US Pre-Market Brief | 2026-08-12 (ET) Wednesday

Coverage window: 2026-08-11 16:00 ET regular-session close → 2026-08-12 08:10 ET (includes 8/11 after-hours earnings, the overnight Asia/Europe session, and this morning's pre-market); Section 9.0 additionally covers the July CPI released at 08:30 ET and the market reaction within 3 minutes of the release (appended through 08:33 ET). Quote conventions:

• Prior close / pre-market price / pre-market change / pre-market volume / 52-week high-low for single stocks and ETFs come from the stockanalysis.com quote API, snapshots 07:43–08:10 ET (main snapshot 08:10, tagged per ticker). "Prior close" is the 8/11 16:00 ET regular-session closing price; "today's pre-market %" is the pre-market change relative to that prior close, and each ticker's snapshot time is tagged in the tables.
• Treasury yields, VIX, crude, gold, the dollar index and bitcoin come from the CNBC quote API, snapshots 08:07–08:08 ET.
• Earnings figures are taken preferentially from SEC EDGAR originals and company press releases (the specific filing is tagged in the tables); where the original was unobtainable, authoritative media reprints are used and noted.
• Overnight Asia/Europe index levels are media-sourced (Associated Press), not raw exchange closing data. Data this brief did not obtain: ① VIX9D at the time of this snapshot was still 8/11's 12.52 and had not updated to today's value, so this brief does not use it for any judgment about today (why that matters is spelled out below); ② Nebius's (NBIS) earnings press release contains no forward guidance, and as of writing its Q2 6-K had not yet appeared on EDGAR; the CEO's quarterly shareholder letter was in fact published at 07:38 ET alongside the results (nebius.com/investor-hub) — this round of data collection simply failed to capture it, it was not unavailable — noted here as a correction so readers do not conclude the information does not exist. On comparison, the full-year guidance in it is a reiteration of last quarter's figures (see Section 5.1), so this brief still does not treat it as a new catalyst today; NBIS's end-of-Q2 ARR was not obtained by this brief (end-of-Q1 was $1.92 billion), and ARR is the central pillar of the company's full-year guidance; the first-hand transcript of CoreWeave's earnings call was not obtained by this brief, so all call-sourced guidance in the text (Q3 revenue/interest/capex, FY26 revenue and capex, year-end ARR) is media-relayed and has been tagged "pending verification" item by item; Lumentum's FY26 full-year cash flow statement was not obtained by this brief (not attached to the 8/11 press release), so its operating cash flow and free cash flow are left blank; Coherent's sell-side consensus figure is questionable (the circulating $1.43 is below the $1.52 low end of the company's own 5/6 guidance), so this brief does not use it as the bar to clear; ③ Option implied volatility comes from the CBOE delayed-quote API, snapshots 07:35–08:08 ET, and implied move data was obtained for three names — COHR, LITE, SMCI (see Section 8); option data for the remaining tickers was not obtained, so volatility judgments about those names in the text are inferred from price action, not from option data; ④ July CPI was released at 08:30 ET, after the main snapshot time — the scenario mapping was written before the release (08:19 ET), and the actual result and market reaction have been appended in Section 9.0, with the original order of the two passages preserved.

---

0. Today in One Sentence

Today is a "strong pre-market ahead of an event": SPY +0.25% pre-market, QQQ +0.66%, while the semiconductor ETF (SMH) is +1.69% and the software ETF (IGV) is −0.99% — what is up and what is down is split cleanly along one line, and that line is "selling shovels" versus "using shovels." The strongest catalysts come from two AI cloud earnings reports: CoreWeave (CRWV) +18.49% pre-market, with an adjusted operating margin of 5% blowing through the market's 2.7% expectation; Nebius (NBIS) +15.5% pre-market, revenue of $582.3M up +454% YoY, adjusted EBITDA swinging from last year's −$21.0M to +$236.2M. The second main line is optical communications: Lumentum's (LITE) Q1'27 guidance of EPS $4.05–4.35 runs straight over the $3.63 consensus, taking it +8.24% pre-market and dragging the whole chain up with it — COHR (+5.64%), CIEN (+4.90%), AAOI (+3.86%), GLW (+3.78%) (FN nominally +5.41%, but on only 47 shares traded, which is not a usable reference) — but one thing needs saying plainly: LITE's pre-market price this morning merely returns it to its August 7 level. The true starting point of this entire move was the 8/4 policy news that "the FCC plans to ban Chinese-made optical modules"; the earnings report's function is to "confirm the policy narrative has not been falsified by fundamentals," not to supply new upside.

The driver type is "earnings + guidance," not macro — but today's genuine first-order event is the July CPI at 08:30 ET, and its risk direction is the opposite of the past two years: the market currently prices September FOMC as roughly a coin flip between a hike and no move, not a cut. [08:33 ET addendum] CPI is out and came in fully in line: headline +3.4% YoY (prior 3.5%), core +2.5% YoY (prior 2.6%), core +0.2% MoM. The market reaction is "risk cleared," not "direction reversed" — 10Y barely moved, VIX flipped from up to down at 15.20, SPY/QQQ strengthened, while the divergence between SMH (+2.19%) and IGV (−1.08%) actually widened. Macro did not take the pricing power; the main lines remain earnings and themes. Pre-market 10Y Treasury 4.662% (−2.2bp), VIX 15.38 (+0.65%), Brent $89.20 — after the US military fired on a Panama-flagged cargo ship attempting to break the blockade of Iranian ports, oil is waiting just below the $90 handle, the second asymmetric risk source besides CPI.

---

1. News Overview

Pre-market broad market and cross-asset (snapshot 08:07–08:10 ET)

Overnight Asia/Europe (media-sourced): Nikkei 225 +0.8% to 67,524.06; Korea's KOSPI +3.7% to 6,579.04 (strongest globally); Hang Seng −0.8% to 25,440.17; Germany's DAX +0.3% to 26,480.04; France's CAC 40 −0.2% to 8,698.79. KOSPI +3.7% is the single most noteworthy overnight item — Korea is the market with the highest memory and semiconductor weighting, and it and this morning's SMH +1.69% in the US are two ends of the same thing.

---

2. Strongest Themes, Descending

---

3. Overall Single-Stock Strength Table

Sorted by total score, descending. "Today's pre-market %" is the pre-market change relative to the 8/11 close; snapshot times are tagged per ticker, earliest 07:43, latest 08:10 ET; readings more than 5 minutes away from the main snapshot (08:10) are noted in the table.

