Amazon’s second quarter looked like one of the best in its history: net income of $62.6bn, earnings per share of $5.75 against $1.68 a year earlier, a stock that jumped 15.3% in a single session. Read the fourth bullet of the company’s own release, though, and the picture changes: “Second quarter 2026 net income includes non-operating pre-tax other income of $53.4 billion, primarily from our investments in Anthropic.” The operating business earned $27.5bn. Roughly six-sevenths of the pre-tax gain that tripled headline EPS came from marking up a stake in a private AI company. That is not an accounting scandal — Amazon disclosed it plainly — but it is the single most important thing to understand before putting a price target on the stock. Ours is a $330 bull case against a $185 bear case, from a last close of $258.63 on 21 August 2026.
Here is the number that gets less attention than it deserves, and it comes from the same release. Amazon’s free cash flow is now an outflow of $7.6bn on a trailing-twelve-month basis. Twelve months earlier it was an inflow of $18.2bn. That is a $25.8bn swing in a single year, and Amazon names the cause without hedging: a $66.1bn year-on-year increase in purchases of property and equipment, which “primarily reflects investments in artificial intelligence.” So the two headline facts of this quarter point in opposite directions. The AI investment cycle is producing a spectacular paper gain on the Anthropic stake at exactly the moment it turns the world’s most reliable cash machine free-cash-flow negative. The bull case and the bear case are both built from that single sentence, which is why the spread between them is wide.
Key facts
- Last close $258.63 — up 16.5% over twelve months and 32.0% off the February low, but 9.9% below the $287.20 intraday high of 3 August — stockanalysis.com, 21 August 2026
- Q2 net sales $200.6bn, up 20%; operating income $27.5bn, up 43% — Amazon Q2 2026 earnings release (8-K, Exhibit 99.1), 30 July 2026
- AWS sales $42.2bn, up 37% — the fastest growth in 18 quarters — on a $169bn annualised run rate, with segment operating income of $16.6bn against $10.2bn a year earlier — same release
- Net income $62.6bn included $53.4bn of non-operating pre-tax other income, “primarily from our investments in Anthropic” — same release
- Free cash flow: an outflow of $7.6bn for the trailing twelve months, versus an $18.2bn inflow a year earlier — same release
- Q2 capital expenditure $54.2bn in the quarter alone, equal to 27.0% of quarterly revenue — Q2 2026 Form 10-Q, filed 31 July 2026
- Q3 guidance implies deceleration: net sales $197.0–202.0bn (growth of 9–12%) and operating income $22.5–26.5bn, a midpoint 10.9% below the $27.5bn just delivered — same release
What actually happened in the quarter
Strip out the investment gain and the operating result was still excellent. Net sales rose 20% to $200.6bn. Operating income rose 43% to $27.5bn, lifting operating margin to 13.7%. All three segments grew: North America up 16% to $116.2bn with $9.1bn of operating income, International up 15% to $42.2bn with $1.7bn, and AWS up 37% to $42.2bn with $16.6bn.
That AWS line is the reason the stock rose 15.3% on 31 July. AWS is now the same size as the entire International segment but produces nearly ten times its operating income, at a segment margin of 39.3%. Its growth rate has accelerated to the fastest in four and a half years, and it did so at a scale where deceleration is the normal expectation. Amazon crossed a $3 trillion market capitalisation days later, on 3 August, and set its 52-week high of $287.20 the same day.
The cost of that growth is on the cash flow statement. Amazon spent $54.2bn on property and equipment in the three months to 30 June, against operating cash flow of $45.4bn in the same period — so the quarter consumed roughly $8.8bn more cash than it generated. Annualise the quarterly figure and you get about $217bn, consistent with the $220bn capital expenditure plan the company raised guidance to in July. Trailing-twelve-month operating cash flow of $161.4bn is up 33%; it is simply no longer enough.
Andy Jassy, President and CEO, framed the quarter this way in the release: “AWS is booming, growing 36.7% year-over-year in Q2—our fastest growth in 18 quarters—and our AI and Chips businesses each eclipsed run rates of more than $25 billion.”
The Anthropic gain, and what EPS looks like without it
Amazon’s disclosure is explicit, so the analysis is straightforward rather than speculative. Of $62.6bn in net income, $53.4bn was non-operating pre-tax other income, primarily the revaluation of its Anthropic stake. Apply a 21% statutory rate to that gain as a rough approximation and roughly $42bn of after-tax benefit falls away, leaving something in the region of $20bn of underlying net income, or about $1.88 per share against the $5.75 reported. That calculation is our estimate, not a company figure — Amazon does not publish an ex-gain EPS — and the real effective rate on the item will differ. But the order of magnitude is what matters: the headline number is roughly three times the underlying one.
