Amazon's cloud business surged as AWS revenue growth accelerated for a fifth straight quarter, rising 37% to $42 billion and boosting the company's stock nearly 10%. The results, driven by massive AI demand, affirm Amazon's strategy of heavy capital investment and positioning its Bedrock platform as a host for multiple leading AI models, rather than relying on developing its own top-tier model.
AWS revenue growth
- ▪The Q2 results marked the fifth consecutive quarter of accelerating cloud-computing revenue growth for Amazon
- ▪Amazon Web Services (AWS) revenue grew 37% year-over-year to $42 billion in the second quarter
- ▪Amazon's overall net sales rose 20% in the second quarter, and its stock increased by nearly 10% in after-hours trading following the earnings report
- ▪For the fiscal year ending June 30, Amazon spent $173 billion on property and equipment, an increase from $107.65 billion the previous year
- ▪Amazon raised its 2026 capital expenditure (capex) forecast from $200 billion to $220 billion to meet AI demand
- ▪Competitor Microsoft's Azure cloud revenue grew 43% in its recent quarter, while Google Cloud revenue increased by 82%
- ▪Amazon is developing its own custom chips, such as the Trainium TPU and Arm-based Graviton processor, to improve margins for its cloud business
Multi-model AI strategy
- ▪Amazon CEO Andy Jassy stated that there will not be a single AI model that dominates the industry
- ▪Amazon's strategy is to provide customers with access to a wide range of leading AI models from various providers, including Anthropic and OpenAI
- ▪Andy Jassy explained that customers want access to multiple models because different systems excel at different tasks and will leapfrog each other in capabilities over time
- ▪Jassy predicted that within a few years, there will be at least a half-dozen AI models that are comparably good, and Amazon's platform will host them
Amazon Bedrock platform
- ▪The rapid growth of Amazon Bedrock is fueled by the trend of customers wanting to use multiple leading AI models for their applications
- ▪According to CEO Andy Jassy, Amazon and its Bedrock service can have a "wildly successful business" even without having its own top-tier frontier model
- ▪Amazon is positioning its Amazon Bedrock service as the central platform for customers to access and use a variety of leading foundation models
Internal model development struggles
- ▪Reports indicate Amazon's internal efforts to build its own AI models have been "in somewhat of a mess."
- ▪Amazon's Alexa is reducing its use of costly models from Anthropic, switching to Amazon's own AI models to lower operational costs
- ▪As part of its strategy overhaul, Amazon is winding down most of the Nova model line and reallocating resources to a new frontier-model initiative
- ▪Amazon's in-house "Nova" line of models struggled to compete with other frontier models, leading the company to overhaul its AI strategy
Industry shift toward efficiency
- ▪Investors are currently treating cloud providers like Amazon, Microsoft, and Google as the most reliable investments in the AI stack
- ▪The AI market is shifting from a focus on raw model performance to an emphasis on efficiency, or "intelligence per dollar."
Story comments
Loading comments…