3.1 Long Side

| Rank | Ticker | Name | Theme | Direction | Bullish level | Total | Core news | Catalyst directness | Fundamentals | Expectation gap | Pre-market action | Main risk | Conclusion | |---|---|---|---|---|---|---|---|---|---|---|---|---| | 1 | NBIS | Nebius Group | AI cloud compute | Bullish | S | 76 | Q2 group revenue $582.3M (+454%, above the $569.9M consensus), adjusted EBITDA swung from −$21.0M to +$236.2M | Very high (own earnings) | Net obligations including leases only about $2 billion; convertible coupons of 1.25%/2.625% | High (EBITDA beat) | +15.5% @ $223.18 (08:03, single source; this field is intermittently missing; another independent source showed +12.5% at the same time), pre-market volume 2.08 million shares | All six circulating guidance items are May-vintage; operating cash flow depends almost entirely on customer prepayments; already gapped up 15% | Priority deep-dive | | 2 | LITE | Lumentum | Optical comms | Bullish | S | 74 | FQ1'27 guidance EPS $4.05–4.35 vs est. $3.63; gross margin above 50% for the first time | Very high (own guidance) | Revenue +109%, gross margin structurally stepping up | High (guidance midpoint about 15% above consensus) | +8.24% @ $888.19 (08:10) | Enormous one-year gain, high valuation; already gapped up pre-market | Priority deep-dive | | 3 | CRWV | CoreWeave | AI cloud compute | Bullish | S | 68 | Q2 revenue $2,575M (+112%), adjusted operating margin 5% vs est. 2.7% (consensus is media-relayed) | Very high (own earnings) | Backlog $104 billion, contracted power 3.7GW; but lease obligations included, total $51.6 billion, equity buffer only 6.5% | High (a real expectation gap on the margin dimension) | +18.49% @ $107.02 (08:10), pre-market volume 3.27 million shares | $10.6 billion maturing within 18 months against $5.5 billion of cash; adjusted net loss margin −11%→−22%; already gapped up 18% | Priority deep-dive (position is elevated) | | 4 | EAT | Brinker | Restaurants | Bullish | A | 63 | FY27 guidance EPS $12.60–13.40; new $750 million buyback authorization | Very high | FY26 EPS $10.87, five consecutive years of positive same-store growth | Medium | +2.99% @ $228.00 (08:09) | Consumer resilience in question; includes a 53rd week | Watch closely | | 5 | HRB | H&R Block | Consumer services | Bullish | B+ | 62 | EPS $2.38 vs $2.21; dividend +10%; FCF $756 million | Very high | Best results in five years, buybacks and dividends together | Medium | +12.41% @ $52.46 (08:06) | Low-growth business; already gapped up 12% pre-market | Watch closely | | 6 | CAVA | CAVA Group | Restaurants | Bullish | A | 61 | Same-store +9% vs est. +7.1%, traffic +5.3% | Very high | Revenue +31.3%, traffic-driven rather than pure price | Medium-high | +18.02% @ $71.77 (08:10) | Full-year guidance only maintained, not raised; already gapped up 18% | Watch closely | | 7 | SMCI | Super Micro | AI servers | Bullish but questionable | B+ | 56 | FY27 revenue guidance $65–72 billion (FY26 $39.1 billion) | Very high | Gross margin 17.5%, net debt about $1.2 billion | Low (core data pre-announced on 7/21) | +8.54% @ $34.30 (08:10) | Q1 guidance implies margins cut nearly in half; export-control independent investigation unresolved | Watch only | | 8 | CSCO | Cisco | Networking | TBD | A | 55 | Reports FQ4'26 tonight; consensus EPS $1.17, revenue about $16.8 billion; AI order target already raised to $9 billion | Medium (event has not happened) | Full-year AI order target raised from $5 billion to $9 billion | Low (pre-event) | +2.03% @ $122.88 (08:10) | Options imply about 8.3% move; already up pre-market | Watch only (pre-event) | | 9 | VRT | Vertiv | AI power | Bullish | A | 53 | Spillover from CRWV's power expansion; already +4.34% yesterday | Medium (derivative) | Leader in data center thermal management | Low | +4.22% @ $293.70 (08:10) | Gapped up two days running; no catalyst of its own | Watch closely | | 10 | MRVL | Marvell | AI chips | Bullish | B+ | 52 | No disclosure of its own today; AI memory platform + Microsoft Maia custom chain continues to develop | Low (no catalyst of its own) | Custom ASIC and interconnect positioning | Low | +3.55% @ $219.85 (08:10), pre-market volume 4.15 million shares (the largest in this brief's sample of 30-odd names) | Huge volume with no news today — note it may be a passive follower | Watch closely | | 11 | COHR | Coherent | Optical comms | Bullish but high risk | A | 50 | Spillover from LITE's print; reports its own results tonight, est. EPS $1.43 (+93.2%) | Medium (spillover) | One of the two optical component leaders | Negative (LITE has raised the bar) | +5.64% @ $347.11 (08:10) | Earnings tonight + already up 5.6% pre-market + expectations raised | Watch only | | 12 | CIEN | Ciena | Optical comms | Bullish | B+ | 49 | LITE spillover | Medium | Optical transport systems | Low | +4.90% @ $406.51 (08:10) | Pure spillover, no catalyst of its own | Watch closely | | 13 | FN | Fabrinet | Optical comms | Bullish | B+ | 48 | LITE spillover (FN is LITE's main contract manufacturer) | Medium-high (direct supply chain) | Optical module contract manufacturing leader | Low | +5.41% (only 47 shares traded, not a usable reference) | The quote is not a usable reference; this brief uses its gain as evidence for nothing | Watch closely | | 14 | IREN | IREN | AI compute | Bullish | B | 47 | AI cloud theme spillover | Low | Bitcoin miner converting to AI hosting | Low | +7.50% @ $42.73 (08:10) | High-beta momentum-follower name | Watch only | | 15 | GLW | Corning | Optical comms | Bullish | B | 46 | LITE spillover (fiber and optical connectivity) | Medium | Optical fiber and glass substrates | Low | +3.78% @ $165.21 (08:10) | Pure spillover | Watch closely | | 16 | APLD | Applied Digital | AI compute | Bullish | B | 45 | AI cloud theme spillover | Low | Data center hosting | Low | +7.41% @ $31.89 (08:10) | High-beta follower | Watch only | | 17 | HYLN | Hyliion | Distributed power | Bullish | C | 44 | Raised full-year revenue guidance to $115–120 million; Navy KARNO contract | Very high (own) | Extremely small revenue base, not profitable | Medium | +25.77% @ $4.93 (08:09) | Small cap, high volatility, already gapped up 26% | Watch only | | 18 | AAOI | Applied Optoelectronics | Optical comms | Bullish | C+ | 43 | LITE spillover | Low | Earnings quality weaker than peers | Low | +3.86% @ $139.51 (08:10) | Weakest fundamentals in the chain | Watch only | | 19 | CBRS | Cerebras | AI chips | TBD | A | 42 | Reports tonight; Q2 core revenue guidance about $194 million (+88%); est. loss per share $0.17 | Medium (event has not happened) | OpenAI 750MW / multi-year agreement worth over $20 billion | Low (pre-event) | +2.23% @ $240.00 (08:09) | Down 27.1% since listing; event risk tonight | Watch only (pre-event) |

3.2 Short / Avoid Side

---

4. Single-Stock Scoring Model (100 points)

| Dimension | Weight | NBIS | LITE | CRWV | EAT | HRB | CAVA | SMCI | CSCO | COHR | |---|---|---|---|---|---|---|---|---|---|---|---| | News source authority | 0–15 | 15 | 14 | 15 | 14 | 14 | 14 | 15 | 10 | 10 | | Catalyst directness | 0–20 | 20 | 20 | 20 | 20 | 20 | 20 | 20 | 12 | 12 | | Earnings elasticity | 0–15 | 14 | 14 | 12 | 8 | 9 | 10 | 8 | 10 | 11 | | Moat and fundamentals | 0–15 | 12 | 11 | 7 | 11 | 11 | 10 | 7 | 12 | 10 | | Expectation gap | 0–10 | 8 | 7 | 8 | 5 | 6 | 7 | 5 | 4 | 3 | | Catalyst persistence | 0–10 | 8 | 9 | 9 | 5 | 4 | 5 | 6 | 7 | 8 | | Trading characteristics | 0–10 | 8 | 7 | 9 | 6 | 6 | 7 | 9 | 9 | 8 | | Risk deduction | 0~−15 | −10 | −8 | −12 | −6 | −8 | −12 | −14 | −9 | −12 | | Total | 100 | 76 | 74 | 68 | 63 | 62 | 61 | 56 | 55 | 50 |

Scoring note one (why SMCI gets only 56 and COHR only 50): neither deduction comes from "the business is bad," but from the expectation gap having already been consumed in advance. SMCI's gross margin and order figures were pre-announced on 7/21 (see Section 5.4), and COHR's bar was raised by LITE's beat last night — both are structures where "the good news is already in the price and the bad news is not yet priced."

Scoring note two (CRWV cut from 72 to 68, NBIS from 78 to 76): both cuts happened after re-checking against first-hand disclosure documents, not because of a change of view. CRWV was cut because the 10-Q disclosed this morning shows a balance sheet tighter than the market's standard framing (lease obligations included at $51.6 billion, equity/assets only 6.5%, marginal financing spread widening 100bp in three months), and because the adjusted net loss margin deteriorated from −11% to −22% YoY (an item widely under-reported); NBIS was cut because all six pieces of guidance circulating in the market turn out on comparison to be the 2026-05-13 vintage, so the expectation gap item was marked down accordingly, and because its operating cash flow comes almost entirely from customer prepayments rather than earning power. Neither name's conclusion label changed; what changed is their score gap versus the names below them.

---

5. Detailed Analysis of Top Names

This section is ordered by how much explanation is needed, not strictly by descending total score (the authoritative ranking is Section 3). The first three are today's main catalysts; SMCI in 5.4 is pulled forward separately because its "good news" conceals a fact pointing the opposite way.

5.1 NBIS — Nebius Group (total 76, priority deep-dive)

Related news (8/12 pre-market, company press release): Q2'26 revenue $582.3M, versus $105.1M a year earlier, +454% YoY; first-half revenue $981.3M (+529%). Adjusted EBITDA +$236.2M, versus −$21.0M a year earlier; adjusted net loss narrowed to −$33.2M (from −$91.5M last year, a 64% narrowing).

⚠️ First, clear up a definition that is being mixed up: media generally write "revenue $574.6M, +466% YoY," while the company's press release headline number is $582.3M, +454%. Both are very likely correct, just on different bases — $582.3M is the group basis (including Avride autonomous driving and TripleTen edtech), $574.6M is the Nebius AI cloud segment basis (compare Q1's structure: group $399.0M / AI cloud segment $389.7M). Valuation must use the group basis $582.3M (market cap corresponds to the group); use the segment number only when discussing business momentum, since mixing them systematically overstates. Group revenue of $582.3M is above the market consensus of roughly $569.9M.

Catalyst logic: the key to this report is not the revenue growth rate (the market already knew it would be very high) but the inflection in profit structure. A year ago adjusted EBITDA was still negative; in the same quarter this year it reached +$236.2M against a consensus at the time of about $173.1M (media-relayed; the original sell-side basis was not obtained) — the beat happened on the profit dimension, not the revenue dimension, which is the same logic behind CRWV being bought today. Operating expenses as a share of revenue fell from 206% last year to 130%, and SG&A from 65% to 30% — genuine operating leverage.

Theme and stage: the mid-fermentation stage of the AI cloud compute theme — past the "story-telling" stage (revenue is already at scale), not yet at the "crowded" stage (the profit model is still being validated).

