AMZN is a difficult stock to trade. At $261, it is only 35% above its Covid-era high.

In the past 6 months, the stock rallied 40%, gave back half of its gains, and then moved to another ATH - all of this happening after the US-Iran war.

Currently, AMZN has a unanimous Long vote across the four models.

The stock's choppy behaviour has put the ML model at a significant disadvantage, as it stayed on the sidelines for too long during the initial climb and then failed to get out of the way quickly enough on the way down.

Meanwhile, the Neural model has been more responsive to the dips and is getting in and out of position more quickly. But it caused a higher churn and still underperformed both the Trend and Regime models.

The Trend model entered long fairly quickly after the initial rally and has stayed long since, not being shaken out by the whipsaw. Here, I’m not sure if one’s technical analysis would have fared any better, as TA would have only entered once the price cleared the moving average clusters anyway, and can get headfaked by at least two breaches of the 200-day MA at this point.

Finally, the Regime model switched to flat only for about a month, from late June to mid-July, capturing the largest share of the upside.

AMZN’s position decision is decided by a majority vote among ML, Trend, and Regime. The least likely off-switch is from Trend, which means we likely need both ML and Regime models to switch to flat to shake the position out.

Amazon runs three businesses:

AWS is the profit engine, with $42.2 billion in revenue in the June quarter at a 39% operating margin, an annualised run rate of $169 billion, producing $16.6 billion of segment operating income. AWS commands roughly 31% of cloud infrastructure with a $496 billion contracted backlog growing at triple-digit rates YoY. Enterprise migration costs make that revenue sticky.

Retail turns roughly $560 billion of annualised revenue at a thin margin, with the marketplace fee piece as the profitable half. Fulfilment is a physical asset that is nearly impossible to replicate in the same density.

Advertising is the highest-incremental-margin business of the three, compounding in the low-20s YoY, thanks to the high purchase-intent signal quality Amazon enjoys.

Amazon sits at the demand end of the hardware chain and the supply end of compute:

Feeding AWS' growth costs roughly $220 billion in capital spending this year, making Amazon the largest single buyer among the hyperscalers, ahead of Microsoft and Alphabet. Capital spending runs at 21.79% of revenue on a trailing basis, 26.60% on 2026 consensus, and 28.70% on 2027 consensus. Free cash flow has gone from a $38.2 billion peak in 2024 to negative $7.6 billion trailing, with consensus at negative $52.7 billion over the next twelve months.

Amazon is scaling through its Digital Transformation playbook. Capex rose 52.1% YoY to $61.1 billion in 2021 and another 4.2% to $63.6 billion in 2022. Free cash flow hit negative $9 billion, then negative $11.5 billion. The shares de-rated to roughly 11–12x forward EBITDA at the 2022 trough, and the recovery took two years.

However, the current build is four times that size, and the assets are shorter-lived: data-centre shells monetise over 30+ years, but servers and networking gear take close to three years to break even against a five-to-six-year useful life, and roughly two-thirds of AI capex goes into that short-lived tier.

At the same time, Microsoft, Alphabet, and Meta all raised their 2026 budgets. Memory suppliers raised prices hard enough to add ~$20 billion to Amazon's own guidance. The AI labs are still funding purchases with equity raises rather than cash flow — Anthropic was valued at $965 billion after a $65 billion round in late May 2026. This may be where caution is warranted, as the backlog's largest single contract is funded by a private capital market rather than a profitably operating business.

AWS revenue reached $42.2 billion, up 36.7% YoY. This was the fastest growth in 18 quarters when AWS was less than half its current size. AWS backlog reached $496 billion, growing triple digits YoY, with an annualised run rate of $169 billion. AWS operating income of $16.6 billion, with margin up 650 basis points YoY. I would argue that AMZN’s entire valuation is attributable to AWS.

Custom silicon run rate is now over $25 billion annually, growing triple digits. The AI revenue run rate is separately over $25 billion, also triple-digit growth. Anthropic and OpenAI have both made multi-year, multi-gigawatt commitments to Trainium (custom chips). Other notable customers include Uber, Pinterest, NEURA Robotics, Odyssey, Poolside and TwelveLabs.

Advertising revenue of $19.8 billion, up 26% YoY. Conversion has improved significantly, as shoppers who click a sponsored prompt convert 48% more often and spend 21% more.

