Amazon to $750: A Compounder Thesis for 2030

This is a four-year compounder thesis on a company I have held for over eight years and continue to hold and add to. The case rests on three accelerating businesses inside one parent that the market is still pricing like a maturing retailer.

Summary

Amazon (AMZN) closed at $272.58 on May 8, 2026, with a market capitalization of $2.93 trillion. The market has spent the last eighteen months pricing Amazon as a mature retail business with a slowing cloud unit. That view is now wrong on the facts, and the data from the last six quarters has flipped the burden of proof.

This memo lays out the case for $750 per share by year-end 2030, approximately a 24% annualized return from current levels. The path runs through three converging drivers: AWS re-accelerating into a 25–30% growth rate as AI workloads convert from talk into bookings, an advertising business now generating $70 billion in annual revenue at 40%-plus margins that the market still treats as a rounding error, and the May 4, 2026 launch of Amazon Supply Chain Services, which opens the company’s logistics infrastructure to third parties and tore $20 billion of market cap out of UPS and FedEx in two days.

The $1,000 case (roughly $11 trillion market cap) is real but requires a P/E re-rating on top of operational bull-case execution. That re-rating could happen if the broader market re-prices megacap tech upward. It is not the base case.

The most credible bear case is not company-specific. It is that hyperscaler capex of $416 billion in 2025 and an estimated $685 billion in 2026 produces returns that disappoint investors, leading to multiple compression across all four hyperscalers regardless of operating performance. This memo engages that risk directly.

Introduction

Amazon reports three segments: North America, International, and AWS. That structure obscures what is actually happening inside the business. In 2025, Amazon generated $716.9 billion in revenue and roughly $80 billion in operating income—but those headline numbers mix a $128 billion cloud business growing 20%-plus, a $68 billion advertising business growing 22%, a $520 billion retail engine on razor-thin margins, and a newly launched logistics-as-a-service business that took $20 billion of market value out of UPS and FedEx in two days.

The thesis of this memo is that consensus is anchored to an old story—mature retail, decelerating AWS—and the new story is already in the data. Six straight quarters of AWS reacceleration. A $69 billion advertising business that the average sell-side model still rounds to a margin assumption inside North America. A logistics product the company has spent five years and tens of billions of dollars in fulfillment infrastructure preparing for, and has now turned on. The price action since the Q1 2026 print and the Amazon Supply Chain Services launch suggests the market is starting to notice. The work below is an attempt to size what that means by year-end 2030.

Company Overview

In 2025, Amazon generated $716.9 billion in revenue, up 12% year over year. AWS contributed $128.7 billion of that (up 20%), North America $426.3 billion (up 10%), and International $161.9 billion (up 13%). Operating income reached an estimated $80 billion, with AWS contributing roughly $46 billion at a 35.4% margin.

The advertising business sits inside North America and International, but Amazon discloses it separately. It generated $68.6 billion in 2025, up 22% from $56.2 billion in 2024. At an estimated 40% operating margin, advertising is contributing somewhere between $25 billion and $30 billion in operating income that the consolidated reporting buries inside the retail segments.

If you reconstruct the income statement by economic reality rather than reporting convention, Amazon in 2025 was: a $128 billion cloud business growing 20%-plus at 35%-plus margins, a $68 billion advertising business growing 22% at roughly 40% margins, and a $520 billion core retail and marketplace business growing 7–9% at razor-thin margins. The high-margin businesses are roughly 27% of revenue but the substantial majority of operating profit and almost all of the marginal economics.

The May 4, 2026 launch of Amazon Supply Chain Services creates a fourth pillar. It opens Amazon’s warehouse, freight, ocean, and parcel infrastructure to third-party shippers. Early customers include 3M and Lands’ End. UPS fell 10.5% the day of the announcement. FedEx fell 9.1%, its worst single day in over a year. The market understood immediately what the company is becoming.

Driver One: AWS Has Re-Accelerated

For most of 2024 and the first half of 2025, AWS grew in the 17–20% range. Analyst consensus was that the business had matured and that Azure and Google Cloud were quietly taking share. Reasonable view at the time. Wrong now.

AWS revenue has grown sequentially in each of the last six quarters: $28.8B, $29.3B, $30.9B, $33.0B, $35.6B, $37.6B. The year-over-year growth rate (ex-FX) over that same period: 19%, 17%, 17%, 20%, 24%, 28%. On the Q1 2026 call, Andy Jassy described it as “AWS is growing 28%, our fastest growth in 15 quarters, on a very large base.”

This is not a one-quarter blip. It is six consecutive quarters of sequential dollar growth and four consecutive quarters of year-over-year acceleration. AWS backlog (the contracted future revenue Amazon discloses in its 10-Q) was $244 billion at the end of Q4 2025, up 40% year-over-year. That is forward visibility, not a guess.

