0%, VIX down to the low-15s) — so there's real two-sided positioning risk heading into a data- and speech-heavy Friday.

Tier 1 — before the open

FRO ($44.45) — Frontline reports Q2 with tanker rates in a different universe than when guidance was set. Q1 was a record quarter (revenue $714.2M, net profit $559.1M, adjusted EPS $1.55, dividend $1.55/share) on VLCC TCE of $103,500/day and Suezmax at $72,400/day. Q2 bookings guided in May were already a big step up — 82% of VLCC days fixed at $181,700/day, 79% of Suezmax at $131,300/day — but spot rates have gone parabolic since on the Hormuz disruption: reports of TD3C (Middle East Gulf–China) earnings above $520,000/day, average VLCC prints north of $170,000/day, Suezmax above $200,000/day, with ship-to-ship transfer activity off Oman and Fujairah surging as owners route around the chokepoint. Consensus sits at $2.58 EPS on $754.6M revenue — a number that was set before the most recent leg of the rate spike and may already be stale to the upside. Frontline's dividend is tied to quarterly earnings (payout historically run close to 100% of adjusted net income), so a beat plausibly means another sequential dividend step-up from $1.55. The offsetting dynamic: shadow-fleet sanctions are doing real work on effective compliant-tonnage supply (roughly 700 vessels on international sanctions lists, about half no longer shipping, per recent tallies), which structurally supports rates for compliant owners like Frontline — but that support is a geopolitical variable, not a structural one. The stock is already trading near its 52-week high (~$45.17) and up sharply YTD, so the print lands against a market that's priced in a lot of the good news; the asymmetry is whether the quarter and forward commentary can outrun a rate environment that could mean-revert fast if there's any Hormuz de-escalation or an Iran-Oman transit agreement. Watch for management's characterization of how much of the current spot spike is sustainable into Q3 versus a transitory chokepoint premium.

The rest, by cluster

No other confirmed reporters before Friday's open on the sheet or in the calendar sweep. The Thursday-night MRVL print (consensus $2.71B revenue, $0.93 EPS, ~35%/39% y/y growth, options pricing roughly a 14% move) and Wednesday-night NVDA print (consensus ~$91.7B revenue against $91B ±2% guidance, ~$2.08 non-GAAP EPS, ~80% y/y growth) will both be fully digested by Friday morning — their aftermath is a macro input to Friday's tape rather than a Friday-dated event itself. Micron isn't reporting this week either, but its last print extended a multi-quarter beat streak on an HBM/DRAM supply squeeze that Goldman has characterized as the tightest DRAM shortage in roughly 15 years, with 2026 HBM supply reportedly sold out — it remains the cleanest read on whether AI capex is showing up as durable end demand or front-loaded ordering, and it's a name likely to get cited on both the NVDA/MRVL calls and in any post-print commentary Friday.

Where the asymmetries sit (landscape, not advice)

On FRO: consensus was set against a rate structure that's since moved sharply higher, but the stock is also near 52-week highs pricing that move — the setup cuts both ways depending on how forward-looking management's Q3 commentary is and whether the desk treats the current rate spike as durable or as chokepoint-premium that fades with any Hormuz resolution. On the macro leg: a market that's been whipsawing between hike, hold and cut framings for September has more room to move on a decisive signal than one that's already converged on consensus — but Warsh's demonstrated preference for saying little raises the odds of a "non-event" outcome that leaves the September debate exactly where it started, which is its own kind of dislocation for anyone positioned for resolution. The PCE print at 8:30am, ahead of the 10:00am speech, sets the data backdrop Warsh will be reacting to (or pointedly not reacting to) in real time.

Dominoes

• FRO's print and commentary read through to the broader tanker/shipping complex (other crude tanker owners) and to freight-cost inputs feeding PPI/goods inflation.
• A hawkish Warsh read hits duration broadly — mega-cap tech, small caps, rate-sensitive REITs — while a dovish or no-signal read reopens September cut pricing and could unwind some of the recent yield backup (10-yr closed 8/21 at 4.738%).
• July PCE either confirms or complicates whatever Warsh signals 90 minutes later; a hot print paired with hawkish framing compounds, a cool print paired with any dovish lean could reprice cut odds sharply in a single session.
• NVDA/MRVL verdicts spill into the semis complex (AMD, AVGO, TSM), power/grid names tied to data-center buildout, and the Micron/memory read-through — all of it already baked into the tape by the time Friday's macro catalysts hit.
• Month-end flows on Monday 8/31 land directly on top of wherever Friday's close settles, with potential to amplify or fade the week's net move.

Flagged as unverified

• September FOMC probabilities are genuinely inconsistent across trackers as of 8/21–8/22 (readings ranged from ~30% hike odds to ~85% cut odds depending on source/timestamp) — confirm live via CME FedWatch immediately pre-print rather than relying on any single cited number here.
• July PCE headline/core consensus figures were not independently confirmed this session; the PCE read-through above is inferred from July CPI (3.4% headline / 2.5% core) and PPI (flat m/m, 4.7%/4.2% y/y) rather than a sourced consensus estimate. The PCE release date itself is reported inconsistently this week (Friday 8/28 per most calendars vs. Wednesday 8/26 per one aggregator) — verify against BEA.
• FRO's forward P/E carries conflicting reads across providers (~6.4x forward vs. ~10.7x trailing) — treat as approximate.
• FRO's 52-week range varies by source (roughly $19.87–$45.17 vs. $18.43–$43.10 vs. $16.25–$40.95 across different platforms) — directionally the stock is near highs; exact bounds are soft.
• A Micron quarterly-revenue figure surfaced in research this session was implausibly large relative to its known run-rate and is deliberately not cited above — treat any specific MU revenue number from this cycle as unverified until confirmed directly.

Sources: GlobeNewswire/Frontline IR, TipRanks, techtimes.com, MNI Markets, CNBC, Al Jazeera, Bloomberg, stockanalysis.com, Simply Wall St, marketchameleon.com, gurufocus.com, Lloyd's List, S&P Global, investinglive.com, tradingkey.com.