Fundamental verification (from the company's unaudited balance sheet as of 2026-06-30):

• Cash and equivalents $8,042.1M (year-end 2025: $3,678.1M)
• Total debt $8,545.7M (current $46.7M + non-current $8,499.0M), net debt only about $504M
• Operating lease liabilities $1,510.4M (non-current) — must be consolidated into the view when comparing with CRWV; total debt-like obligations including leases are about $10,056M, against cash of $8,042.1M, for net obligations of about $2 billion
• Total deferred revenue $5,975.2M (year-end 2025: $1,577.5M), of which the non-current portion is $4,995.8M
• ⚠️ This is this brief's single most important correction on NBIS: Q2 operating cash flow of +$2,246.1M looks extremely strong, but it comes almost entirely from customer prepayments (deferred revenue rose about $4.38 billion net in the first half, corresponding almost exactly to first-half operating cash flow of about $4.5 billion). Compare Q1, where the split is available: within operating cash flow of +$2,258M, the change in deferred revenue contributed +$3,198M, so stripping out prepayments, Q1 operating cash flow was actually −$940M. In other words, NBIS's "operating cash flow" is in economic substance closer to a form of interest-free financing than to earning power — it is an obligation that must be repaid in future compute delivery. For comparison, prepayments accounted for about 37% of CRWV's first-half operating cash flow, and it remained positive after stripping them out. Treating this money as sustainable internally generated cash flow to plug the capex gap will systematically understate financing risk. And of the $5,975.2M, $4,995.8M is non-current, meaning most of the prepaid amounts correspond to delivery more than 12 months out
• Q2 capex −$5,657.4M, first half −$8,130.3M
• Additional paid-in capital (APIC) rose from $2,360.9M to $7,394.8M — an increase of about $5 billion — but this is not all cash raised: it includes $137.8M of stock-based compensation and the share-based consideration portion of the $605.6M of new goodwill added this period. The correct reading is "APIC up about $5 billion, dilution has already happened," not "$5 billion of cash financing"
• Total shareholders' equity $10,340.5M; shares outstanding 271,855,218
• Q2 operating loss −$175.9M, interest expense $119.1M, net loss from continuing operations −$190.4M
• Stock-based compensation expense $102.5M (14% of operating expenses), the main source of the difference between adjusted and GAAP figures; add it back when reading non-GAAP numbers

Moat / market position: an independent AI cloud headquartered in Europe (Amsterdam), with contracts with both Meta and Microsoft. Its financing structure is far better than peers': $4.3375 billion of convertible notes with coupons of only 1.250% (due 2031) and 2.625% (due 2033), NVIDIA's $2 billion equity investment, and a first $775 million senior secured note priced at SOFR + 2.50% on 2026-07-17 (collateralized by deployed GPUs and investment-grade customer contract cash flows). The difference versus CRWV is not technology, it is the balance sheet and the cost of capital. ⚠️ But the structure of the Meta contract has to be unpacked: the disclosed $27 billion consists of two parts — $12 billion is a five-year compute purchase that only begins in early 2027; $15 billion is an option for Nebius to sell capacity to Meta on predetermined terms, or to its own customers at market prices. The second part is not a Meta purchase commitment, so treating the whole $27 billion as "locked-in orders" overstates it.

Pre-market and technicals: +15.5% @ $223.18 (08:03 ET), pre-market volume 2.08 million shares. ⚠️ This is the least reliable quote in this brief: the API's pre-market field intermittently returns empty values for NBIS (2 of 4 pulls were empty), and another independent source showed +12.5% at the same time — the two differ by 3 percentage points, so read it as a range (a re-check at 08:33 ET showed +17.12%, same direction). The 52-week range is $62.01–$299.86 — the current price sits at roughly the 68th percentile of the range, still 34% below the 52-week high, not an obviously extended position.

⚠️ Another trap when reading the financials: NBIS's GAAP net income is positive in some quarters (for example +$621.2M in Q1'26), but that includes +$781M of non-cash, non-operating gains from revaluing the ClickHouse stake; this quarter's Q2 net loss from continuing operations is −$190.4M, and the year-ago +$502.5M likewise included one-off revaluation gains. Figures like a "93% net margin" served up directly by third-party data sources come from this basis, and citing them would be seriously misleading.

Final judgment: priority deep-dive. It is the only large-cap today where a profit inflection, extremely low net debt, and a price not yet at the 52-week high all hold at once, and its cost of capital is an order of magnitude below peers'. But two things must be stated clearly, or old news gets treated as a new catalyst:

1. This press release contains no forward guidance at all. Comparing item by item against the company's 2026-05-13 Q1 results and original shareholder letter, the six guidance items circulating in the market — year-end ARR of $7–9 billion, 2026 capex of $20–25 billion, full-year revenue of $3.0–3.4 billion, an adjusted EBITDA margin of about 40%, contracted power >4GW, and year-end connected power of 800MW–1GW — not one of them was newly disclosed today; all were given three months ago (and the "raise" to $20–25 billion of capex itself also happened in May). Treating them as new catalysts today would repeat exactly the "cumulative-period figures are old news" class of error.
2. The only genuinely new information today is the actual Q2 numbers: group revenue $582.3M (above the $569.9M consensus), adjusted EBITDA of $236.2M turning meaningfully positive for the first time, cash of $8.04 billion, deferred revenue of $5.98 billion. The rally should be attributed to "the profit inflection being confirmed," not to "guidance being raised." The risk takes the form of equity dilution, not a solvency crisis — paid-in capital already rose about $5 billion in the first half (dilution has happened), and the company still has an untapped ATM plus about 46 million treasury shares. Full-year guidance requires revenue to nearly double again within two quarters (first half $981 million vs full year $3.0–3.4 billion), leaving very little room for error.

5.2 LITE — Lumentum (total 74, priority deep-dive)

Related news (8/11 16:05 ET): FQ4'26 EPS $3.23 (est. $2.95), revenue $1.01 billion (est. $984.6 million), +109% YoY; non-GAAP gross margin above 50% for the first time, which the company said it had not originally expected to achieve at the current revenue level. FQ1'27 guidance: EPS $4.05–4.35 (consensus $3.63), revenue $1.225–1.275 billion (est. $1.16 billion), non-GAAP operating margin 39.5%–40.5%.

Catalyst logic: guidance matters more than the reported quarter, and this stock is the textbook case. The quarter's EPS beat by $0.28, while the next quarter's guidance midpoint of $4.20 is about 15.7% above the $3.63 consensus — and it comes alongside revenue guidance about 7.8% above consensus, which says the company considers the gross margin step-up sustainable. The hardest item here is actually not EPS but the operating margin guidance of 39.5%–40.5% — it speaks more directly than EPS to this round of profitability improvement not being one-off. This was the eighth consecutive quarter of growth and the third consecutive quarter of >20% sequential growth.

One piece of price action that must be spelled out — and a popular explanation that must be corrected: this report drew almost no reaction in yesterday's after-hours session (about +0.05%); the stock only rallied overnight into this morning's pre-market (+8.24%). Yesterday's recap judged this to be "IV crush / already priced." After checking the 8-K original, a different explanation is more likely, and it has first-hand evidence:

LITE reported a GAAP net loss of $7.8 billion this quarter, GAAP loss per share of −$84.65. That loss came from a one-off non-cash loss on debt extinguishment of about $7.8 billion generated by the equitization of part of its convertible notes this quarter.

That is to say, the headline number that hit market terminals first at 16:05 last night was "loss per share of $84.65." In a stock priced above $800 with only about 101 million fully diluted shares, after-hours liquidity is already very thin, and any trade matched off the headline or the EPS field would be frozen by that number first. Only once the $7.8 billion of non-cash loss was stripped out and non-GAAP EPS of $3.23 and the FQ1'27 guidance became visible did repricing occur — and by then it was overnight. That unusual stillness of "almost exactly +0.05%" is better explained by parsing lag than by IV crush.

⚠️ But some restraint is due on the word "overturn" itself. Strictly, the only thing this brief can confirm is: the "already priced" inference drawn yesterday from two hours of after-hours prices was falsified by this morning's +8.24%. As for whether the true cause was "after-hours liquidity too thin + a misleading GAAP headline" or something else, this brief has no data that can distinguish between them. One piece of side evidence worth adding: options priced LITE's implied move into this report at about ±12.2%, while yesterday's actual move was +8.3%, falling inside the implied distribution — that is "a beat that did not exceed the pricing," which is a different thing from "IV crush." The lesson is therefore more precise: two hours of after-hours price is not enough to characterize a report with a guidance beat, and especially not when GAAP and non-GAAP differ enormously. (New conclusions produced by self-correction need evidence too; being a "correction" does not exempt them from checking.)

⚠️ More important: the true starting point of this move was not last night's earnings but the August 4 policy news.

This morning's pre-market price merely returns the stock to its 8/7 level; it has not made a new high. The attribution for this whole move belongs to the FCC's proposed ban on Chinese optical modules as a policy narrative, with the earnings report functioning closer to "confirming the policy narrative will not be falsified by fundamentals" than to supplying new upside. Framing today's optical comms move purely as "LITE earnings-driven" is incomplete — it is essentially a policy trade, and that policy is still at the "proposed / media-reported" stage and has not been enacted.

Fundamental verification (independently recomputed): FY26 revenue $3.014 billion (+83.2%), non-GAAP gross margin up from 34.7% to 46.0% (FQ4 alone 50.4%), non-GAAP operating margin up from 9.7% to 29.8% — revenue up 83% with operating margin up more than 2000bp, the strongest operating leverage among the three hardware companies here. Balance sheet: cash and short-term investments $2.738 billion, interest-bearing debt $1.637 billion, net cash of +$1.10 billion (in FQ4 the equitization of convertible notes cut debt from $2.573 billion to $1.637 billion, at the cost of heavy share dilution). ⚠️ A share-count trap that must be corrected: third-party data sources show about 77.8 million shares outstanding, while the company's own disclosed FQ4 non-GAAP fully diluted share count is 101.1 million (the difference comes from convertible preferred issued in March 2026 and unconverted convertible notes). Using 77.8 million shares understates market cap by about 30%. On the fully diluted basis, market cap is about $89.8 billion, EV/Sales about 29.4x and TTM non-GAAP P/E about 102.5x. ⚠️ LITE's FY26 full-year cash flow statement was not obtained by this brief (not attached to the 8/11 press release), so operating cash flow and free cash flow are left blank pending the 10-K. That is precisely what is needed to judge the quality of the 50% gross margin, and this brief cannot verify it.

