Amazon posted second-quarter 2025 revenue of $167.7 billion, surpassing analysts’ expectations. Earnings per share reached $1.68, also beating consensus forecasts. On the surface, this looked like a strong quarter, marking a continuation of Amazon’s post-pandemic momentum.

However, the company’s shares fell nearly 7% in after-hours trading. The driver of this reaction wasn’t Q2’s performance, but Amazon’s profit guidance for Q3, which disappointed markets. The company flagged rising expenses tied to its rapidly expanding artificial intelligence infrastructure, along with operational challenges in scaling its cloud business.

More Articles: 

Amazon’s management made it clear: the business is entering an investment-heavy phase, particularly in AI, and this will temporarily pressure margins even as top-line growth continues.

AWS Grows Revenue but Margins Compress Under AI Load

Amazon Web Services (AWS), the company’s most profitable business unit, reported revenue of $30.9 billion—up 17.5% year-over-year. While this represents a healthy growth rate, AWS operating income fell to $10.2 billion, with margins shrinking to 32.9%. This is the lowest margin AWS has reported since late 2023.

Multiple factors contributed to this margin compression. AI-related capital expenditure is rising faster than revenue. Additionally, increased depreciation of cloud assets, stock-based compensation for AI and engineering teams, and foreign exchange fluctuations weighed on profitability.

Management acknowledged that while demand for generative AI and foundation models is surging, the costs of building and deploying the underlying infrastructure are front-loaded. CFO Brian Olsavsky referred to the margin dip as a “necessary investment phase,” but investors weren’t convinced this spending would yield short-term returns.

Massive AI Investments Signal Strategic Shift

Amazon plans to spend over $100 billion in capital expenditure this year, the largest annual investment in its history. Much of this will go toward expanding AI data centers, scaling proprietary AI chip development, and strengthening cloud infrastructure across North America and Asia.

CEO Andy Jassy emphasized that the company is fully committed to building a vertically integrated AI stack, from silicon (Trainium and Inferentia chips) to software deployment. However, this comes at a cost. Investors are now questioning whether Amazon’s AI strategy can monetize fast enough to justify the upfront spend.

Unlike previous infrastructure buildouts (such as during the early AWS growth years), the AI race involves direct competition from well-capitalized rivals like Microsoft and Google, both of which are also ramping investments aggressively.

Data Center Constraints Limit AWS Scalability

Another factor that dimmed investor sentiment was Amazon’s acknowledgment of infrastructure constraints. The company admitted that it cannot deploy cloud capacity as fast as demand is growing—largely due to physical limitations.

Power availability, chip shortages, and construction delays are all slowing down the rollout of new data centers. Jassy noted that AWS continues to experience “high customer interest,” but the inability to fulfill demand immediately will likely restrain growth in the second half of 2025.

This operational bottleneck adds a layer of complexity: Amazon is pouring capital into expansion, but logistical delays mean the return on that investment may take longer than Wall Street would like.

Retail and Ads Provide Support Amid Cloud Challenges

While AWS faced margin challenges, Amazon’s core retail operations and advertising divisions provided stability. Retail revenues grew 11% year-over-year, buoyed by a successful Prime Day and robust demand across North America and international markets.

Advertising revenue jumped 23%, making it one of the fastest-growing segments in the company’s portfolio. Amazon continues to benefit from its first-party data advantage, enabling high-margin targeted ad offerings.

Another key development is Amazon’s investment in logistics automation. The company is deploying robotic systems and AI-driven warehouse tools, with the goal of reducing fulfillment costs over the long term. Management projected up to $10 billion in annual savings by 2030 through increased automation and operational efficiency.

Strategic Insights and Non-Googleable Takeaways

1. AI Investments Are Diluting AWS Margins in the Short Term
Unlike previous cycles of tech investment, AI infrastructure costs are rising faster than revenue growth. Amazon is in the middle of a cash-heavy buildout phase, and short-term profitability is taking a hit.

2. Physical Constraints Will Cap AWS Revenue Growth
Even with aggressive capital expenditure, AWS is constrained by factors outside its control: electricity availability, server chip supply, and regulatory hurdles in multiple countries. This means cloud revenue may not scale in proportion to investment.

3. Proprietary Chips Are a Long-Term Cost Hedge
Amazon’s internal chips, such as Trainium and Inferentia, are designed to reduce reliance on third-party GPUs like NVIDIA. While not yet a commercial differentiator, they could provide long-term cost advantages once adoption matures.

4. Automation Spending Is a Defensive Move
By investing heavily in warehouse robotics, Amazon is building operational leverage. This is a proactive hedge against cost inflation and thinning retail margins, and it positions the company for long-term savings.

5. Investor Skepticism Reflects Uncertainty, Not Performance
Amazon’s fundamentals remain strong, but guidance ambiguity and vague AI monetization pathways are creating a confidence gap. Investors want clearer time horizons and ROI metrics.

Disclaimer:
This article is for informational purposes only and does not constitute investment advice or a recommendation to buy or sell any securities. Procapitas does not provide personalized financial advice. All investment decisions should be made in consultation with a licensed financial advisor. The information presented is based on publicly available sources and Procapitas’ independent research and analysis, which are believed to be reliable but are not guaranteed for accuracy or completeness.