The thesis under review: agents run on cloud CPUs, inference silicon and agent platforms as much as on GPUs; AWS (Graviton5, Trainium, Bedrock) is where that demand lands; Amazon's stock has lagged the AI rally and is 13% below its August record, so a strong third quarter and re:Invent take it back to ~$290 within three months. We test each link, read the chart, and price the bullish call and put structures that pay if it happens.
Update, 30 October 2026, after the third-quarter report and the open. Amazon reported on Thursday evening: AWS +39% to $46.0 billion at a 38.7% operating margin, its fastest growth in five years, with a $578 billion backlog and growth guided above 40% for both 2027 and 2028; operating income $28.1 billion (+61%, a record 13.9% margin); and the market first sold the cost of it. Capex was $75.9 billion in the quarter ($210.7 billion over twelve months, about $220 billion for 2026 and "meaningfully higher" in 2027), free cash flow −$33.5 billion (−$42.4 billion trailing), advertising grew 11% against a $20.8 billion consensus, the fourth-quarter operating-income guide of $28.5–33.5 billion sits $2 billion under consensus, and headline EPS of $1.89 compares with a year-ago $1.95 that contained a $9.5 billion Anthropic gain (this quarter's other income was −$1.2 billion; the stake is next marked at the listing in November, upward). Pre-market the stock was indicated at $225, on the June–July lows; it opened at $232.53 (−4.6%), printed $229.36 in the first minute and was $239.62 (−1.7% on the day) within fifteen minutes: the dip toward the shelf was bought before it got there, with the sell-side unanimous that the AWS number is the one that matters (Morgan Stanley $335 from $350, Goldman $370, Evercore $355, Bank of America $317). The recommendation below is rewritten with live quotes from 14:52 CET: the core strike is the March $225, with the break-even under the shelf; the pre-market version of this page had moved it to $215 for a stock indicated at $225, and the strike moves back up with the price. Tiles, strategy table, chain and charts are live from Interactive Brokers; dealer positioning uses Thursday's open interest (footnote under the chart). The 26 October note that follows is kept for the record.
Update, 26 October 2026. This report was written on 27 September for a December target; the author now wants to be paid to own Amazon with a put expiring in February or March 2027, funded by a cash bonus, and the data and the trade below are refreshed for that. Where the stock is: $250.65 at midday on Monday, back at the bottom of the $244–268 range it has held since mid-September after a 4.6% fall on 22 October with the AI-hardware sell-off (Thursday's $248.00 low sits on the 16 and 28 September lows at $244–245 and the dealers' $240 put wall). The Q3 report is on Thursday 29 October after the close: guidance is $197–202 billion of revenue and $22.5–26.5 billion of operating income, consensus $202 billion and $1.96 of EPS, AWS expected at about +35% after +37% last quarter, and 2026 capex was lifted to about $220 billion in July. Implied volatility is 44% (74th percentile; the options price the report at about ±7%), which is the reason the long-dated puts pay what they do this week. Anthropic has filed to list (Amazon owns an estimated 15–20%); AWS says capacity is sold through 2027. The long-dated put sale is in the new recommendation; the January structures below it are the original September trade, left for the record. Data refreshed 28 October before the open (Interactive Brokers, Amazon $253.77 after Tuesday's +1.2%): the tiles, the strategy table, the chain and the charts are live; the prices quoted in the recommendation's prose are Monday's close; after the rally the March $225 put is about $9.95 (break-even $215.05) and the February $225 about $8.55.
Our view in one paragraph. The first link holds: Meta, Snowflake, Uber, OpenAI and Anthropic have all tied Graviton and Trainium spending explicitly to agent workloads, AWS grew 36.7% last quarter (its fastest in 18 quarters) with a $496 billion backlog, and Andy Jassy says capacity is sold out through 2027. The second link is the problem: the market saw that print, paid +15% for it on 31 July, and has since sold Amazon down 13% on $220 billion of capex, negative free cash flow, a new FTC advertising suit and 5.2% Treasury yields. What Amazon has going for it is exactly what AMD lacks: sentiment is washed out, the consensus target ($330) sits a third above the price, the stock is the cheapest hyperscaler on EV/EBITDA, and the chart is in an orderly pullback to its 200-day average rather than a blow-off. $290 is a retest of the August high; the market prices a chance and we would say one in four, because the 29 October report is a coin flip between an AWS-acceleration gap up (+10–15%) and a capex-shock gap down (−8%). That asymmetry favours a structure that owns the upside cheaply and gets paid to wait: a Jan-2027 260/290 call spread funded three-quarters by a 235/210 put spread, for a net outlay of about $2.30 a share and a maximum gain of 101% on the capital at risk.