Pre-market and technicals: +8.24% @ $888.19 (08:10), pre-market volume 333,000 shares (average daily volume about 5.99 million), a healthy volume ratio. 5-year price percentile 94.9%, 3-year 91.5%, 12-month gain +605.8%. (Note: the 52-week low of $111.20 is an intraday price; on a closing basis the low is $114.62.) Options imply a move of about ±12.2% for this week.

Final judgment: priority deep-dive, but reduce the weight placed on "earnings-driven." The results and the operating leverage are real, and the net cash is real; but the price is already at the 95th percentile of its 5-year range, and this move is driven mainly by a policy that has not been enacted. 29x EV/Sales is only self-consistent on the premise that "+130%-magnitude growth continues for several more quarters." The current price already embeds about an 8% pre-market gap, and whether it fills is the thing to watch after the open; the actual progress of the FCC ban is the real tracking variable for this main line.

5.3 CRWV — CoreWeave (total 68, priority deep-dive, but position is elevated)

Related news (8/11 16:05 ET, SEC EX-99.1 original):

Catalyst logic — there is a definitional issue here that must be made clear: what the market is buying is that the adjusted operating margin of 5% blew through the 2.7% consensus. That is a real expectation gap. But the same metric is down year over year (16% last year). In other words: relative to market expectations this is a beat; relative to a year ago it is deterioration. Both sentences are true, and saying only one is misleading. The market chose to price on the "relative to expectations" basis today because it treats CRWV as a "scale-for-margin" growth story — and that story's premise is that margins have already bottomed and are recovering, with Q2's 5% being only the first data point. One quarter does not make a trend.

A reminder on the backlog definition: the 8-K original says "backlog of approximately $104 billion as of 2026-06-30, excluding more than $25 billion of net new customer commitments added early in Q3." Some media phrase this as "backlog rose from $104.2 billion to $129.2 billion (as of August 11)" — the two are not contradictory, but the latter adds figures from two different points in time, so cite it only with the timestamps stated.

⚠️ A number uglier than "adjusted operating profit declined," and widely missed: adjusted net loss of −$567M (net loss margin −22%), versus −$130M (−11%) a year earlier. The loss amount grew 4.4x and the loss margin doubled — and this happened on the company's own chosen adjusted basis, not as GAAP accounting noise.

Fundamental verification (from the 10-Q original disclosed this morning, as of 2026-06-30):

On the debt, four things must be spelled out:

1. The market's "about $35 billion of debt" framing is correct, but it is an understating framing. Interest-bearing debt is indeed $35.07 billion, but within six months CRWV took operating lease liabilities from $8.20 billion to $16.32 billion (long-term data center leases, which are in substance rigid payment obligations). Total debt-like obligations including leases are $51.61 billion. Net interest-bearing debt is $29.54 billion; net debt including leases is $46.08 billion.

2. The equity buffer is only 6.5%. Equity of $5.024 billion / total assets of $77.070 billion = 6.5% — meaning a 6.5% write-down on the asset side would wipe out all equity. (The same metric for NBIS is about 32.5%, 5x higher.)

3. $10.6 billion of debt matures within the next 18 months, against $5.52 billion of unrestricted cash on hand. This is not a question of "can it repay," it is a structural dependence on continuous refinancing. The first half bore this out: operating cash flow +$3.663 billion, capex −$14.117 billion, free cash flow −$10.454 billion, with the gap plugged by net financing of +$13.985 billion (of which net debt issuance about +$11.5 billion and equity +$2.98 billion). Growth is 100% driven by external financing.

4. ⚠️ The most informative item, and one the media essentially did not cover: the marginal cost of financing deteriorated markedly within three months.

The spread on recourse term loans widened 100bp within three months and carried a debt-service coverage covenant for the first time. The debt footnote in the 10-Q lists existing effective interest rates as high as 15% (DDTL 1.0), 12% (Magnetar), and 11% (DDTL 2.0 and OEM financing). This is the credit market's real-time pricing of CRWV, not an analyst's opinion. For comparison, the first secured note NBIS obtained on 2026-07-17 was priced at SOFR + 2.50% — in the same month, the two companies' marginal secured financing costs differ by 300bp.

Why the interest gap is structural rather than one-off: working back from the 10-Q cash flow statement, Q2 depreciation and amortization was $1,393M. Adjusted EBITDA $1,510M − D&A $1,393M ≈ $117M — 92% of EBITDA is eaten by depreciation. GPU depreciation is not accounting noise; it is the real economic cost of "assets losing value that must be bought again." When depreciation eats 92% and interest ($640M per quarter) is several times what is left, a net loss of −$626M is an arithmetic necessity, not a one-off item (see below for a correction on the precise multiple). (One further note: $79M of interest was capitalized into fixed assets in Q2, so reported interest expense understates the true cost of capital; that portion will return later in the form of depreciation.)

Valuation (on the 10-Q cover share count of 551,536,602): the pre-market $107.02 implies a market cap of about $59 billion, EV (excluding operating leases) of about $88.6 billion, and EV (including operating leases) of about $104.9 billion.

• Against the FY26 revenue guidance midpoint of $12.8 billion: EV/Sales about 6.7–7.1x (ex-leases), 7.9–8.5x (incl. leases)
• Against adjusted EBITDA: EV/adjusted EBITDA about 11.7x (FY26E) — below Microsoft (19.6x), Google (23.6x) and Amazon (18.2x)
• But against adjusted operating profit: EV/adjusted operating profit about 85–173x, versus about 22–26x for Microsoft/Google

"Expensive or cheap" depends on the denominator, and three denominators give three opposite answers — I think the EV/EBITDA answer is the deceptive one. An EV/EBITDA discount presumes "comparable EBITDA quality": depreciation eats 92% of CRWV's EBITDA, while for hyperscalers that ratio is typically 30–40%. The higher the depreciation share, the more EV/EBITDA understates true valuation — CRWV's EV/EBITDA discount roughly equals the premium of its depreciation burden, and once the two offset, the "cheapness" disappears. This is not a "cloud computing valuation" problem; it is closer to a "highly levered infrastructure / leased-asset valuation" problem, and applying a software-cloud multiple framework gets the framework itself wrong, whether it concludes expensive or cheap. One coincidence worth noting: EV including leases of about $104.9 billion ≈ the backlog of $104 billion. The market is currently capitalizing the entire signed contract value into enterprise value 1:1 — with no delivery cost deducted, and with that backlog definition also including a company-"estimated" non-RPO portion (as noted in the 8-K footnote).

⚠️ A piece that must be added: the full set of guidance given on the 8/11 17:00 ET earnings call. CRWV's 8-K press release itself contains no forward guidance (it states explicitly that guidance is given on the call), but the call ended about 15 hours before this brief was written, and the guidance, corroborated by two independent sources, is as follows:

This information changes a key conclusion in this section, and requires a self-correction:

The first draft of this brief used "Q2's single-quarter adjusted operating profit of $128M × 4 = $510 million annualized" against "$2.56 billion of annualized interest" to conclude that "interest is about 5x adjusted operating profit." That method is wrong — it linearly extrapolates a single quarter when the company had already given full-year figures that same evening. Recomputing with the company's own FY26 guidance: FY26 interest is about $3.0–3.2 billion (Q1 actual $536M + Q2 actual $640M + Q3 guidance $860–940M + Q4 extrapolated), against FY26 adjusted operating profit guidance of $960 million–$1.15 billion, a multiple of about 2.6–3.3x, not 5x.

⚠️ This is exactly the same criterion this brief uses in Section 5.4 to criticize SMCI ("one quarter does not make a trend"), applied in the opposite direction to CRWV. The correction:

The direction of the conclusion is unchanged, the magnitude is: even on the company's own most optimistic full-year guidance, interest expense is still more than 2.6x adjusted operating profit — the coverage shortfall is structural, but not as extreme as the first draft said. One further note: whether the FY26 revenue guidance is a "raise" or a "reiteration" is disputed across sources (some say the range was already given in 2026-02 and reiterated on 5/7), and this brief did not obtain a first-hand transcript, so this item is tagged pending verification — if it is a reiteration, the nature of today's +18% is different.

Pre-market and technicals: +18.49% @ $107.02 (08:10), pre-market volume 3.27 million shares (one of the largest among large caps today). The 52-week range is $60.55–$153.20 — the current price sits at roughly the 50th percentile ((107.02−60.55)/(153.20−60.55)), still 43% below the 52-week high. It has been added to the Nasdaq 100 index.

Final judgment: priority deep-dive, but position is elevated (score cut from 72 to 68). The 5% adjusted operating margin blowing through the 2.7% expectation is real, the backlog and the power are real, and the volume is real. But after the 10-Q disclosure this morning, the balance sheet is tighter than the market's framing: $51.6 billion of obligations including leases, an equity buffer of only 6.5%, $10.6 billion maturing within 18 months against $5.5 billion of cash, a marginal financing spread widening 100bp in three months and carrying a financial covenant for the first time, and an adjusted net loss margin deteriorating from −11% to −22% YoY. The thing to watch is whether the gap fills after the open — the current price already prices in 18.5% of this report's good news.

5.4 SMCI — Super Micro (total 56, watch only) — the most important item in this brief

Related news (8/11 16:05 ET, SEC EX-99.1 original):

FY26 full year: revenue $39.1 billion (FY25 $22 billion), gross margin 10.8% (FY25 11.1%), non-GAAP EPS $3.63. End-of-quarter cash $7.5 billion; bank debt and convertibles totaling $8.7 billion (net debt about $1.2 billion). FY27 guidance: revenue $65–72 billion. FQ1'27 guidance: revenue $14.5–15.5 billion, non-GAAP EPS $1.01–1.10.

