Amazon’s $220 Billion Flywheel

Post-close earnings analysis, July 30, 2026. Amazon reported a quarter that looked extravagant from almost every angle: $200.6 billion of revenue, 37% growth at AWS, $27.5 billion of operating income and a new $220 billion capital-spending plan. The shares rose more than 9% after hours.

The most important number was not the stock move. It was the acceleration. Amazon’s consolidated revenue growth increased from 17% in the first quarter to 20% in the second. AWS growth accelerated from 28% to 37%, its fastest pace in 18 quarters. At a moment when investors are separating AI spending that produces revenue from AI spending that produces only expense, Amazon delivered proof.

The company is more diverse than the ticker suggests

North American sales rose 16% to $116.2 billion and produced $9.1 billion of operating income. International sales increased 15% to $42.2 billion and earned $1.7 billion. AWS also generated $42.2 billion of revenue—but produced $16.6 billion of operating income, up from $10.2 billion a year earlier.

That mix is Amazon’s quiet advantage. Retail supplies traffic and purchasing intent. The third-party marketplace turns other merchants’ sales into fees. Prime strengthens retention. Logistics monetizes the network. Advertising sells access to a customer near the moment of purchase. AWS supplies the high-margin infrastructure beneath an expanding share of the digital economy.

The same household can buy a product, watch a video, pay for a subscription and generate an advertising impression. The same enterprise can rent storage, databases, custom silicon and artificial-intelligence capacity. Amazon is not one flywheel. It is several flywheels sharing data centers, distribution nodes, customers and capital.

AWS is still the gold

AWS represented roughly one-fifth of quarterly revenue but about three-fifths of the operating income generated by Amazon’s three reported segments. That disproportion is why 37% growth matters so much. Every incremental cloud dollar arrives in a business with structurally better economics than first-party retail.

Chief Executive Andy Jassy said both Amazon’s AI business and its custom-chip business now exceed $25 billion annual revenue run rates and are growing at triple-digit rates. Trainium commitments from Anthropic and OpenAI span multiple years and multiple gigawatts. Graviton commitments roughly tripled sequentially. The company is also seeing demand outstrip available capacity through 2027, with what Jassy called “striking” demand already visible for 2028.

This is the bullish interpretation of the build: Amazon is not spending to invent demand. It is spending to catch demand it already has.

The $5.75 EPS needs an asterisk

Amazon reported net income of $62.6 billion, or $5.75 per diluted share, versus $18.2 billion and $1.68 a year ago. That headline is not a clean measure of operating earnings. The quarter included $53.4 billion of non-operating pretax income, primarily related to Amazon’s investment in Anthropic.

The more durable figure is operating income, which rose 43% to $27.5 billion and exceeded the high end of Amazon’s own $20 billion to $24 billion guidance. Revenue also exceeded the prior guidance range of $194 billion to $199 billion. In other words, the operating business delivered even after removing the mark-to-market fireworks.

This distinction matters for valuation. A trailing P/E that incorporates a huge unrealized investment gain can make the shares look cheaper than the recurring business actually is. Investors should normalize Anthropic-related gains and focus on segment operating income, cash generation and the returns earned on new infrastructure.

The bill arrived before the cash return

Trailing-12-month operating cash flow climbed 33% to $161.4 billion. Free cash flow, however, moved to negative $7.6 billion from positive $18.2 billion a year earlier. The main reason was a $66.1 billion year-over-year increase in purchases of property and equipment net of incentives, largely for AI infrastructure.

Amazon now expects approximately $220 billion of cash capital spending in 2026, up from the prior $200 billion plan and $128 billion in 2025. Jassy said higher memory costs were the principal reason for the increase. That is favorable evidence for suppliers such as Micron, but it raises Amazon’s execution bar. Capacity must turn into revenue fast enough to outrun depreciation, power costs and financing demands.

The spending is not inherently wasteful. A data center is a productive asset Amazon can rent for years. But “AI capex” is not a magic phrase that suspends arithmetic. Utilization, pricing, customer concentration and useful life determine the return.

The guidance is deliberately messy

For the third quarter, Amazon expects sales of $197 billion to $202 billion, or growth of 9% to 12%, and operating income of $22.5 billion to $26.5 billion. The sales midpoint sits below the consensus cited by FactSet, but the comparison is distorted because Prime Day moved into the second quarter. Amazon said growth would be almost four percentage points higher without the timing shift. Foreign exchange is expected to reduce growth by roughly 0.8 percentage point.

The market’s positive after-hours response suggests investors looked through that calendar noise and concentrated on AWS acceleration, the operating-income beat and capacity demand. Tomorrow’s regular session will show whether that judgment survives a full day of price discovery.

What would prove—or break—the thesis

The favorable case is straightforward: retail margins remain disciplined, advertising keeps compounding, AWS sustains growth above 30% and the new capacity is absorbed quickly. In that outcome, Amazon’s diverse revenue base funds an infrastructure advantage that smaller competitors cannot match.

The failure case is also visible: tariff costs compress retail, memory and power inflation persist, AWS growth slows before depreciation peaks, or customers concentrate too much bargaining power in a few giant contracts. The next proof points are AWS backlog and margins, capital intensity, retail unit economics and the pace at which free cash flow recovers.

Amazon’s quarter did not make $220 billion look small. It made the opportunity look large enough that the number can be debated seriously.

Sources

• Amazon second-quarter 2026 results, cash flow and guidance
• Amazon first-quarter 2026 results for sequential comparisons
• Post-earnings share move, capital-spending update and consensus context

Information verified after Amazon’s July 30, 2026 earnings release and call. After-hours prices can change materially before the next regular session.

Author positions: The author holds long positions in AMZN and MU. Compensation: Neither Aria Vantage nor the author received compensation from any issuer or third party in connection with this article. This article contains analysis and opinion. It is general market commentary for informational and educational purposes, not individualized investment advice or a recommendation. Read the full disclosures.