Amazon sales rise 13% in Q2 but profit outlook softens

Amazon has welcomed better than expected trading in its second quarter, as revenues rose by 13% to $167.7bn (£130.8bn), but has warned that Q3 profits may not meet expectations amid ongoing tariff uncertainty.

Operating income hit $19.2bn (£15.0bn) in the quarter ended 30 June 2025, compared with $14.7bn (£11.5bn) in the second quarter of 2024. 

In addition, net income rose to $18.2bn (£14.2bn), or $1.68 (£1.31) per diluted share, compared with $13.5bn (£10.5bn), or $1.26 (£0.98) per diluted share.

It comes as North America sales rose by 11% year-over-year to $100.1bn (£78.1bn) in Q2, while international sales were up by 16% to $36.8bn (£28.7bn).

The company’s AWS segment also saw sales rise by 17.5% year-over-year to $30.9bn (£24.1bn). Over the quarter Amazon signed multiple new AWS agreements, including with PepsiCo, Airbnb, Peloton and London Stock Exchange.

Despite this strong performance, the group warned that Q3 profits may fall below expectations, with operating income now expected to be between $15.5bn (£12.1bn) and $20.5bn (£16.0bn), compared with $17.4bn (£13.6bn) in the third quarter of 2024.

Over the period Amazon introduced DeepFleet, an AI model that will coordinate the movement of more than a million robots across its fulfilment network. According to the group, it will improve the travel time of its robotic fleet by 10%, “enabling us to deliver packages to customers faster and at lower costs”.

The technology has been built by using the company’s data sets of inventory movement within its sites and leveraging AWS tools, including Amazon SageMaker.

It is also built on AI that “learns and improves over time”, and will “continue to find new ways to optimise how our robots work together”.

Andy Jassy, president and CEO of Amazon, said: “Our conviction that AI will change every customer experience is starting to play out as we’ve expanded Alexa+ to millions of customers, continue to see our shopping agent used by many millions of customers, launched AI models like DeepFleet that optimise productivity paths for our 1M+ robots, made it much easier for software developers to write code with Kiro, launched Strands to make it easier to build AI agents, and released Bedrock AgentCore to enable agents to be operated securely and scalably.

“Our AI progress across the board continues to improve our customer experiences, speed of innovation, operational efficiency, and business growth, and I’m excited for what lies ahead.”