The first fact that must be spelled out: the two numbers the market is treating as "new good news" were already public three weeks ago. On 2026-07-21 SMCI filed an 8-K preliminary business update that already stated: ① FQ4 revenue would land at the low end of the $11.0–12.5 billion guidance range; ② GAAP and non-GAAP gross margin were expected at 15%–17%, significantly above the prior 8.2%–8.4% guidance, mainly due to improved customer and product mix; ③ more than $60 billion of new orders in FQ4. (7/21 8-K original¹⁶) In other words, neither the "gross margin surge" nor the "$60 billion of orders" was new news last night, and the revenue miss had been flagged in advance too. The only genuinely new information last night was two items: FY27 revenue guidance of $65–72 billion, and the FQ1'27 EPS guidance.

The second fact that must be spelled out: the company's own next-quarter guidance implies margins cut nearly in half. Working back on the company's own guidance basis (non-GAAP diluted share count 761M):

Revenue grows 30–40% sequentially while earnings per share fall 30–40% sequentially. ⚠️ But the attribution must be cleanly separated and not charged entirely to gross margin: the diluted share count rises from FQ4's actual roughly 705M to the FQ1'27 guidance assumption of 745M, about 5.7% of dilution in a single quarter (sourced from the FY26 issuance of 7.00% mandatory convertible preferred of $4.231 billion + common stock of $1.407 billion). That is, of the −35% to −41% sequential EPS decline, about 6 percentage points come from dilution and the rest from gross margin regression. Stripping out dilution, the remaining decline is still mainly explained by gross margin, so the direction of the conclusion does not change. Working backwards, FQ1'27's implied gross margin returns to about 9.8% (assumptions: a 20.5% tax rate, operating expenses moderately above FQ4's $357 million ex-SBC, net interest about −$60 million; assumptions listed so the math can be redone), the same order of magnitude as FY26's full-year 10.8% and FQ3'26's 9.9%. Put differently: the 17.5% gross margin is a one-off mix benefit the company itself does not expect to persist, and the 7/21 wording "primarily due to a favorable customer and product mix" already said as much about its nature. Method note: the reverse calculation above holds up because the company itself gave the diluted share counts assumed in the guidance in the 8-K (GAAP 745M / non-GAAP 761M), so this is a reconstruction of the company's own guidance basis, not outside conjecture.

One point in the company's favor: FQ4 actual gross margin was 17.5% GAAP / 17.6% non-GAAP, above the top of the 15%–17% range pre-announced on 7/21. That is the one piece of hard data on 8/11 that is "better than three weeks ago," and it should be acknowledged.

The third fact that must be spelled out: working capital is deteriorating badly, and it is the hardest negative evidence in this brief.

Inventory of $12.896 billion (+175.5% YoY) and receivables of $6.125 billion (+177.9% YoY) are both expanding about 2.3x as fast as revenue growth (+77.8%), while days payable have not budged, which says suppliers grant no credit terms and the entire working capital expansion is funded by shareholders and creditors. The result: FY26 full-year operating cash flow of −$6.81 billion and free cash flow of −$6.972 billion, against GAAP net income of +$2.23 billion over the same period — a net-income-to-cash conversion ratio of about −305%. The gap was plugged by financing: FY26 net financing inflow of $9.479 billion, including $4.469 billion of new borrowings, $1.407 billion of common stock issuance, and $4.231 billion of 7.00% mandatory convertible preferred. (FQ4's single-quarter operating cash flow turning positive at +$747 million is a good sign, but the first three quarters of FY26 burned about $7.56 billion combined — one quarter does not make a trend.)

The fourth fact: the empirical record on guidance delivery is not good.

Over the last 6 quarters: 3 misses, 1 scrape of the low end, 1 met, 1 big beat; full-year guidance has drifted just as violently (FY25 was revised down from $23.5–25.0 billion to an actual $21.97 billion). And SMCI has filed 3 additional Item 2.02 pre-announcements outside its regular reports (2025-04-29, 2025-10-23, 2026-07-21), so mid-quarter guidance revisions have become routine. FY27's $65–72 billion should be read as "a median expectation with very large variance," not a promise.

The fifth risk: the regulatory issue is heavier than the single phrase "board independent investigation" suggests. A detail of wording: the 7/21 8-K said "alleged export-control issues," while the 8/11 filing dropped "alleged" and wrote "export-control issues" directly. The change in wording may itself carry meaning and is worth noting. Beyond the board's independent investigation, the 10-Q original also records two facts that have been mentioned less often: the company received a subpoena from the SEC's Division of Enforcement on 2024-11-19, and another on 2026-04-28; and on 2026-03-19 federal prosecutors for the Southern District of New York (SDNY) charged three individuals then employed by or affiliated with the company with conspiracy to violate export controls (the company itself was not named as a defendant). On 4/7 the company announced an independent investigation led by the lead independent director and the audit committee chair, supported by Munger Tolles & Olson and AlixPartners, with no completion deadline set to date. Last night's 8-K still states: the investigation's findings could affect the company's forecasts, these results, and results of prior periods. There are also two derivative suits in the Delaware Court of Chancery and a securities class action in the Northern District of California.

The sixth fact, and the easiest one to miss: last night's report is "preliminary, unaudited data." The 8-K describes itself throughout as preliminary / unaudited and states explicitly that the financial close process is not complete, that the independent registered public accounting firm has performed no audit or review procedures on these figures, and that final results could differ materially. Combined with SMCI's history of filing an NT 10-Q (late filing), these numbers should not be treated as final until the FY26 10-K is out.

Final judgment: watch only. The FY27 revenue guidance of $65–72 billion is genuinely striking (equivalent to 66%–84% growth over FY26's $39.1 billion, not "a doubling"), but anyone buying today is paying an +8.5% gap premium for a combination of "revenue surging, margins cut nearly in half, full-year operating cash flow of −$6.8 billion, 3 misses in the last 6 quarters of guidance, and an unaudited report" — while also carrying a regulatory and accounting risk with no timetable. A statistical artifact worth explaining in passing: the media's line today that "revenue of $11.12 billion missed the $11.73 billion expectation" is essentially the product of consensus not having been marked down after the 7/21 pre-announcement — the company told the market three weeks ago that revenue would land at the low end of guidance. That "miss" was not a genuine surprise. This is not a bearish view of the company; it is a bearish view of the risk-reward at today's price. To be clear, SMCI currently trades at a TTM non-GAAP P/E of about 9.5x and EV/Sales of about 0.70x, and the low multiple is not a discount in itself but the market's risk compensation for the six items above.

5.5 COHR — Coherent (total 50, watch only)

Related news: pre-market +5.64% @ $347.11, with no new disclosure from COHR itself. ⚠️ Attribution requires care: it is synchronous in time with LITE's guidance, but the same day saw SMH +1.69% (+2.19% after CPI) and Korea's KOSPI +3.7% overnight, with the entire AI semiconductor sector strengthening — "COHR has no news of its own" can rule out a single-stock event but cannot rule out sector beta. This brief therefore phrases it as "co-moving with LITE's guidance, with sector beta not excludable," rather than "driven entirely by LITE spillover." COHR reports FQ4'26 after the NYSE close tonight (8/12), with the call at 16:30 ET.

⚠️ First, correct a widely cited number that is very likely wrong: media generally write "consensus EPS $1.43 (vs $0.74 a year earlier, +93.2%)." Checked against the company's original disclosures, neither end of that basis lines up:

• COHR's own FQ4 guidance in its 2026-05-06 8-K was non-GAAP EPS $1.52–1.72 — the so-called "consensus of $1.43" is below the low end of the company's own guidance, which is highly unusual for a company that has given an explicit range;
• "$0.74 a year earlier" does not line up either: COHR's Q4 FY25 non-GAAP diluted EPS was actually $1.00.

This brief therefore does not use $1.43 as the bar to clear. A more reasonable threshold is above the guidance midpoint of $1.62. ⚠️ Until the composition of the sell-side consensus is obtained, please do not use $1.43 to judge whether tonight is a beat — this is one of the items this brief explicitly tags as pending verification.

How options price tonight (CBOE delayed quotes, snapshots 07:35–08:08 ET): at the 347.5 strike closest to the pre-market price, the 8/14-expiry at-the-money straddle implies a single-day move of about ±12.4% (about ±10.5% after the 0.85 adjustment factor), with ATM IV of about 164%, corresponding to a market cap swing of about $7.1–8.4 billion. For comparison, LITE implied ±12.2% this week and SMCI implies ±16.5% (front-end IV about 217%, the highest event risk premium of the three).

Why the spillover is actually a risk — three layers:

1. The bar was raised by a peer: LITE delivered revenue +109%, non-GAAP gross margin above 50% for the first time, and next-quarter revenue guidance +134% YoY; COHR's corresponding numbers are next-quarter revenue guidance of about +29.5% YoY, non-GAAP gross margin guidance of 39–41%, and an operating margin of about 20%. Even delivering at the top of guidance, COHR still trails LITE by an order of magnitude on all three of growth, gross margin and operating margin — and tonight the market will use LITE's print as the yardstick.
2. Today's +5.6% is itself a raising of expectations: this move also started with the 8/4 FCC news ($262.89 → $379.13, +44% in five trading days), and after falling −14.2% on 8/10, today's bounce has raised the bar once more heading into the print.
3. A two-way, underappreciated risk: indium phosphide (InP) supply. COHR lists "doubling InP capacity by year-end and doubling again in 2027" as a core growth pillar, and InP's upstream is highly connected to the Chinese supply chain. The policy positive of "the FCC plans to ban Chinese optical modules" and "China's ability to retaliate on indium supply" are two sides of the same coin. If management is asked about InP licensing on tonight's call and gives a vague answer, it could directly offset the policy narrative.