This is the best-evidenced link in either of our reports. In April Meta signed a multi-year, multibillion-dollar agreement for "tens of millions" of Graviton5 cores explicitly for "orchestrating multi-step agent tasks", making it a top-five Graviton customer; Snowflake raised its AWS commitment to $6 billion anchored on Graviton for agentic workloads; Uber moved trip-serving to Graviton4 and is piloting Trainium3. Graviton5 is used by 98% of AWS's top 1,000 customers and revenue commitments for it tripled quarter-on-quarter. On the inference side, Anthropic committed more than $100 billion over ten years and up to 5 GW of Trainium (about 1 GW online by year-end, more than a million Trainium2 chips already running), and OpenAI expanded its AWS deal to $100 billion over eight years including 2 GW of Trainium, with Amazon investing $50 billion. In August Nvidia added 2 million more Blackwell Ultra and Rubin GPUs plus Vera CPUs "for agentic and physical AI" to the AWS pipeline.
The operating numbers agree. AWS revenue growth has accelerated for six straight quarters, from 17% to 36.7% ($42.2 billion in Q2, 39.4% operating margin). AI services and custom chips each run above $25 billion annualised and both are growing at triple digits. Bedrock processed more tokens in the first quarter than in all prior years combined, the AgentCore SDK passed 2 million downloads in five months, Kiro usage tripled in a quarter, and the backlog doubled in six months to $496 billion. Jassy on the call: "Even at $220 billion, we will still not have enough capacity to meet all the demand we have in 2026, and I believe this dynamic will also be true in 2027."
Verdict: true, quantified and directly tied to agents by the customers themselves. The caveat is that it is not exclusive: the same demand is landing at Google Cloud (+82%), Azure (+43%, guiding 45%), Oracle, CoreWeave and a growing list of neoclouds. AWS's share of the cloud market has drifted from 32% in 2021 to 28%.
Here the chain frays. Amazon delivered the best AWS quarter in years on 30 July and the stock jumped 15% to a record $284 on 3 August. Since then it has given back 13% while the Nasdaq made new highs. The reasons are specific: 2026 capex raised to ~$220 billion (from $200 billion; the original $200 billion guide caused an 8% drop in February), trailing free cash flow at −$7.6 billion for the first time since 2023, an FTC and 22-state lawsuit (31 August) alleging a hidden ad-auction "soft reserve price" that extracted $20 billion from advertisers, a Teamsters strike at its largest warehouse, confirmed permanent loss of customer data in its Bahrain region, a $1.9 billion delivery-driver pay plan, and a Street-low $230 target from Rothschild Redburn arguing that generative AI is "dilutive to returns". Add a Fed that hiked in September and a 10-year yield above 5.2%, and the market is currently paying for visible consumer AI products and picks-and-shovels, not for capex-absorbing platforms.
Verdict: the weakest link. A re-rating needs AWS to accelerate again on 29 October and, more importantly, a 2027 capex message the market can live with ("high-$200-billions, largely pre-reserved by customers" rather than a step-function surprise).
Amazon is up this year against for the Nasdaq-100, and roughly in line with Microsoft and Alphabet. It trades at 26.6× the next twelve months' earnings and about 24× 2027 consensus EPS of $10.48, with an EV/EBITDA of 16.7× against 20× for Microsoft and 23.6× for Alphabet: the cheapest of the three despite the fastest-accelerating cloud. Fifty-nine analysts carry an average target of $329.54 (range $230–$405) with no sell ratings, and September brought target increases from Wells Fargo ($338), Pivotal ($333), Bernstein and BofA ($320) alongside the Redburn cut. The Anthropic stake, marked at $190 billion on 30 June, faces an IPO in mid-to-late October at a reported ~$2 trillion valuation, which would add another nine-figure non-operating gain and reinforce the "Amazon owns the winner" narrative just before the earnings call.