Two smaller one-offs sit inside operating income itself, both disclosed in the 10-Q rather than the press release. Amazon received roughly $640m of tariff refunds under the International Emergency Economic Powers Act, credited to cost of sales and hitting mainly North America — and it notes this “represents the significant majority of refunds we expect to receive.” A further $551m was an unrealised gain on energy derivatives, primarily affecting AWS. Together they are about 4.3% of consolidated operating income. Strip both and operating income grew roughly 37% rather than 43%, and North America’s margin was about 7.3% rather than 7.85%. Neither repeats.
Two consequences follow. First, any valuation multiple built on trailing GAAP EPS is currently meaningless for Amazon. At $258.63 the stock looks optically cheap on reported earnings and is not. Second, the gain is a mark, not a realisation. Amazon has not sold Anthropic shares; it has revalued them. If the private AI funding market re-rates downward, the same line reverses, and it reverses into the income statement with the same visibility it arrived with. Investors who cheered a 245% jump in net income should expect symmetry.
This is a pattern worth recognising across the AI complex rather than a quirk of one company. The same dynamic — enormous capital expenditure financed against enormous paper gains on AI-adjacent holdings — is visible at Meta and, from the other side of the trade, at Nvidia. What distinguishes Amazon is that it is simultaneously the buyer of the compute, the seller of the compute through AWS, and the holder of an equity stake in one of the largest buyers. Those three roles do not net out; they compound.
Valuation and the guidance nobody priced
At $258.63 with roughly 10.89 billion diluted shares, Amazon carries a market capitalisation near $2.82trn. Trailing twelve-month revenue is running above $760bn, so the shares trade around 3.7 times sales — historically undemanding for Amazon, and the number the bulls point at.
The guidance is the part the market has been slower to absorb. Amazon told investors to expect third-quarter net sales of $197.0–202.0bn, which is growth of 9% to 12% against the 20% just reported. Operating income is guided to $22.5–26.5bn, a midpoint of $24.5bn — 10.9% below the $27.5bn the company had just delivered. In fairness to Amazon, a large part of that revenue deceleration is a calendar artefact rather than demand: Prime Day fell in July in 2025 but on 23–26 June in 2026, so it sits in the prior quarter this year. The company says so in the release — “Excluding the impact of Prime Day in both 2025 and 2026, third quarter 2026 year-over-year growth would be nearly 400 basis points higher” — which would put underlying growth nearer 13–16%. The operating income guidance is harder to explain away. A business does not usually guide sequential operating income down while its highest-margin segment is accelerating, unless the cost base is climbing faster than the revenue. The stock has fallen 9.9% from its 3 August high, which suggests the market has begun to price this, but not that it has finished.
| Measure | Bull reading | Bear reading |
|---|---|---|
| AWS +37% | Fastest in 18 quarters, 39.3% segment margin | Bought with $54.2bn of quarterly capex |
| Net income $62.6bn | Anthropic stake is a real asset | $53.4bn of it is a mark that can reverse |
| Free cash flow | Investment phase, not impairment | −$7.6bn TTM, a $25.8bn swing in a year |
| Q3 guidance | Conservative, as Amazon usually is | Growth 9–12% vs 20%; op income midpoint −10.9% |
| Valuation | ~3.7x sales, undemanding for Amazon | Trailing GAAP P/E is meaningless right now |
| Retail | Record Prime delivery speeds | International margin still only 4.0% |
Regulatory and structural risk
Three structural issues sit underneath the equity story.
The first is the circularity of the AI trade. Amazon sells compute through AWS, buys compute and custom silicon at enormous scale, and holds an equity stake in Anthropic — which is itself a major consumer of AI compute. Revenue, capital expenditure and investment gains are therefore not independent variables. A slowdown in AI demand would compress AWS growth, strand capital expenditure, and mark down the Anthropic stake simultaneously. Investors accustomed to Amazon’s diversification should note that these three exposures are correlated, not offsetting.
The second is that the capital expenditure is largely irreversible. Data centres and custom silicon are long-lived, purpose-built assets. Unlike inventory or headcount, they cannot be flexed down quickly if demand disappoints. Amazon has committed to roughly $220bn of it for the year, and the company’s own framing — investment “primarily reflect[ing] investments in artificial intelligence” — makes the concentration explicit.