Fundamentals (pre-FQ4 basis): TTM revenue $6.602 billion (single quarter +20.5% YoY), non-GAAP gross margin about 39.5%, TTM non-GAAP net income $937 million / EPS $4.86; cash and short-term investments $2.418 billion, interest-bearing debt $3.425 billion (net debt about $1.01 billion); TTM free cash flow −$539 million, with FQ3 single-quarter capex at 16.1% of revenue — negative cash flow here is an active capacity-expansion choice, entirely different in nature from SMCI's working capital bleed. On 196.4 million fully diluted shares, market cap is about $68 billion, EV/Sales about 10.5x, TTM non-GAAP P/E about 71.2x, and the 5-year price percentile is 95.9%. An interesting comparison: LITE's absolute multiple (EV/S 29.4x) is 2.8x more expensive than COHR's (10.5x), but adjusted for growth COHR is actually the more expensive one (EV/S ÷ growth: LITE 0.132 vs COHR 0.295). The two are a trade-off between "high multiple, high slope" and "medium multiple, medium slope," not a simple expensive/cheap comparison.

Final judgment: watch only. The day before an event, already up 5.6% pre-market, expectations raised by a peer, implied volatility of ±10–12%, and even the "bar to clear" number itself in doubt — five conditions stacked together make a classic asymmetric downside structure. In terms of information state, the only name in the optical comms chain to have released first-hand information today is LITE; COHR's information does not come out until after tonight's close — the two are at completely different information stages.

5.6 CAVA — CAVA Group (total 61, watch closely)

Related news (8/11 16:05 ET): Q2 revenue $365.4M (+31.3%), net income $23M / EPS $0.19; same-store +9.0% (est. +7.1%), of which traffic +5.3% and price/mix +3.7%. Full-year guidance maintained: 75–77 new openings, same-store +4.5%~6.5%, adjusted EBITDA $181–191 million.

Catalyst logic: the quality of the same-store beat is high — 5.3 of the 9% comes from traffic rather than price increases, which is scarce in the current consumer environment. ⚠️ There are two revenue definitions, both from the same 8-K, and neither is a reporting error: GAAP total revenue is $368.4M, while the company's self-defined "CAVA Revenue" is $365.4M (excluding other revenue). Media citing $368.4M are using the GAAP basis. This brief uses $365.4M so it lines up with the same-store basis; please state which one you are citing.

Why only "watch closely" rather than "priority deep-dive": ① full-year guidance was only maintained, not raised — and this question actually has a partial answer inside the same press release: the company maintained full-year restaurant-level margin guidance of 23.7%–24.3% while Q2 actual was 25.7%, i.e. the company itself guided to margin regression in the second half; and Q2's 25.7% was in fact compressed by about 60bp YoY (salmon product launch investment + a rising third-party delivery mix). In other words: same-store is very good, but unit-level profitability is being diluted by investment and channel mix, and that is the reason the full year was not raised. ② The prior close of $60.81 is not far above the 52-week low of $43.41 and far below the high of $98.79, which looks more like the repair of an oversold stock than the breakout of a trending one; ③ pre-market has already gapped +18.0%, pricing one quarter of good news all at once.

Final judgment: watch closely. The fundamental quality is real, but the risk-reward after an 18% gap is no longer attractive.

5.7 CSCO — Cisco (total 55, watch only · earnings tonight)

Reports FQ4'26 and full-year results after tonight's (8/12) close. Consensus EPS $1.17, revenue about $16.82–16.85 billion (about +18% / +15% YoY). The core focus is not the reported quarter but FY27 guidance and the speed at which AI orders convert into revenue — the company's full-year AI order target has already been raised from $5 billion to $9 billion, but related revenue expected to be recognized this fiscal year is about $4 billion, only about 6% of total revenue. The options market implies a move of about ±8.3% on this report, which is a large implied move for a trillion-dollar-scale networking leader. Pre-market +2.03%. Conclusion: watch only — information is incomplete pre-event, and it cannot be priced until FY27 guidance lands.

5.8 HRB — H&R Block (total 62, watch closely)

FQ4'26 EPS $2.38 (est. $2.21, a 7.7% beat); full-year revenue +4.9%, EBITDA +8.3%, adjusted EPS +13.9%, free cash flow $756 million, with $714 million returned through dividends and buybacks over the year; quarterly dividend raised 10% to $0.46; FY27 guidance revenue $4.11–4.16 billion, adjusted EPS $6.04–6.24, with plans to repurchase about another $400 million. Pre-market +12.41% @ $52.46. This is a low-growth name with very solid cash flow and shareholder returns, and the best results in five years plus a dividend raise plus a bigger buyback is a complete package. The risks are that the business itself has a low ceiling and that it has already gapped up 12% pre-market. Conclusion: watch closely.

5.9 EAT — Brinker (total 63, watch closely)

FY26 revenue $5.81 billion, net income $487 million, diluted EPS $10.87; Chili's posted a fifth consecutive year of positive same-store growth, with FQ4 same-store +5.6% (Maggiano's −2.5%). FY27 guidance revenue $6.15–6.27 billion, non-GAAP EPS $12.60–13.40 (including a 53rd operating week, which must be backed out when reading growth rates). FY26 buybacks totaled $400 million, and the board authorized a new $750 million on 8/10; after year-end the company used its revolver to redeem $350 million of 8.25% notes (reducing the interest burden), and will acquire 12 Chili's locations on 8/27. Pre-market +2.99% @ $228.00. Conclusion: watch closely. Guidance is solid, capital allocation is clear, and it is up only 3% pre-market, one of the few delivery-type names today that has not been front-run by a gap.

5.10 MRVL — Marvell (total 52, watch closely) — a volume anomaly

Pre-market +3.55% @ $219.85, with pre-market volume of 4.15 million shares, the largest in this brief's sample of 30-odd names, exceeding CRWV (3.27 million) and SMCI (1.40 million). ⚠️ This brief did not scan the whole market, so it cannot be asserted that this is the largest market-wide. But MRVL has no disclosure of its own today — earnings are not until 8/27. Recent drivers are the launch of its AI memory infrastructure platform and its positioning in Microsoft's Maia 200/300 custom accelerator programs.

Why single it out: a stock with no news that day printing pre-market volume this large has two explanations, but the evidentiary weight of the two is not equal:

• Baseline explanation (evidence-supported): money is using it as a liquidity vehicle for the whole AI semiconductor theme — SMH +1.69% (+2.19% after CPI) and a broad rally across the AI hardware chain both support this.
• Alternative explanation (zero positive evidence): there is news not yet public. Independent checking found no company disclosure or SEC filing within the window (earnings are not until 8/27), so this item is retained only as something that cannot be ruled out; it is not an equal other half. Hence only "watch closely," tagged as a volume anomaly rather than news-driven. If volume persists after the open with no news landing, the baseline explanation holds.

---

6. Bearish / Avoid List

---

7. Within-Theme Rankings

7.1 AI Cloud Compute (Neoclouds)

7.2 Optical Comms / Optical Modules

7.3 AI Power and Data Center Infrastructure

---

8. Open-Verification Signals

Pre-market (before 09:30)

• Gap direction and volume of the theme leaders: CRWV (+18.5%, 3.27 million shares pre-market), NBIS (+15.5%, 2.08 million), LITE (+8.2%, 330,000) all have healthy pre-market volume; this is not a thin-tape ramp. Watch particularly whether 09:15–09:30 holds the pre-market highs — for names gapped 15%+, weakening in the final 15 minutes before the open sharply raises the odds of a gap fill.
• Whether MRVL's 4.15 million shares of pre-market volume is accompanied by news landing. If volume persists after the open with still no news, judge it a theme vehicle rather than a single-stock event.
• FN traded only 47 shares pre-market, so its +5.41% quote carries no reference value; do not use it for any judgment before the open.
• A 08:29 ET re-check ahead of CPI (19 minutes later than the main snapshot): the main names not only held but strengthened slightly — CRWV +18.69% (pre-market volume up to 3.84 million shares), NBIS +16.45% (3.63 million), SMCI +8.99%, LITE +8.17%, COHR +6.20%, SPY +0.35%, QQQ +0.82%. The gaps did not decay ahead of the event and pre-market volume was still rising, which says this is not a thin-tape ramp. But it also means: all the good news is already in the price, and the CPI result will be priced from here.

Intraday (09:30–10:00)

• Whether the first 30 minutes is gap-and-go or gap-fill: CRWV / NBIS / CAVA / HRB, the four names gapped 12%–18%, are today's best observation sample.
• Sector ETF confirmation: the test of whether the theme is real is whether the SMH/IGV divergence persists. If SMH rises and IGV turns higher too, it is broad beta, not a theme; only if the divergence persists does the "selling shovels vs using shovels" judgment hold.
• Volume confirmation: the optical comms chain (LITE / COHR / CIEN / FN / GLW) needs to see at least three names rise on higher volume before this can be confirmed as a theme rather than a single-stock event in LITE.

