📊 Ainvest Option Flow Digest — 2026-07-28

On a Red Day for AI & Power Names, the Big Money Went Neutral

⚡ The 30-Second Read

It was an ugly session for the AI-infrastructure trade — Corning −16% (soft guidance), Bloom Energy −16% (a pre-earnings meltdown), Intel −5.9%, Nvidia −5% — and the options flow tells you how the pros played it: mostly by NOT taking a side. Across seven names we counted ≈$97M of premium, but the three biggest tickets are all hedged or neutral structures, not directional bets:

• 🔷 INTC — the day's biggest, ≈$24M, is a delta-hedged deep-ITM $70 call paired with ≈744,800 short shares (≈106% match). Long call + short stock = a synthetic-put / financing structure, NOT a bullish Intel bet.
• 🛡️ CRWD ≈$15M — a delta-hedged covered-call overwrite (short $195 LEAP calls + 176,800 long shares), harvesting rich long-dated IV. Neutral.
• 🔬 GLW ≈$7M — a delta-hedged long-convexity bet (long $140 calls + 357,000 short shares) that Corning makes a big move after today's 16% drop. Direction-neutral on day one.

The genuinely directional flow was smaller and split:

• 🟢 NVDA ≈$17M — an unhedged Dec-2027 $210 LEAP call, a bullish multi-year dip-buy after Nvidia's ≈5% drop. · ⚛️ FRMI ≈$7.2M — a deep-ITM put sale on Rick Perry's AI-power startup Fermi = a bullish synthetic-long bet.
• ⚡ BE ≈$17M — a naked $300 LEAP call sale (short upside vol / income, bearish-lean) after BE's crash. · ⚡ TLN ≈$10M — a deep-ITM put buy on the AI-power name Talen (bearish-lean / hedge).

🔬 Delta-hedges verified on both tapes today. INTC: 9,200×100×0.762 ≈ 701K vs a 744,800-share short block (≈106%). CRWD: 2,600×100×0.66 ≈ 172K vs a 176,800-share long block (≈103%). GLW: 10,200×100×0.335 ≈ 342K vs a 357,000-share short block (≈104%). That's what makes "delta-hedged" a fact, not a label. The genuinely unhedged directional tickets: NVDA + FRMI (bullish) and BE + TLN (bearish-lean).

🔁 OI Review — Last Session's Provisional Flags, Now Resolved (covers 2026-07-27)

The July 28 pre-market open-interest snapshot is in, and Monday's tape did not survive it intact. Of ten ⏳ flags, five resolved as published and five did not — including three outright inversions. This moved the day's genuinely-new premium from ≈$146M down to ≈$77M.

🔄 Inversions

🐺 WOLF — the "$59.6M bearish put ladder" was mostly a short-put unwind. Two legs closed: $25P 32,182 → 9,693 (−22,489) and $30P 50,379 → 19,266 (−31,113) — open interest only falls when contracts are retired, so those buys were buy-to-close (the desk spent ≈$44.5M covering existing short puts). The $20P was a pure transfer (+4). The only genuinely new position is a $22.50/$17.50 ratio bear spread for ≈$8.4M — a roll-down, not a fresh pile-on. Corrected delta ≈−0.48M shares, not −4.25M, and because the short leg exceeds the long, it actually loses in a true collapse.

🛡️ SPY — we called it a transfer; it was a ≈291,000-contract FRESH open. $525P 10,933 → 207,281; $625P 37,562 → 132,651. Root cause: the "flat OI" came from pre-print morning snapshots — OPRA publishes OI for the prior close, so neither number could contain a same-day 09:50 ET trade. A genuine ≈$6.8M net tail-hedge open, not a transfer.

🥈 SLV — not a bearish silver bet at all; it was a conversion. Both legs opened, which sent us back to the tape — every leg carried a stock+option cross marker (2,000,000 shares against the 20,000-lot, exactly 100 per contract). Long stock + synthetic short at the same strike = a conversion: deltas cancel, ≈3.1% annualized locked, no view on silver whatsoever. Our "unhedged short" framing was wrong on every count.

❗ Size corrections

🌐 NBIS — the second clip was exchange-cancelled; settled ≈$10.7M, not $21.5M (fresh-open read holds). · 📊 QQQ — only ≈52% opened; genuinely new overwrite ≈$13.7M credit, not $27.4M.