The skeptic’s response is that Azure grew 40% and Google Cloud grew 63% in the same quarter, so AWS is still losing relative share. That is true. It is also less damaging than it sounds.

Two reasons. First, the base sizes are radically different. AWS generates an estimated $150 billion in annualized revenue. Azure generates roughly $100 billion. Google Cloud generates roughly $50 billion. Growing 28% off a $150 billion base produces more incremental dollars in absolute terms than growing 63% off a $50 billion base. Second, the operating margin reality. AWS produces an estimated $50–55 billion in annual operating profit at a 35–38% margin. Google Cloud just turned its first full year of operating profit in 2024 at an estimated 12–14% margin. Azure does not disclose standalone margins, but consensus puts them below AWS. AWS still produces more cloud profit dollars than the next two combined.

There is also a chip story embedded inside AWS that the market is undervaluing. Amazon disclosed in Q1 2026 that its silicon business (Graviton CPU, Trainium AI accelerator, Nitro) reached a $20 billion annual revenue run rate, growing at triple-digit percentages year over year. Amazon landed 2.1 million AI chips over the trailing twelve months, more than half of which were Trainium. The company will deploy over one million NVIDIA GPUs starting in 2026.

What this means in plain terms: Amazon is the only hyperscaler with credible silicon optionality. Microsoft runs Maia and Cobalt, but both are smaller and earlier. Google has TPU, which is mature but is essentially Google-internal. Trainium is being sold to external customers (Anthropic is the most visible) and is generating real revenue. This reduces AWS dependence on NVIDIA in a way Microsoft cannot replicate, which structurally protects margins as compute mix shifts from training to inference over the next three years.

Driver Two: Advertising Is Hidden in Plain Sight

Amazon’s advertising business will exceed $80 billion in 2026. Most retail analysts cover Amazon. Most do not write about ads. That gap is the opportunity.

The growth trajectory in numbers: $20 billion in 2020, $31 billion in 2021, $38 billion in 2022, $47 billion in 2023, $56 billion in 2024, $69 billion in 2025. The 22% growth rate in 2025 was actually an acceleration from the 21% rate in 2024. Q4 2025 advertising revenue of $21.3 billion was larger than Amazon’s entire revenue in Q2 2012. It took the rest of Amazon thirteen years to grow to where the ad business now sits as a TTM run rate.

Three things make Amazon advertising structurally different from Google search ads or Meta social ads, and all three are underappreciated.

Conversion at the point of intent

Amazon ads target consumers at the point of purchase intent. A “blender” search on Amazon is a buyer looking for a blender. A “blender” search on Google may be that, or it may be someone curious about smoothies. Amazon’s bottom-funnel position commands premium CPCs and predictably high conversion rates, which advertisers will keep paying for as long as the conversions hold up.

Prime Video CTV inventory

Amazon launched ads on Prime Video in January 2024 with the default tier serving ads to all subscribers unless they pay extra. That decision instantly added the largest CTV inventory pool outside of Netflix, attached to roughly 200 million Prime subscribers. CTV advertising CPMs are 5–10x mobile display CPMs. The ramp has been faster than analysts modeled.

Closed-loop measurement

Amazon can show advertisers exactly what they got for their spend, because Amazon owns the entire path from impression to purchase. Google can do this for shopping. Meta can do this for some direct response. Amazon does it natively for everything, which is why brand budgets keep migrating from Google Search and Meta into Amazon. This is structural and gets stronger with scale.

The right reference point for valuing Amazon ads is not Google’s media-and-search business at 22x. It is Meta at 26x, or the digital ad pure plays like Trade Desk at 60x. Amazon ads runs at margins comparable to Meta’s family-of-apps business (mid-to-high 40s operating margin per analyst estimates) and is growing faster.

The advertising business alone, in bull case 2030, is worth nearly as much as Amazon’s entire current market cap.

Driver Three: Logistics Monetization Just Started

On May 4, 2026, Amazon announced Amazon Supply Chain Services (ASCS). The pitch is simple. Companies that ship products can now plug into Amazon’s logistics infrastructure—warehouses, freight forwarding, ocean freight, parcel delivery—as a third-party service. The first announced customers were 3M and Lands’ End. The market reaction was instant: UPS down 10.5% on Monday, FedEx down 9.1%, GXO Logistics down double digits, Forward Air down double digits, Old Dominion down 7%.

That market reaction is the single most important data point for this thesis, and it happened a week ago. Roughly $20 billion of market value evaporated from the comp set.

That price action is the market’s first-take estimate of what ASCS is worth, and it is almost certainly an under-pricing of the structural shift.

