Amazon's Earnings Beat Wasn't What You Think — And That Changes the Risk/Reward

The headline number is $5.75 per share. Wall Street expected $1.82. That's a beat so large it looks like the market has no reason to doubt Amazon's trajectory anymore. Shares surged roughly 20% over the week following the Q2 report, pushing the stock toward its $285 area and a market cap above $3 trillion.

But the market's reaction is misreading what happened. The real story sits behind the EPS headline, and it changes how I view the setup.

The EPS 'Beat' Wasn't Really a Beat

Here's what the $5.75 figure hides: Amazon booked $53.4 billion in non-operating income during the quarter, primarily from its investments in Anthropic. That one-time gain is the difference between a solid earnings report and the monster beat the headline suggests.

Stripping that out, operating income was $27.5 billion, up 43% year-over-year from $19.2 billion. That's strong execution, no question. But comparing the $5.75 EPS against the $1.82 estimate is misleading — the estimate never included the Anthropic windfall, and the result doesn't reflect a fundamental shift in how the business generates earnings. If you want to understand Amazon's operating momentum, look at operating income, not diluted EPS inflated by an investment mark-up.

AWS Reacceleration Is the Real Story

Where Amazon actually exceeded expectations is cloud. AWS revenue hit $42.2 billion, up 37% year-over-year. That's the fastest growth rate AWS has posted in 18 quarters — and it matters because the market had been watching for a slowdown that never came.

AWS operating income jumped to $16.6 billion from $10.2 billion a year ago, with the segment margin expanding to 39.4% from 32.9%. The AI business eclipsed a $25 billion annual revenue run rate, growing at triple-digit percentages. Amazon's custom chip unit — Trainium and Graviton — also cleared the $25 billion run rate, with multi-gigawatt commitments from Anthropic, OpenAI, Uber, and Pinterest.

This isn't a cyclical blip. It's the kind of acceleration that validates why the market has kept buying Amazon through tariff fears and spending anxiety. AWS annualized revenue is now at $169 billion, and the backlog is at a record $496 billion. Demand isn't the problem.

The $220 Billion Question

The problem — or at least the question — is what Amazon is willing to spend to sustain it. Capital expenditures hit $53 billion in the second quarter alone, up 69% from a year earlier. Management raised full-year 2026 capex guidance to $220 billion. That's more than Amazon spent on physical assets in the entire prior year.

The consequence shows up in free cash flow. Trailing twelve-month free cash flow is now negative $11.6 billion. Amazon generated $161.4 billion in operating cash flow but deployed $173 billion in capital expenditures. For a company that once prided itself on cash generation, the shift is dramatic.

This is where the risk lives. Goldman Sachs estimates the four largest hyperscalers — Amazon, Microsoft, Alphabet, and Meta — will spend a combined $5.3 trillion on AI infrastructure over the coming years. The spending is real and accelerating. The question that separates conviction buyers from cautious observers is whether AWS growth stays high enough to justify the investment. Analysts at Jefferies and others have noted that if AWS growth falls below roughly 28%, the $220 billion program risks becoming the most expensive capital deployment mistake in corporate history.

The market is currently treating the capex as proof of demand rather than a risk. That's a reasonable read as long as AWS keeps growing at 37%. It's a dangerous read if growth decelerates to mid-20s while capex stays at $200 billion.

Valuation at the Top

At $274, Amazon trades at a forward P/E of 41.9x. The trailing P/E of 21.9x looks cheap, but that's because last year's earnings haven't caught up to the $220 billion spending ramp. The forward multiple is the one that matters.

The PEG ratio — forward P/E divided by the revenue growth rate — sits at 0.25, which is compelling on paper. But that calculation uses the 15.8% trailing revenue growth, not the forward trajectory. If growth holds above 20% through 2027, the forward P/E compresses into a zone where Amazon looks attractively priced relative to its growth rate. If growth slows to high-teens, the 41.9x forward multiple starts asking a lot from margins that haven't proven they can carry it.

Compared to peers, Amazon's EV/EBITDA of 17.6x sits below Microsoft at 19.0x and Alphabet at 24.2x, but above Meta at 13.7x. Amazon is in the middle of the pack, which makes sense — it's not the cheapest, but it's not the most expensive either. The valuation gap between Amazon and its higher-multiple peers reflects the free cash flow risk I noted above.

Price Action: Extended, Not Overbought

The stock is up 38% over the last four months and 12% over the last 20 days. The RSI sits at 64, above the 50-day moving average of $247.5 and the 200-day average of $236.5. The MACD is positive at 7.58. The stock is clearly in momentum mode.

RSI at 64 is not overbought territory — that would be above 70. But the stock is trading near its 52-week high of $287.2 and has covered nearly half the distance from its 52-week low of $196 in a short window. Momentum is supportive, but the risk/reward for buying at or near all-time highs is rarely ideal.

The Verdict

Here's where I land. The Q2 earnings report is genuinely strong on the operating side. AWS reacceleration at 37% is the kind of data point that justifies optimism about the AI infrastructure buildout. The custom chip story is maturing faster than most investors expected. Amazon's competitive moat in cloud infrastructure — the breadth of services, the enterprise lock-in, the Trainium pipeline — remains intact.

But the risk/reward for chasing at current levels is poor. The stock has run 38% in four months. The EPS headline was inflated by a one-time investment gain that won't repeat. Free cash flow is negative and the capex bill is still rising. I don't think the market has baked in too much doom here — the opposite is arguably true. It has baked in too much confidence that AWS growth will stay at 37% while capex runs $220 billion a year.

A pullback to the $245-$250 area — near the 50-day moving average — would offer a much more attractive entry. That's where the forward P/E starts looking justified by the growth rate and where the risk/reward tilts in favor of buyers. I'd reassess my conviction if AWS growth drops below 25% in a consecutive quarter or if management raises capex guidance beyond $250 billion without a corresponding increase in revenue visibility.

Don't chase. Wait for the dip. Amazon's thesis is still strong, but timing matters.

Rating: Buy (Wait for Pullback)