✅ Confirmations

🔴 AMD $760C +916 ✅ OPEN · 💾 DRAM $55P +7,161 ✅ OPEN (hedge intact) · 💉 MRNA $85C +2,321 ✅ OPEN · 🎮 TTWO $280C +1,917 ✅ OPEN (OI now 10×) · 💽 WDC $500C +452 ✅ OPEN.

Three hard-earned lessons: (1) a same-morning OI snapshot can never settle a trade that prints later that day; (2) check the condition code for a stock leg before calling anything unhedged (SLV was a conversion all along); (3) the standing one — a big premium headline tells you dollars moved, not that exposure was created (WOLF moved ≈$59.6M of cash and created ≈$8.4M of new risk).

📋 Today's Flow at a Glance

Ticker	Premium	Expiration	Catalyst	Option Play	What It Means
🔷   INTC	≈$24M	Nov 2026 (Monthly)	Q3 earnings ≈Oct 22	Deep-ITM $70 call + 745K short-stock	Delta-hedged financing/synthetic   — NOT a bullish bet
⚡   BE	≈$17M credit	Jun 2027 (LEAP)	Q2 earnings Jul 28   (crashed pre-print)	Naked $300 LEAP call   SALE	Short vol / income   (bearish-lean) after the ≈55% drawdown
🟢   NVDA	≈$17M	Dec 2027 (LEAP)	Q2 earnings Aug 26	$210 LEAP call buy (unhedged)	Directional bullish   — multi-year AI dip-buy
🛡️   CRWD	≈$15M credit	Jun 2028 (LEAP)	Q2 earnings Sept 2	$195 LEAP call SALE + 177K long-stock	Delta-hedged covered-call overwrite   (neutral)
⚡   TLN	≈$10M	Dec 2026 (Quarterly)	Q2 earnings Aug 5 + FERC	Deep-ITM $370 put buy (unhedged)	Directional bearish / hedge   on the AI-power name
🔬   GLW	≈$7M	Sep 2026 (Quarterly)	Citi TMT Sept 9 (earnings after expiry)	$140 call + 357K short-stock hedge	Delta-hedged long-vol / convexity   (neutral)
⚛️   FRMI	≈$7.2M credit	Dec 2028 (LEAP)	Anchor contract (un-dated) + Q2 Aug 13	Deep-ITM $12 put   SALE   (synthetic long)	Directional bullish   on the AI-power startup

Net/structural premium ≈$97M. INTC, CRWD and GLW (≈$46M combined) are delta-hedged/neutral. Directional views: NVDA + FRMI bullish, BE + TLN bearish-lean.

🔍 The Stories Behind the Prints

🔷 INTC — the biggest ticket of the day is a financing trade, not a bet. A desk paid ≈$24M for deep-in-the-money Intel $70 calls and, at the same instant, shorted ≈744,800 shares — a ≈106% delta match. Long call + short stock is, by put-call parity, a synthetic put / delta-one financing structure: it carries almost no directional view. That's notable given the day (Intel −5.9%) and the setup (a stock re-rated ≈4× off its ≈$20 lows on a U.S. government stake, a $5B Nvidia investment, and 18A silicon, now with a genuinely two-sided Street at $75–$155). When you see a $24M "call buy" that's fully hedged with a short-stock block, it isn't conviction — it's plumbing.

⚡ BE — selling the crash's rich volatility. Bloom Energy melted down ≈16% into its July 28 earnings (a short-seller report, a rejected gas pipeline for Oracle's Project Jupiter, and "H2 math" skepticism after a 248% first-half run), and a desk used the spike in implied volatility to sell 3,919 June-2027 $300 calls for ≈$17M credit — a naked, short-upside-vol income bet that BE stays below $300 (≈+90%) through mid-2027. It's the mirror image of chasing: harvesting fear rather than paying for it. (The gain is capped at the credit and the loss is large on a melt-up — a professional's trade, not a starter one.)

🟢 NVDA vs ⚡ TLN — the day's clearest directional split. On NVDA, someone bought ≈$17M of unhedged Dec-2027 $210 calls into the ≈5% dip (sparked by a report Nvidia may backstop $250B of OpenAI's financing) — a patient, multi-year bullish bet on the AI-compute cycle into the Aug 26 print. On TLN, a desk bought ≈$10M of deep-ITM Dec puts on Talen Energy — a bearish-lean or hedge on the nuclear/AI-power name after a ≈3.5× run, into FERC co-location risk and a capped PJM capacity price. Same AI-infrastructure theme, opposite directions.