Amazon has been preparing this launch for years. The logistics buildout that crushed free cash flow in 2021–2022 created an infrastructure footprint with utilization that runs well below capacity outside of peak shopping windows. Sweating that infrastructure with third-party volume is the same playbook AWS ran with cloud computing in 2006–2010. AWS was originally built to run Amazon’s own e-commerce. Once the operating system was proven internally, opening it to outside developers turned a cost center into the highest-margin business in the company.

The same arithmetic applies here. Amazon’s North American 1P shipping costs are estimated at $80–90 billion annually. Selling spare capacity at any positive margin creates marginal profit dollars on infrastructure already paid for. The medium-term TAM is large. UPS and FedEx combined produce $179 billion in revenue. Even at 10% share capture of just the parcel piece, ASCS becomes a $20 billion revenue business. Add ocean freight, freight forwarding, and warehousing-as-a-service, and the credible 2030 range for ASCS revenue is $40–80 billion.

The retail segment also benefits in a way the segment reporting will not show clearly. Core retail (NA + International ex-advertising) generated roughly $520 billion in 2025 revenue at an estimated 1–2% operating margin. Logistics utilization improvement and 3P seller mix expansion lift that margin. By 2030, an operating margin of 4–5.5% is achievable on a $720–780 billion revenue base.

This is the segment most analysts have been most wrong about, and where the next twelve to twenty-four months of disclosure (Amazon will eventually break out ASCS revenue, probably in 2027 or 2028) will force a reappraisal.

The Magnificent Re-Rate

Quick aside on the broader market backdrop, since it affects whether $750 or $1,000 is the right number.

If Apple, Alphabet, Microsoft, Nvidia, and Amazon all reach $10 trillion market cap by 2030, the Magnificent 5 combined would be worth $50 trillion, up from roughly $19 trillion today. The S&P 500 currently sits at approximately $67 trillion in total market capitalization. The Mag 5 reaching $50 trillion would put them at around 75% of the current index, which mechanically requires the broader index to expand substantially. Either the Mag 5 thesis is wrong, or the S&P 500 is heading meaningfully higher over the next five years.

I think the second is the more likely scenario. AI-driven productivity, if real, accrues most heavily to the largest enterprises that have the scale and capital to deploy it. The S&P 500 doubling over the next five years would imply roughly a 15% CAGR, above the long-term historical average of approximately 10% but in line with the post-2020 actual experience. Aggressive but not unprecedented.

That backdrop matters for the $1,000 case specifically. At bull-case 2030 operating income of $341 billion, the implied EV/OI multiple at each price target is shown below.

$750 requires a 24x multiple, below Amazon’s current 33x P/E and roughly in line with Meta’s 26x today. Defensible without re-rating. $1,000 requires a 32x multiple, almost exactly Amazon’s current P/E and below Microsoft’s 35x. If the broader market re-rates upward, $1,000 is reachable. If multiples compress, $750 is still defensible.

The framing I would adopt: $750 is the base case, achievable on operational execution alone. $1,000 is reachable in the upside scenario where operations meet bull case and the macro environment supports current or expanded megacap multiples.

Valuation: Sum of the Parts

The sum-of-the-parts valuation:

Scenario	AWS	Advertising	Retail + ASCS	Other	EV	$ / Share
Bear	$1.72T (20x · 33%)	$0.89T (18x · 38%)	$0.21T (12x · 2.5%)	—	$2.82T	$252
Base	$3.10T (24x · 38%)	$1.73T (22x · 45%)	$0.56T (16x · 4%)	$0.16T	$5.55T	$500
Bull	$4.32T (27x · 40%)	$2.75T (26x · 48%)	$1.15T (20x · 5.5%)	$0.39T	$8.61T	$783

Bear case ($2.82 trillion, $252/share): AWS revenue $260B at 33% margin, 20x multiple. Ads $130B at 38% margin, 18x. Retail $640B at 2.5% margin, 12x. ASCS $20B at 10% margin. This is a flat-to-down scenario where Azure and Google Cloud meaningfully erode AWS share, advertising growth decelerates, and ASCS underperforms.

Base case ($5.55 trillion, $500/share): AWS revenue $340B at 38% margin, 24x. Ads $175B at 45% margin, 22x. Retail $720B at 4% margin, 16x. ASCS $40B at 15%. This represents a continuation of current operational momentum.

Bull case ($8.61 trillion, $783/share): AWS revenue $400B at 40% margin, 27x. Ads $220B at 48% margin, 26x. Retail $780B at 5.5% margin, 20x. ASCS $80B at 18%. Operations execute against the upper end of plausible outcomes.

The bull case lands at $783 per share, almost exactly at my $750 price target. The base case ($500) sits below the target but represents 13.9% annualized returns, which is roughly double the long-term equity market return.

The $1,000 target requires bull-case operations and a multiple environment that sustains AMZN at roughly its current 32x P/E. If that combination holds, $10.75 trillion is reachable. I assign roughly 25–30% probability to the $1,000 outcome and roughly 50–55% probability to $750 or above.