Verdict: the valuation gap versus peers and the washed-out sentiment give the trade a better entry than the AMD thesis has; they do not by themselves make the stock go up.
$290 is from Friday's close: a retest of the 3 August intraday high of , about 0.75 of a standard deviation over the period at today's implied volatility, and a market value of about $3.1 trillion. On consensus it is 27.7× 2027 earnings, still below Alphabet's multiple. The path is (i) Q3 AWS growth at or above 38% with the backlog above $550 billion and a Q4 operating-income guide with a midpoint of $30 billion or more; (ii) 2027 capex framed as pre-sold capacity rather than a surprise; (iii) the Anthropic IPO pricing well; (iv) re:Invent (30 Nov–4 Dec) delivering Trainium4 and agent metrics, historically a positive window for the AWS narrative. Two of the last four earnings reports produced +9.6% and +15.3% moves on AWS acceleration; the other two produced −8% moves on capex shocks.
What confirms it: a post-earnings close above $270 (the September breakdown level) that holds for a week; backlog growth still above 100%; consensus 2027 EPS moving above $11.
What confirms it: a post-earnings close below the 200-day average (~$241) and the September low ($244); a Q4 operating-income guide below $25B; capex commentary that surprises to the upside.
| Date | Event | Why it matters for the thesis |
|---|---|---|
| 6–7 Oct | Prime Big Deal Days | First read on the holiday consumer; watch Alexa for Shopping and ad-load commentary. |
| ~mid–late Oct | Anthropic IPO (Nasdaq, ~$2T reported) | Marks Amazon's $190B stake; sentiment on "Amazon owns the winner" days before earnings. |
| Late Oct–Nov | EU DMA final decision on AWS gatekeeper status | Egress, interoperability and self-preferencing obligations; headline risk for AWS margins. |
| 27–28 Oct | FOMC; Microsoft and Meta results (~28 Oct) | Rates, and the bar Azure sets (guide 45%) the day before Amazon reports. |
| ~29 Oct | Amazon Q3 results and Q4 guide | The make-or-break event: AWS growth vs 36.7%, backlog vs $496B, Q4 operating-income guide, first 2027 capex signal. |
| 3 Nov | US midterm elections | Tariff and regulatory path. |
| ~17 or 25 Nov | Nvidia Q3 FY27 results | Rubin/Vera cadence; sets the sector multiple. |
| 27–30 Nov | Black Friday to Cyber Monday | Holiday-quarter read on retail and ads. |
| 30 Nov–4 Dec | AWS re:Invent | Trainium4, Anthropic capacity, agent-platform metrics; historically a positive narrative window. |
| 8–9 Dec | FOMC with dot plot | Last macro event before the target date. |
| 18 Dec | December options expiry | Last monthly expiry inside the window. |
| 24 Dec | Target date | Three months from the date of this report. |
| Q4 (TBD) | Amazon Leo consumer launch; Section 232 "Phase Two" chip/server tariff | Cost and capex risk on the infrastructure side. |
Daily closes from Interactive Brokers. Hover for values; the table shows the same data.
| Total return | 1 week | 1 month | 3 months | 6 months | YTD | 1 year |
|---|
Amazon has spent 2026 tracking Microsoft and Alphabet rather than the index, and over the last month it is the only one of the four that is down. The relative-strength argument is a mean-reversion one: the stock has lagged an index it is 5% of, during a rally driven by the AI infrastructure it sells.
Blue candles closed up, red closed down. Orange dashed: dealer-gamma levels (call wall, put wall, gamma flip); swing support and resistance are listed in the key-levels table. Hover for the crosshair: the price is tagged on the price axis and the date on the time axis. Right panel: net dealer gamma per strike on the same price axis (blue where dealers are long gamma, red where short).
