Amazon: AWS Acceleration Is Real, But A $220B Capex Gamble And The Stock's Rally Leave Too Little Room For Error

Amazon reported its first-ever $200.6 billion in net sales and AWS grew 37%, its fastest rate in 18 quarters. The stock has surged roughly 36% over the past four months and sits within striking distance of its 52-week high. The operating numbers are strong. The question is whether the multiple, after this rally, has room for the enormous capital commitment management is now demanding the market trust.

What happened

Q2 2026 revenue hit $200.6 billion, up 20% year over year and above the $197 billion consensus. Every segment grew: North America 16%, International 15%, AWS 37%. Operating income rose 43% to $27.5 billion, with the consolidated operating margin expanding to 13.7% from 11.4% a year earlier. Advertising grew 26% to $19.8 billion.

The headline net income figure of $62.6 billion requires a caveat: $53.4 billion of pre-tax, non-operating mark-to-market gains on Amazon's Anthropic stake inflated the bottom line. Strip that out and underlying diluted EPS comes to roughly $1.95, still above the $1.82 consensus. The operating picture, not the one-time gain, is what moved the stock.

AWS is the centerpiece. Revenue of $42.2 billion implies a $169 billion annualized run rate - Amazon's own framing. Operating income in the cloud unit surged 64% to $16.6 billion, pushing AWS operating margin to 39.4% from 32.9%. The customer backlog of future revenue commitments reached $496 billion. CEO Andy Jassy said AWS is now in its fifth consecutive quarter of accelerating growth, and that AI and chips businesses each exceeded $25 billion in annualized revenue.

The capex problem

Here's where the risk-reward gets harder. Amazon raised its full-year 2026 capital expenditure guidance from roughly $200 billion to $220 billion, citing higher memory-chip costs. Trailing-twelve-month capex sat at $173 billion. Operating cash flow for the same period was $161.4 billion, up 33%, but that leaves free cash flow at an outflow of roughly $7.6 billion - flipped from an $18.2 billion inflow a year earlier. For context, free cash flow is what remains after a company pays for the assets needed to grow the business. Amazon hasn't been this deep in the red on FCF since 2023.

Jassy framed the squeeze as a timing issue. Data centers require capital about two years before servers are installed and revenue flows; the servers themselves run on a shorter purchase cycle and take roughly three years to break even against a useful life of five to six years. Most AI capacity is contracted for five years or longer. The argument is that this burn is temporary and the monetization window is long.

That argument can work - but only if AWS growth sustains and margins don't compress as those facilities come online. The company has a proven track record of AI spending paying off at scale, and the $496 billion backlog gives some comfort that demand is there. But $220 billion in capex is a massive commitment, and the market is asking Amazon to prove the payoff before handing it another leg of multiple expansion.

Valuation after the rally

Amazon trades at 22.6 times trailing earnings, but that figure is distorted by the $53.4 billion Anthropic gain. Forward P/E sits at 43.4 times, the market cap is $3.06 trillion, and EV/EBITDA is 18.2 times. The PEG ratio - forward P/E divided by expected earnings growth - looks attractive at 0.26, but PEG ratios built on earnings that include one-time investment gains are misleading. The underlying growth story is real, but the trailing multiple isn't a fair anchor.

Compared to peers, Amazon's trailing P/E of 22.6x sits below Microsoft's 27.1x and Alphabet's 18.7x, while EV/EBITDA of 18.2x is in the same band as Microsoft's 18.5x and above Meta's 13.7x. Revenue growth of roughly 16% YoY leads the group. The case is that Amazon's combination of AWS acceleration, advertising growth, and operating leverage justifies the multiple. The counter case is that a stock up 36% in four months, trading near highs, with negative free cash flow and a $220 billion capex commitment, has priced in a lot of AI optimism already.

The catalyst clock

Q3 earnings, expected late October, will be the next inflection. Wall Street is looking for roughly $1.57 in EPS and $178 billion in revenue. The market will want to see AWS growth holding near 37% or better, and some clarity on whether the capex spend is beginning to show a path back to positive free cash flow. Jassy said demand will still outpace capacity in 2027, which should keep AWS bookings strong, but the question for the stock is whether that narrative continues to support a 43x forward multiple when FCF is negative.

Risks and what would break the call

• AWS growth decelerates below 30%. The thesis depends on AWS maintaining its acceleration curve, not just holding steady. A slowdown would pressure the multiple.
• Capex extends beyond 2026 without FCF recovery. If the cash burn continues into 2027 and 2028, the story shifts from timing gap to structural drag.
• Margin compression in AWS. The CFO noted roughly $600 million of AWS operating income in Q2 came from mark-to-market gains on energy derivative contracts, with no repeat assumed in Q3 guidance. That pressure is modest but worth watching.
• The tariff angle: Amazon confirmed it received about $600 million in tariff refunds following the Supreme Court ruling on IEEPA duties, with plans to pass some back to customers. That's a nice tailwind for retail margins, but it's a one-time event, not a structural driver.

Investor takeaway

I am keeping Amazon at a Hold. The AWS reacceleration is genuine, operating leverage is real, and the AI infrastructure buildout has evidence behind it - not just a narrative. But the stock has run hard. A 36% move in four months, a 43x forward multiple, and a $220 billion capex commitment that's turned free cash flow negative create a setup where the market has already done a lot of the work. I need to see at least one more quarter of AWS at this growth rate with a credible sign that capex intensity is coming down before the risk-reward tilts back to Buy. Until then, holders can ride the momentum, but new money at these levels is paying for a great story that the balance sheet hasn't yet earned.