The EBITDA margin spike includes roughly $600 million of tariff refunds in North America and roughly $600 million from fair value changes on energy derivative contracts, primarily in AWS.

What could concern investors is whether the free cash flow margin continues to widen on the downside.

FY2026 cash CapEx was raised to approximately $220 billion from approximately $200 billion, with the entire $20 billion increase attributed to higher memory costs. Q2 cash capex alone was $53.1 billion.

Here are the key financial metrics from the past four quarters. I designed the table this way so we can focus solely on the key top-line, profit, and cash flow metrics.

Here are Wall Street's forward estimates. Revenue growth is expected to slow to low-mid teens in the next 12 months, although AMZN is very likely to beat these expectations.

What Keeps The Stock Sideways

AWS decelerates from 36.7% YoY toward the high-20s through 2027 as comparisons harden, advertising compounds in the low-20s YoY, and 2027 EBITDA lands near the consensus $271.7 billion.

The multiple compresses to 10.5x forward EBITDA as depreciation from the build reaches the income statement, and free cash flow stays negative. That gives roughly $260 a share, flat against $261.31 today. Two years of building, no re-rating, and the shareholder is paid in optionality rather than cash.

The Bullish Case

AWS holds above 33% YoY growth at a margin near 39% through 2027, while the 2027 capital budget is NOT raised again. This would be the combination that converts the $496 billion backlog on the 2021 to 2023 schedule rather than a longer one. EBITDA beats to about $290 billion and the multiple holds at 13x. That gives about $330, +26%. Alternatively, the valuation multiple expands to 15x while revenue and profits stay on target.

Bear case

The opposite pairing. A 2027 capital budget guided above $270 billion while AWS growth slips under 30% YoY, which is spending accelerating into decelerating demand. EBITDA misses to about $245 billion and the multiple de-rates to 9.5x, below the 11x to 12x trough of the 2022 overbuild.

AMZN’s EV/EBITDA has moved to a multi-year low, as its NTM EBITDA margin is expected to decrease meaningfully. (EBITDA NTM growth turns negative).

In terms of the target valuation multiple, EV/ Sales is too rudimentary for Amazon’s conglomerate of businesses across retail, AWS, and marketing.

EV/ EBITDA is a slightly better metric, as it accounts for the profit margins too. At an expected 5% decrease NTM but an overall 27% increase over the next 3 years, 15x is a reasonable valuation here. It implies a stock price of $332 in one year’s time, a 27% increase. To be clear, that price increase is entirely driven by the valuation multiple expansion in the near term (from 11.6x).

The current price would be agreed by the DCF model, if assuming AMZN grows revenue around mid-teens through to 2031, raises the EBIT (GAAP) margin to 22%, and gradually reduces CapEx as a percentage of revenue from near-30% to mid-teens.

These assumptions are not difficult to achieve, provided continued excellent execution. If AMZN can raise its revenue growth and EBITDA margin while reducing its CapEx simultaneously, the stock can re-rate to even higher.

Amazon’s valuation is entirely attributed to its high-growth, high-margin AWS business, which, as a standalone, would be a Fortune 50 business by revenue. Therefore, the most important thing Amazon must get right is to repeat the Digital Transformation playbook as the business increased both its CapEx and actual revenue.

You can think of today’s price as being at or near “fair value” for AWS, assuming continued execution, but there is further upside if AWS accelerates in both revenue and margins. The retail and marketing businesses are “freebies” to the stock, which is a nice-to-have optionality here.

AMZN has seen rallies since 2024 in three waves: R1, R2, and the current R3.

R2 replicated the 100% extension of R1, while R3 has so far completed about 75% of R1 or R2. R3 currently stalls at the upper channel resistance, but I favour it to make a breakout next.

R3 can be further divided into R3.a and R3.b. If R3.b replicates R3.a, it arrives at the same point of the 100% extension of R1 and R2, which is my overall R3 target.

Amazon is executing very well, but faces all the same concerns as other hyperscalers in its high CapEx vs ROI. However, there is some assurance that the management has done this type of buildout (though in a smaller scale) in its recent history, and AWS is moving very fast, even given its market-leading scale.

Valuation here is not a deterrent to entry. I am more than happy to follow the Systematic Portfolio choice on this one.