Options sentiment (CBOE delayed quotes, snapshots 07:35–08:08 ET)

How to read this: LITE's actual move yesterday fell inside the pre-event implied range, making it "a beat that did not exceed the pricing." COHR is priced for a ±10–12% single-day move tonight while its "bar to clear" is itself in doubt (see Section 5.5) — big volatility plus an unclear benchmark is the hardest combination to trade.

• ⚠️ Implied volatility data was not obtained for tickers other than the four above, so related judgments are inferred from price action.

Risks

• Reversal after the gap: 6 names are up more than 7% pre-market today, and historically the gap-fill rate on days like this is not low.
• Lone risers with no sector follow-through: HYLN (+25.8%) and HRB (+12.4%) are single-stock events with no sector support.
• Index futures versus Treasuries diverging: this morning equity futures are up while 10Y/2Y/30Y yields are all down 2bp in tandem — both lean bullish, so there is no divergence, an unusually consistent signal this morning. But it is a consistency established before CPI, which limits how much it is worth.
• ⚠️ The most important item: VIX 15.38 (+0.65%), while VIX9D at the time of this snapshot was still 8/11's 12.52, not updated. Yesterday's recap already noted that "with only one night left before CPI, short-dated protection is actually cheaper." This morning VIX only ticked up 0.65%, meaning that "under-protection" asymmetry was still unrepaired on the morning of the event. If CPI surprises to the upside, the market lacks a cushion.

---

9. Final Conclusions

9.0 CPI Scenario Mapping → now released; result and market reaction below

The scenario table in this section was written at 08:19 ET (before the CPI release), and the "actual result" below was appended after the 08:30 ET release; the chronological order of the two is preserved so the usefulness of this mapping can be checked.

9.0.1 ✅ Actual July CPI result (BLS original, released 08:30 ET)

Composition: the shelter index rose +0.1% MoM, contributing about two-thirds of the total monthly increase; food +0.1% (food away from home +0.3%); energy −1.5% MoM (but still +14.7% YoY — the Hormuz effect has not yet washed out of the year-over-year figure). Within core, medical care, airfares, communication, education and recreation rose, while motor vehicle insurance fell. (BLS original¹⁷)

9.0.2 Immediate market reaction (08:32–08:33 ET, about 3 minutes after release)

Verdict: this lands in the "in line" branch of the table above. Three pieces of evidence agree: ① yields at both ends barely moved (10Y only −0.4bp), showing this data did not change the pricing of the rate path; ② VIX flipped from up to down, clearing event risk; ③ the SMH/IGV divergence did not converge but widened (+2.19% vs −1.08%).

What this means (judgment): the theme ranking in Section 2 of this brief holds under this branch, and it is the most favorable branch — macro did not take the pricing power, and the main lines are led by earnings and thematic logic. But note that the stance in Section 9.4 has not changed: what CPI cleared was "data risk," not "position risk." CRWV has already gapped 21%, NBIS 17%, SMCI 10%, so the good news is priced more fully, not more cheaply. ⚠️ Also, energy at +14.7% YoY shows the geopolitical premium is still inside inflation, and there is still one more August CPI and one more employment report before the September FOMC — has the hike pricing been removed? No, it has merely not been reinforced.

Market expectations: headline CPI +3.4% YoY (June +3.5%), +0.1% MoM (June −0.4%); core CPI +2.5% YoY (June +2.6%), +0.1% MoM (June flat). Key background: the ripples from the Iran conflict still dominate monthly volatility — oil fell in early July on optimism about talks and recovered late in the month as fighting resumed, with gasoline prices netting out at about −2% to −3% in CPI. Even so, inflation remains above the +3.2% average hourly earnings growth over the same period. Policy background (this is the pivot of the brief's framework and must be stated first): the next FOMC is September 15–16. On the CME FedWatch basis, the probability of a 25bp hike in September is relayed variously at about 55%–65% (corresponding to different snapshots around 8/4–8/6); this brief did not obtain a live reading at the time of writing, so it uses a range rather than a point estimate. ⚠️ The "roughly a coin flip between a hike and no move" phrasing elsewhere in the text should be read against this range — what can be said with certainty is the direction: the probability of a hike is no lower than that of no move, and a cut is essentially not in the pricing. The July 29 decision was to hold, but the vote was 9 to 3, with three officials favoring a hike. In other words, the risk direction of this CPI print is the opposite of the past two years: the bad case comes from "too hot," not "too cold." July payrolls came in far below expectations, giving the "high inflation + weak employment" combination a stagflationary character, which is also why September pricing has been unable to converge.

What this table is meant to convey: the theme ranking this brief gives (AI compute and optical comms at the top) holds in both the "in line" and "hot" scenarios, but is the weakest tier in the "cold" scenario — because a cold print favors the de-rated software and small-cap names, not the winners that have already gapped. The theme judgment is conditional on the scenario; do not execute it in isolation from the CPI result.

9.1 The 5 Stocks Most Worth Watching Today

9.2 Today's 3 Strongest Themes

9.3 What to Avoid Today, and Why

1. Software / SaaS (NOW, CRM, WDAY, DDOG, PLTR) — today's direct evidence is the divergence between IGV −0.99% (−1.08% after CPI) and SMH +1.69% (+2.19% after CPI), the cleanest single item of the day. But keep them separate: the divergence is today's fact, while "SaaSpocalypse" is a half-year background narrative, not today's catalyst.
2. ONON, KTB — the guidance cut and revenue miss are genuine negatives, but ONON is already −20.29% in a single day and sitting against its 52-week low; and it simultaneously raised gross margin guidance, with DTC channel constant-currency growth of 34.3% — what was killed was growth, not profitability. Avoid; but the bad news is largely reflected in the price, so it is not listed as a short-watch candidate.
3. Names reporting tonight that are already up pre-market (COHR +5.6%, CBRS +2.2%, CSCO +2.0%) — information is incomplete pre-event, and the pre-market gains have already raised the bar. COHR is especially dangerous: its rally comes entirely from LITE spillover, and that spillover raises its own expectation bar.
4. SMCI — see Section 5.4. Its revenue guidance and margin guidance point in opposite directions, and the market has priced only the former.
5. Micro-cap movers group (OFAL, RMCF, BOXL, PLAG, WXM) — poor quote reliability, no verifiable disclosure.

9.4 Final One-Sentence Judgment

[Conclusion after the 08:33 ET addendum; direction unchanged] CPI has landed fully in line, and what it cleared was "data risk," not "position risk" — after the release CRWV went from +18.5% to +21.0% and SMCI from +8.5% to +10.0%, so the good news is priced more fully, not more cheaply. The passage below, written before CPI, therefore still holds; the release has only made the answer to "should you pay for the gap" a little more expensive:

Today's real question is not "which AI stock to buy" but "whether to pay for a pile of good news that has already gapped 8%–21%." The information on the earnings side is hard — CRWV's and NBIS's operating leverage and LITE's guidance all come from first-hand disclosure and are mutually independent; but the information on the price side is hard too: that good news was already priced 8% to 21% before the open, and after CPI landed each of them added another 1–2.5 percentage points. Whatever upside the clearing of macro risk offered has already been eaten by the price itself.

The overall stance of this brief is therefore: thematically bullish, conservative on execution tempo. The logic hardness of the AI compute and optical comms main lines is acknowledged — they are backed by first-hand disclosure, and after CPI the SMH/IGV divergence widened rather than converged, showing the themes still hold the pricing power; but be aware that the names up the most today also have their good news priced the most fully, and the list still contains names not yet front-run by a gap (such as EAT, up only 3% pre-market). Whether the gaps fill is the single variable most worth watching after the open. Two more things to remember: ① tonight's three reports from CSCO, COHR and CBRS reintroduce event risk, and all three are already up pre-market; ② energy is still +14.7% YoY, the geopolitical premium has not left inflation, and there is one more CPI and one more employment report before the September FOMC — hike pricing has merely not been reinforced, it has not been removed.

---

Internal Data-Collection Record (not sent to clients)

Failures and workarounds in this round of data collection

• Fetching SEC.gov directly via WebFetch returned HTTP 403 (twice: SMCI 8-K, CRWV 8-K). Switching to curl with a compliant User-Agent declared (StarrQuantLab Research <email>) returned 200 successfully. This is SEC's crawler policy requirement, not a failure on our side, but the WebFetch path is unusable for sec.gov — go straight to curl in future.
• Benzinga /premarket, TheStreet and the CNBC pre-market article all returned 403 via WebFetch; the businesswire NBIS press release timed out on WebFetch at 60s. The NBIS press release was ultimately captured in full from the EQS-News relayed version (including the complete balance sheet and income statement), consistent with the company's own figures.
• NBIS's stockanalysis pre-market quote field is intermittently empty: in 4 pulls, 2 returned empty strings for ep/ecp (though epv, pre-market volume, returned normally). This is an intermittent API problem, not a trading halt. This brief uses the successful 08:03 ET reading (+15.5% @ $223.18). ⚠️ Note: NBIS's pre-market snapshot time in the body (08:03) is 7 minutes earlier than the other tickers (08:10), and this is tagged in the tables.
• NBIS's Q2 6-K had still not appeared on EDGAR as of 08:10 ET (the most recent 6-K was 7/27), and the shareholder letter PDF was not obtained. So the new guidance on ARR and capex is not confirmed by this brief.
• An early round of WebSearch returned self-contradictory micro-cap data: the same names (WXM, PLAG) appeared in one summary as "the top two gainers at +175% / +104%," while the direct stockanalysis API showed them as "the top three decliners at −35.19% / −77.97%." Everything was switched to timestamped direct stockanalysis data, and this group of names was placed wholesale on the avoid list in the body.
• The same round of search also described ONON as "−16.5% in this morning's pre-market", while direct quotes show ONON was −20.29% in the 8/11 regular session and +1.20% in this morning's pre-market. This is a variant of the "silent one-day lag" class of failure — the search summary pinned yesterday's intraday drop onto this morning's pre-market. Corrected against the direct quote.