A reference point worth noting: at current revenue mix and growth rates, I project Amazon to generate roughly $250 billion in operating income by 2030 in the base case. Apple’s current operating income is roughly $125 billion. Microsoft’s is roughly $130 billion. By 2030, Amazon may be the most profitable company in the world by operating income, even if the market does not yet appreciate that fact in 2026.

The Bear Case Worth Taking Seriously

The standard bears—antitrust, retail margin compression, Azure share gains—get more airtime than they deserve. The real risks are different.

Risk 1: Capex returns disappoint and the entire hyperscaler complex de-rates

This is the most credible bear case and the one this memo needs to engage directly.

The combined capex of Amazon, Google, Meta, and Microsoft was $416 billion in 2025 and is guided to approximately $685 billion in 2026. Amazon’s portion alone went from $83 billion in 2024 to $132 billion in 2025, and management has guided to roughly $200 billion in 2026. Amazon’s free cash flow collapsed from $38 billion in 2024 to $11 billion in 2025 as a direct result.

The bull frame is that this is AWS’s 2006–2010 playbook scaled up. Heavy investment now produces a multi-year revenue and margin tailwind once the infrastructure is depreciating against utilization. The bear frame is Cisco in 1999–2000. A profitable, growing business, undeniably real, that priced in growth assumptions that took ten years to materialize. Cisco did not go bankrupt. It traded flat for a decade while its earnings caught up.

The risk for Amazon is not bankruptcy or even a 50% drawdown. It is the possibility that the stock trades sideways at $250–300 through 2028 or 2029 while the business grows into its current valuation. That outcome would represent zero return for three years, which is an opportunity cost rather than a drawdown but still a real negative outcome for capital deployed today.

The defense against this is that Amazon, unlike Cisco in 2000, generates the capex from operating cash flow rather than from external financing. The four hyperscalers combined produced over $400 billion in operating cash flow in 2025. Even at approximately $685 billion in 2026 capex, they are not at solvency risk. They will, however, see free cash flow continue to compress, which will pressure their multiples until revenue catches up.

Risk 2: AWS continues to lose relative share

Azure grew 40% in Q1 2026 and Google Cloud grew 63%. AWS at 28% is still losing share in percentage terms even as it accelerates in absolute terms. Over five years, persistent share loss compounds. If AWS share of global cloud infrastructure drops from 30% today to 22% by 2030, AWS revenue in 2030 looks more like $260 billion than $340 billion, which reduces the SOTP value by $1+ trillion.

The counter is that the global cloud infrastructure market is expected to reach $1.5+ trillion by 2030, up from roughly $500 billion today. The base case projects AWS revenue of $340 billion in 2030, which represents 22–23% share of the larger 2030 market versus 30% today. AWS loses relative share in this scenario but still grows revenue meaningfully in absolute terms because the total market triples. The bear case for AWS only triggers if share loss outruns market growth, which would require a structural failure of AWS positioning rather than just relative growth gaps.

Risk 3: Antitrust action that materially changes the business structure

The FTC’s case against Amazon (filed September 2023) remains active. A Democratic administration taking power in 2029 could re-prioritize enforcement. Forced separation of the 3P marketplace from 1P retail, or a forced AWS spinoff, would destroy meaningful value through diseconomies of separation and lost data flywheel effects. I assign 10–15% probability to materially adverse antitrust outcomes over the next four to five years. That is not zero and deserves the discount it implies, but it does not gate the thesis.

Conclusion

The consensus pricing on Amazon assumes a story that the last two earnings cycles have falsified. AWS has re-accelerated. Advertising is a $70 billion business that the sell-side still treats as a margin assumption. Logistics monetization launched a week ago. None of this is forecast. All of it is in the most recent 10-Q.

The base case math gets the stock to $500 by year-end 2030 on continuation of current momentum, which is 13.9% annualized returns. The bull case lands at $783, in line with my $750 price target. The $1,000 case is reachable only if operations execute against the upper end and megacap multiples hold or expand from current levels.

The bear case worth taking seriously is not company-specific. It is multiple compression from hyperscaler capex producing returns that disappoint investors, a Cisco 2000 outcome where the business is healthy and the stock trades sideways for years. That risk is real and prices the bear case at $252 per share.

Price target $750 by year-end 2030. Upside scenario $1,000. Downside scenario $252. The position thesis is that consensus is anchored to an old story, and the new story is already in the data.

---

Suggested Citation: Rettke, Sterling. “Amazon to $750: A Compounder Thesis for 2030.” sterlingrettke.com, May 14, 2026.

Disclosure: The author has held a long position in AMZN for over eight years. This analysis is for informational purposes only and does not constitute investment advice.