Trend and candles: an orderly pullback that has compressed into a two-week box. The six-month chart shows higher lows from the March low ($199) to the June–July lows ($226), the 31 July earnings gap to $270–287, and a seven-week retreat that has now retraced to the rising 200-day average. The last two weeks are a compression: a sell-off to $244.30 on 16 September, a recovery to $258–259 on 21–22 September, then a 23 September red candle (−$4 on 24.9 million shares, the heaviest volume in a month) answered the next day by a green reversal off $245.60. Bodies are small (four of the last twelve candles have bodies under 10% of their range) and Friday's close of sits in the middle of the box: the market is undecided between $244 and $259.
Levels. Overhead: $258 (the 28 August swing high, with the 20- and 50-day averages at $254–256 just below it), $268 (the level broken in September and the call spread's break-even), $277 and the $287 record. Below: $244 (the 16 September low), the 200-day at $241, the June–July lows at $226, and the March low near $199–202 as the floor of the year's range. The dealer-gamma levels bracket the box almost exactly: the $260 call wall just above it and the $240 put wall just below.
Volume. The pullback has a mild distribution signature: the down days (15, 16 and 23 September) came on above-average volume and the up days on below-average volume, and the 20-day average has crept up to shares. The stock is being sold, not ignored. A close above $259 on more than 20 million shares would say the sellers are finished; a close below $244 on the same volume would say they are not.
Technical verdict: constructive for a bull trade. A pullback to the 200-day with higher lows intact, small candles at support, a dealer-supported range of $240–260 and the put spread's short strike sitting at the gamma flip. Buy the call spread now and sell the put spread into the earnings run-up. A close below $236 turns dealers short gamma and makes $226 the next stop; that is the technical stop for the thesis.
| Level | Price | vs spot |
|---|
Support and resistance are clustered swing lows and highs from the last eight months of daily bars; the gamma levels come from CBOE open interest.
What the profile says for Amazon. Dealers are long roughly of gamma per 1% move, in the next 35 days, and the book is lopsided: the $260 strike carries $129 million of call gamma, the single largest strike on either of our reports, with $250, $255, $265 and $270 all large; the biggest put strikes are $240 ($69 million) and $245. In positive gamma dealers sell rallies and buy dips, so $240–260 is a hedging-supported box: rallies stall into $260 and dips get bought at $240–245, which is exactly what the last two weeks look like. The book flips negative below about $236: under that level dealer hedging accelerates a decline instead of cushioning it, which is the argument for capping the put spread with the long $210 put rather than selling a naked $235 put. For the trade, expect the $260 wall to hold into the 16 October expiry; breaking it needs the 29 October earnings gap, after which October's call gamma has expired and the wall dissolves.
How to read this. Market makers who sell options hedge them with stock; how much stock they must buy or sell as the price moves is their gamma. We estimate it from CBOE open interest for every listed strike and expiry, recompute each option's gamma with Black-Scholes from its own implied volatility, and sign it by the usual convention: dealers are long the calls customers sell (covered calls, spreads) and short the puts customers buy for protection. Positive net gamma means dealers sell as the stock rises and buy as it falls, dampening moves; negative means they chase the move and amplify it. The "walls" are the strikes carrying the most call or put gamma near spot, where hedging flows are strongest; the "flip" is the price below which the whole book turns negative. Two caveats: the sign convention is an assumption (a customer who buys calls outright leaves the dealer short that gamma), and open interest is end-of-day, so the profile changes after every expiry, most of all the third Friday of the month.
Implied volatility is against realized volatility of : options are priced a few points above how the stock has actually moved, which is normal with an earnings date inside the window. Over the past year IV sits at the th percentile (range –), so premium is neither cheap nor rich. That is the case for a spread: buy and sell volatility at the same time, rather than pay for it outright.
The one-standard-deviation move to the target date is ±, a range of roughly $211–$289, which puts $290 right at the top of the cone and gives a risk-neutral probability of . Amazon's average absolute earnings-day move over the last eight reports is 7%, the last four were +15.3%, +4.1%, −8% and +9.6%, and the market currently implies about ±7–8% for 29 October. Expect a 4–6 point drop in implied volatility after the report; the recommended structures are short as much premium as they are long, so they are largely insulated from it.