The third is that the International segment still earns a 4.0% operating margin on $42.2bn of sales. It is the same size as AWS by revenue and produces roughly a tenth of the operating income. Any consumer weakness lands there first, and it has the least cushion.
Regulatory exposure across the FTC and the EU’s Digital Markets Act remains live but has not produced a decision that changes the earnings model this quarter. It belongs in the risk list rather than the forecast.
The $330 bull case and the $185 bear case
The bull case to $330 is roughly 28% above the current price and about 15% above the all-time high. It requires AWS to hold something close to its current trajectory while capital expenditure plateaus rather than compounds. If AWS exits 2026 on a run rate approaching $200bn at a high-30s margin, and the retail business converts record delivery speeds into the operating leverage it has historically produced, then the group’s operating income can grow into the capital base and free cash flow turns positive again during 2027. On that path the market stops discounting the cash burn and starts capitalising the AWS earnings stream, and 3.7 times sales is the wrong multiple. Note what the bull case does not require: it does not need another Anthropic mark-up, and it should not be underwritten on one.
The bear case to $185 is roughly 29% below the current price and below the $196.00 low set in February. It does not require AWS to stall. It requires only two things that are already visible in the company’s own numbers. First, that capital expenditure keeps running ahead of operating cash flow, so free cash flow stays negative through 2027 and Amazon is valued as a capital-intensive infrastructure business rather than an asset-light compounder. Second, that the Anthropic mark reverses, taking the reported earnings base back toward the roughly $20bn annualised underlying figure and exposing how much of the current optical cheapness was borrowed from a revaluation. The Q3 guidance — 9–12% growth and a 10.9% sequential decline in operating income at the midpoint — is the first datapoint on that path, not the last.
Between those two sits the honest reading: Amazon is a very good business in the most capital-hungry phase of its life, and the market is currently unable to see its earnings power through a $53.4bn accounting gain. The stock is 9.9% off its high with the operating business compounding at 43%. Both of those statements are true, and the next four quarters of free cash flow will decide which one matters. For a comparison with a mega-cap that has already been through its own capex-versus-cash-flow reckoning, see our Apple bull and bear analysis.
Frequently asked questions
What is the Amazon (AMZN) share price today?
Amazon closed at $258.63 on 21 August 2026. The stock is up 16.5% over twelve months and 32.0% above its February low of $196.00, but sits 9.9% below the 52-week intraday high of $287.20 reached on 3 August, when Amazon crossed a $3 trillion market capitalisation.
What is a realistic AMZN price prediction for the next 12 months?
Our range is $330 on the bull case and $185 on the bear case. The bull case assumes AWS sustains high-30s growth while capital expenditure plateaus and free cash flow recovers during 2027. The bear case assumes capex keeps outrunning operating cash flow and the Anthropic investment gain reverses.
Why did Amazon’s net income more than triple?
Because $53.4bn of the $62.6bn was non-operating pre-tax other income, which Amazon states came primarily from its investments in Anthropic. Operating income was $27.5bn. The gain is a revaluation of a holding, not a realised sale, so it can reverse if private AI valuations fall.
Is Amazon free cash flow negative?
Yes. On Amazon’s own definition, free cash flow was an outflow of $7.6bn for the trailing twelve months to 30 June 2026, against an inflow of $18.2bn a year earlier. The company attributes the swing to a $66.1bn year-on-year increase in property and equipment purchases, primarily for artificial intelligence.
How fast is AWS growing?
AWS grew 37% year on year in the second quarter of 2026 to $42.2bn, which Amazon describes as its fastest growth in 18 quarters, on an annualised run rate of $169bn. Segment operating income was $16.6bn against $10.2bn a year earlier, a margin of 39.3%.
What is Amazon’s guidance for the third quarter?
Net sales of $197.0–202.0bn, representing growth of 9–12% against the third quarter of 2025, and operating income of $22.5–26.5bn. The operating income midpoint of $24.5bn is 10.9% below the $27.5bn Amazon reported for the second quarter.
This article is editorial analysis and is not investment advice. Figures are drawn from Amazon’s Q2 2026 earnings release and Form 10-Q filed with the SEC; market data from stockanalysis.com as of 21 August 2026. The ex-gain earnings estimate is a FinanceFeeds calculation using a 21% statutory rate and is not a company figure. Price targets are FinanceFeeds estimates and may not be achieved. Featured image: the Amazon Spheres, Seattle, by Sea Cow, licensed under CC BY-SA 4.0, via Wikimedia Commons.