⚛️ FRMI — a synthetic-long lottery on Rick Perry's power startup. Fermi Inc. — a pre-revenue AI-datacenter/power REIT that IPO'd at $21, spiked to $37, and now trades ≈$6 — drew a ≈$7.2M sale of deep-ITM 2028 $12 puts. Selling a deep-ITM put is synthetically long the stock: the desk keeps the full credit if Fermi clears $12 by 2028 (≈+89%) and otherwise takes assignment near a ≈$4.80 basis. It's a patient bet the "Project Matador" campus lands an anchor hyperscaler contract — against real risk: ≈$700M/yr cash burn, a terminated $150M anchor deal, and a live securities class action. Speculative, defined-thesis, and not a template to copy.

📅 Upcoming Catalysts — Mapped to the Right Expiration

(Catalyst date ≠ option expiration — kept separate on purpose.)

Catalyst	Date	Ticker(s)	Option Expiration That Captures It
TLN Q2 earnings	Aug 5 (after close)	TLN	Dec 2026
FRMI Q2 earnings	Aug 13	FRMI	Dec 2028 (LEAP outlives it)
NVDA Q2 earnings	Aug 26 (after close)	NVDA	Dec 2027 (LEAP outlives it)
CRWD fiscal-Q2 earnings	Sept 2 (after close)	CRWD	Jun 2028 (LEAP outlives it)
Citi Global TMT Conference	Sept 9	GLW	Sep 18 (the one dated event in-window)
INTC Q3 earnings   (est.)	≈Oct 22	INTC	Nov 2026
BE Q3 earnings   (est.)	≈late-Oct/Nov	BE	Jun 2027
FRMI anchor-tenant contract	un-dated binary	FRMI	Dec 2028

👥 How Four Different Traders Might Read Today

🎲 The YOLO trader sees the NVDA dip-buy and wants the lottery ticket. Fair — but note it's a 2027 LEAP, a multi-year thesis, not a this-week trade, and you'd be buying into an AI-bubble/circular-financing debate that just knocked 5% off the stock. If you play it, size it as money you can lose and give the thesis years, not days. FRMI is the spicier version — a pre-revenue name that can double or halve on a single contract headline.

📈 The swing trader should read today as a positioning day, not a breakout. The three biggest tickets are hedged/neutral — big desks are financing, overwriting, and buying convexity, not picking a direction into a red tape. That's a market bracing for a move, not calling one. Wait for the catalysts (TLN Aug 5, NVDA Aug 26) to set the tone.

💵 The premium collector has two textbook examples today: BE's naked call sale (harvesting a post-crash volatility spike) and CRWD's delta-hedged covered-call overwrite (monetizing rich long-dated IV). The lesson isn't "sell naked calls" — BE's is a professional structure with large tail risk — it's that the disciplined version is defined-risk (a call spread, or a genuinely covered position), and that selling into elevated IV beats selling into calm.

🌱 The beginner gets the cleanest lesson of the week: the biggest number on the tape (INTC's $24M) had no directional opinion at all. It was a hedged financing structure. Before you ever act on "unusual flow," learn to ask: Is it hedged? Is it a buy or a sell? And is the catalyst before or after the option expires? Today, those three questions turn a scary-looking $97M tape into a mostly-neutral, risk-managed one.

🛡️ The Standing Reminder

On a down day for AI and power names, the smart money mostly refused to chase — three of the four biggest tickets were hedged financing, overwrite, and convexity structures with no directional signal, and even the directional flow was split (NVDA/FRMI bullish, BE/TLN bearish). And yesterday's OI review is the exclamation point: a $59.6M "bearish ladder" (WOLF) turned out to be an ≈$8.4M roll, a "transfer" (SPY) was a ≈$291K-contract fresh open, and a "bearish silver short" (SLV) was a no-view conversion. Dollars moving is not exposure being created; a big premium is not a signal to chase. Wait for the catalyst, check whether open interest confirms new money, and never mistake someone else's hedge for your signal. Provisional flags (most of today's legs are size<OI) resolve with tomorrow's pre-market snapshot.

Not investment advice. Options carry risk of total loss. Do your own research.