| Structure | How it is built | What you want | Risk | Best used when |
|---|---|---|---|---|
| Long call | Buy a call | A big move up, fast | The premium (100% if it expires below the strike) | You expect a move larger than the market prices; implied volatility is low |
| Bull call spread (debit) | Buy a call, sell a higher call | The stock at or above the upper strike at expiry | The net debit | You have a price target and a date; you want the highest return per dollar to that target |
| Cash-secured short put (credit) | Sell a put, hold the cash to buy the shares | The stock to stay above the strike; you keep the credit | Owning the shares at the strike in a sell-off (large, stock-like) | You would be happy to own the stock lower; volatility is high |
| Bull put spread (credit) | Sell a put, buy a lower put | The stock above the short strike at expiry; keep the credit | Spread width minus credit (defined) | You are bullish-to-neutral and want to be paid to wait with a known worst case |
| Risk reversal | Sell a put, buy a call (often for zero cost) | A large rise; behaves like stock above the call strike | Stock-like below the put strike; margin required | You want stock exposure without paying premium and accept the downside |
| Put-financed call spread | Sell a put, buy a call spread | The stock at the top of the spread; the put pays for it | Stock-like below the put strike | You want a call spread for free and accept assignment risk |
| Put spread + call spread | Sell a put spread, buy a call spread | The stock above the call spread's top strike | Both spreads' widths minus net credit (defined both ways) | You want to be paid to wait and keep a leveraged payout at the target, with the worst case known |
Returns on the short-put structures below are shown on a cash-secured basis (capital = strike minus credit). Brokers typically require about 20% of that as margin, so the return on margin is roughly five times the figure shown, with the same dollar risk.
The brief is to profit as much as possible if Amazon is around $290 on 24 December, while using structures that suit a stock with 31% volatility, a binary earnings event and a target only 0.75 standard deviations away. We priced every $5-strike call from $215 up and every put from $190 to $260 in the December, January and February monthlies from Interactive Brokers' closing quotes and model Greeks, built the seven structures above from them, and valued each on the target date under four scenarios with the implied-volatility change each scenario would bring.
Scenarios on 24 December: Bear $212.50 (−15%, the bottom of the one-sigma cone and a capex-shock reaction) · Flat $250 · Base $267.50 (+7%, a solid quarter) · Bull $290 (+16%, the thesis).
| Trade | Net | Max loss | Break-even | Delta | PoP |
|---|
Net is the mid-price debit (paid) or credit (received) per share; one contract is 100 shares. Max loss is the capital at risk per share (for a short put, the strike minus the credit). PoP is the lognormal probability, at today's implied volatility, of finishing above break-even at expiry. Scenario cells show profit or loss as a percentage of the capital at risk on 24 December; −100% means a total loss. Spreads valued before expiry show less than their maximum because the short leg still has time value; held to expiry with the stock at or above the top strike, every call spread pays its full width and every put spread keeps its full credit. Highlighted rows are the recommendation.
| Trade | Capital at risk | P&L | Return |
|---|
Pick up to four trades. Values on the target date use Black-Scholes with each contract's own implied volatility, shifted −5 points for prices above spot and +5 points below. Note the risk reversal's line keeps falling to the left: its downside is the stock's.
Thursday's report set the reaction the 26 October plan was waiting for. Pre-market it looked like −8%; the market opened Amazon at $232.53 (−4.6%), touched $229.36 and bid it back to $239.62 (−1.7% on the day) inside a quarter of an hour. It had printed the best AWS quarter in five years and was sold for the capital it takes to deliver it ($75.9 billion in the quarter, free cash flow −$33.5 billion, 2027 "meaningfully higher"), then bought because the AWS number is the one the stock has traded on all year. For a put seller content to own the stock the picture has three parts. The level: the June–July lows at $225.55, $226.12 and $226.16 (25–26 June and 29 July), a shelf tested three times and not reached today; under it the chart is empty to the February–March lows at $196–200. The positioning: Thursday's open interest re-priced at $239.62 makes the book +$102 million per 1% (positive: dealers sell rallies and buy dips, which dampens), with the flip at $238.4 just under the price, the put wall at $230 and the call wall at $250; pre-market, at $225, the same book read −$196 million with the flip above the price, which is how far fifteen minutes moved the hedging picture. Today's heavy put volume reshapes the strikes when Monday's open interest prints. The history: every gap of 4% or more lower in the last two years (April 2025, August 2025, February 2026) was higher twenty and sixty days later, after a further 2–10% dip from the gap-day close ($161 after $178; $211 after $215; $196 after $210); a 10% dip from here is about $216, and the break-even of the $225 put sits below it.