A few methodology notes worth recording

1. SMCI's "pre-announcement trap": the SEC 8-K captured in the first round of searching (accession 0001375365-26-000019) looked like the earnings report, but was actually dated 2026-07-21, a preliminary business update. It took listing the filings via data.sec.gov/submissions/CIK0001375365.json to discover that the real 8/11 report is 0001375365-26-000021. Lesson: the first thing to do with an SEC filing is verify the filing's own date; do not trust the context supplied by search results. Without that check this time, a three-week-old pre-announcement would have been written into the top theme as last night's new catalyst — the same pit as "cumulative-period order figures are old news" in memory.

2. LITE overturned a judgment from yesterday's recap: yesterday's recap judged, from "a big earnings beat but only +0.05% after hours," that this was IV crush / already priced; this morning's pre-market +8.24% (08:10 snapshot; briefly +8.46% at 08:05) shows the real repricing happened overnight. Conclusion: two hours of after-hours price is not enough to characterize a report with a guidance beat — the window is too short. This has been written into Section 5.2 of the body.

3. NTAP was an "earnings report" conjured out of thin air by a search summary. An early round of searching returned "NetApp +5.85% after hours, adjusted EPS $2.05 beating $1.89, revenue $1.71 billion." It was briefly written in at number 7 on the overall single-stock table. What triggered suspicion was the quote data itself: NTAP had traded only 1,041 shares pre-market, +0.52% — impossible volume for a company that had just reported. Independent verification confirmed: NetApp's FQ1'27 earnings date is 2026-09-02; it did not report at all on 8/11; the "$2.05" in the search summary was actually the low end of the company's FQ1'27 guidance range of $2.05–2.15, mistaken for an actual figure. The whole entry was deleted and the rankings re-ordered. The criterion to remember here is: when "news" and "volume" contradict each other, trust the volume. Pre-market volume is the hardest field to fake.

4. Output from the two fundamentals sub-agents materially changed three conclusions, worth recording for the input-output ratio:

   • CRWV: the agent captured the 10-Q disclosed only this morning (I had only the 8-K when writing the first draft), and from it found that operating lease liabilities doubled in six months to $16.3 billion, that equity/assets is only 6.5%, that there is a $10.6 billion maturity wall within 18 months, and that the marginal secured financing spread widened from SOFR+2.25% to SOFR+5.50% in three months and carried a DSCR covenant for the first time. It also caught the adjusted net loss margin −11%→−22% that I had missed. Score cut from 72 to 68.
   • NBIS: the agent downloaded the original 2026-05-13 Q1 results and shareholder letter PDFs and compared item by item, confirming that all six pieces of guidance circulating in the market are three months old. This corroborated the first draft's suspicion, but upgraded "unconfirmed" to "confirmed as old." It also pointed out that operating cash flow comes almost 100% from customer prepayments (Q1 was −$940 million after stripping them out). Score cut from 78 to 76.
   • SMCI: the agent independently recomputed the implied gross margin (about 9.4%–9.7% after accounting for operating expenses and interest, validating the method by back-solving the original FQ4 guidance), and added five items: deteriorating working capital (cash conversion cycle 94→156 days), FY26 free cash flow of −$6.97 billion, 3 guidance misses in the last 6 quarters, two SEC subpoenas, and the report describing itself as unaudited. The conclusion label did not change but the evidentiary strength rose substantially. Conclusion: these two agents each ran about 14–16 minutes, but both changed substantive judgments — worth the wait. Consistent with "research sub-agents are worth waiting for" in memory.

5. Two field traps in yfinance (reported by an agent, avoided): ① LITE's sharesOutstanding is 77.8 million shares, while the company's disclosed non-GAAP fully diluted count is 101.1 million, so using the former understates market cap by about 30%; ② LITE's forwardEps = $31.51 is badly inconsistent with the company's own FQ1'27 guidance of $4.05–4.35, so that field cannot be used for a forward P/E. Also, NBIS's GAAP net income includes a +$781M non-cash gain from the ClickHouse revaluation, which is where third-party sources' "93% net margin" comes from.

6. Quality-control (risk-auditor) results and disposition — 25 red flags in total, handled as follows:

   Substantive errors corrected (7):

   • The entire block of CRWV call guidance was missing (most serious). The company had already given, at 8/11 17:00 ET, Q3 revenue of $3.4–3.6 billion, FY26 revenue of $12.4–13.2 billion, FY26 adjusted operating profit of $960 million–$1.15 billion, and FY26 capex of $35–39 billion. This directly overturns the first draft's core bear argument that "interest is 5x adjusted operating profit" — recomputed with the company's full-year guidance it is 2.6–3.3x. The first draft committed exactly the error this brief criticizes SMCI for in Section 5.4 (linear extrapolation of a single quarter), and an explicit self-correction has been made in the body.
   • CRWV's 52-week percentile was miscalculated: 57% → 50.2% (arithmetic error).
   • The CAVA "media got the revenue wrong" claim was a misjudgment: $368.4M is GAAP total revenue and $365.4M is the company's self-defined CAVA Revenue, both from the same 8-K; CNBC did not get it wrong. Changed to a definitional note.
   • CAVA's "why didn't they raise the full year" actually has an answer: the company maintained full-year restaurant-level margin guidance of 23.7%–24.3% while Q2 actual was 25.7% (−60bp YoY) — the answer was in the same press release, and the first draft left a checkable fact as an open question.
   • SMCI's implied net margin base was too high: 11.7% → 11.2% (the first draft applied the forward share count of 761M back to FQ4 actual, where the actual diluted count was about 705M); and the layer of share count 705M→745M, 5.7% single-quarter dilution was added, splitting about 6pct of the "−35%~−41% sequential EPS" out of the gross margin attribution.
   • Naming errors: "S&P futures / Nasdaq" were actually SPY/QQQ pre-market ETF prices; SMH is the VanEck semiconductor ETF, not the Philadelphia Semiconductor Index (SOX corresponds to SOXX).
   • The NBIS shareholder letter "not obtained" phrasing was inappropriate: it was published at 07:38 ET alongside the results — this round failed to capture it, it was not unavailable. The wording has been corrected.

   Wording and compliance issues corrected (6):

   • Several passages constituted executable trading instructions ("do not initiate a position," "the correct name is LITE," "wait for the gap fill to enter," "do not chase the short"), conflicting with the disclaimer at the end. All rewritten as observational statements, retaining the five-tier label system prescribed by the skill.
   • Universal quantifier: MRVL's "largest pre-market volume market-wide" → "largest in this brief's sample of 30-odd names" (this brief did not scan the whole market).
   • FN self-contradiction: Section 8 declares its 47-share quote unusable, yet Sections 3.1 and 7.2 used +5.41% as spillover evidence. Now tagged as unusable consistently at every citation.
   • COHR had co-movement written as causation: "the driver comes entirely from LITE spillover" rules out a single-stock event but does not rule out sector beta (SMH +1.69% and KOSPI +3.7% the same day). Theme 2's logic hardness was downgraded from "hard" to "medium."
   • SaaSpocalypse was propped up at A-level by a half-year-old source: downgraded to "background (not an in-window event)," with IGV −0.99% retained as the only evidence today.
   • MRVL's two explanations were written as equals: in fact one has sector evidence and the other has zero, so it was rewritten as "baseline explanation / cannot-be-excluded item."

   Source tags added (3): the three consensus figures driving the "expectation gap" scores (CRWV 2.7%, NBIS $173.1M, SMCI $53.29 billion) are all media-relayed, now tagged at each occurrence; FedWatch's September hike probability is relayed variously at 55%–65%, now expressed as a range with a note that no live reading at the time of writing was obtained; the snapshot times were corrected from "08:06–08:10" to the actual "07:43–08:10, tagged per ticker."

   Parts quality control confirmed as correct and needing no change: all SMCI earnings figures and the 7/21 pre-announcement judgment, all CRWV Q2 figures and the backlog definition reminder, all NBIS figures and the causal chain "the deferred revenue surge is the source of positive OCF," all LITE figures and the two recomputations, the event information for CAVA and COHR/CBRS, the distinction between ONON's drop yesterday and its pre-market this morning, the broad-market and cross-asset conventions, and the internal-section stripping mechanism. Quality control also confirmed that Teledyne's acquisition of Varex does not count as a missed item (the 8-K acceptance time precedes the start of the coverage window).

   Unresolved, left for a human (1): the risk disclaimer lacks a holdings/conflict-of-interest statement. This brief does not add one on its own — that requires real holdings facts and cannot be fabricated by the generating end. Suggest adding it uniformly in the footer template in send.py.

7. yfinance was partially rate-limited this round (one agent reported 429s on .history() and the option chain, with .info and the three statements available; another agent was rate-limited in its second half). The main quote path went through the direct stockanalysis API throughout with no rate limiting; option data was switched to CBOE cdn.cboe.com/api/global/delayed_quotes successfully. stooq is still blocked by a JS challenge and was not used.

---

⚠️ Risk disclaimer: this list is pre-market information organization and observation only and does not constitute investment advice. US equities carry high volatility and pre-market gap risk, and post-earnings IV crush and guidance reversals occur; automatically generated content may contain stale information or factual errors. Rely on company disclosures and SEC filings, and do not use this directly as a basis for trading.