What changed in the pricing is the stock, not the volatility: March implied volatility is 34% (38% before the report); the thirty-day figure has collapsed from 45% pre-market to 31% as the weeklies' event premium came out. At $225 pre-market the March $225 put was indicated at $16; at $239.62 it is $12.03, still well above Monday's $10.90, because the strike is 5% below the price instead of 10%. That is the honest framing of "well priced": the premium is larger than it was, and so is the delta. The option prices in this card are live quotes from Interactive Brokers at 14:52 CET, twenty minutes into a session that has already moved 4%; use them as the level to work, not as fills.
Is the bought dip a good sign? Not on the record. In Amazon's sixteen post-earnings sessions since 2016 that opened 3% or more lower, the eight that closed above the open (today's pattern) went on to a median +0.3% after twenty sessions and +6% after sixty, against −0.3% and +3.7% for the eight that closed below the open: no edge from the rescue itself. Across twelve mega-caps (157 such sessions since 2015), the bought gaps did slightly worse over the next month (median −2.2%, positive 39% of the time) than the sold ones (+0.8%, 53%). What the sample does say is that the day-one low is revisited: in 75% of Amazon's cases and 79% of the wider sample the first day's low was undercut within twenty sessions, by a median 4% (from $229.36 that is about $220); in six of Amazon's sixteen the stock traded more than 7% under the day-one low within a month (2016, the 2018 and 2022 sell-offs, early 2023), which is where the $225 put's break-even sits. For a put seller that is the useful reading: the bounce is not a reason to hurry, the retest is the better entry, and the strike should be chosen for the retest, not for today.
Own the upside with a call spread whose top strike is the target, and pay for most of it by selling a put spread below every nearby support. Use the January expiry: it costs almost the same as December, covers earnings, re:Invent and the target date with three weeks to spare, and lets the short legs keep decaying into year-end.
| Strike | Bid | Ask | Mid | Break-even | IV | Delta | Theta/day | Volume |
|---|
26 October refresh: prices from Yahoo Finance and option quotes, Greeks, open interest, term structure and dealer gamma from the CBOE delayed chain at 09:35 New York time on 26 Oct 2026 (IB Gateway was not running; quotes are delayed 15 minutes and the far-dated puts are $1–2 wide, so work the mid); Q3 guidance, consensus and the capex figure from Amazon's Q2 release and previews of 20–25 Oct 2026 (Pepperstone, MarketBeat, 24/7 Wall St, Seeking Alpha); Anthropic listing and AWS capacity: Yahoo Finance and 24/7 Wall St, 21–25 Oct 2026. Original report: Prices, implied volatility, option quotes and Greeks: Interactive Brokers via IB Gateway, fetched (market closed; quotes are Friday's close). Technicals computed from daily bars; option values from Black-Scholes at a 4% risk-free rate. Fundamentals and news: Amazon Q2 2026 results, release and call transcript (30 Jul 2026), Amazon Q1 2026 commentary (30 Apr 2026), AWS press releases on Meta/Graviton5 (24 Apr 2026), Nvidia (26 Aug 2026), Uber (7 Apr 2026); Anthropic–Amazon compute announcement (20 Apr 2026); OpenAI–Amazon partnership (27 Feb 2026); Synergy Research Q2 2026 cloud market data (30 Jul 2026); Microsoft and Alphabet Q2 2026 results; StockAnalysis consensus and statistics (25 Sep 2026); TipRanks/The Fly on Rothschild Redburn (21 Sep 2026); FTC press release on the ad-auction suit (31 Aug 2026); European Commission DMA preliminary finding (25 Jun 2026); MLex on the FTC trial date; Federal Reserve statement (16 Sep 2026); Census retail sales (16 Sep 2026); University of Michigan sentiment (Sep 2026); AlphaQuery volatility statistics (25 Sep 2026); Earnings Watcher and OptionSlam on historical earnings moves; CNBC and Reuters on the Anthropic IPO timing (5 Sep 2026); AWS re:Invent agenda.