Tech Catch Up – Amazon’s Agony

The have-not is hurting

• Amazon is rapidly losing ground in the cloud to its upstart rivals, and despite the claims of Andy Jassy, I think its weak position in AI is largely to blame.  
• Amazon reported good headline Q2 2025 results, but the performance of AWS was way below what I would have expected given how well its competition is performing.  
• Q2 2025 revenues / EPS were $167.7bn / $1.68 ahead of estimates of $162.4bn / $1.33 but the performance of AWS and its profit forecast for Q3 2025 underwhelmed.
• By far the most serious problem, in my opinion, is AWS, which posted 17% YoY revenue growth to $30.9bn.
• Both Microsoft and Google saw a reacceleration of their revenue growth to 39% YoY and 32% YoY, respectively.
• Amazon is still by far the biggest of the cloud providers, and as such, recording high growth is more difficult, but this only accounts for so much in my opinion.
• In H1 and Q2 2024, AWS recorded similar levels of revenue growth as so far in 2025, but both Microsoft and Google were significantly lower than their current levels.
• Hence, Microsoft and Google have seen a pick-up that has passed Amazon by.
• Andy Jassy was quick to explain this as a capacity constraint, but I don’t think that this explanation holds water.
• This is because the demand for cloud capacity has been evident for some time now, and all players (including Amazon) have reacted by upping capex.
• In Q4 2024 and Q1 2025, both Google and Microsoft complained of capacity constraints, and their increases in capex have now allowed revenue acceleration.
• However, Amazon has yet to see the benefits of its increased investments, and I think the reason is more problematic than capacity constraints.
• RFM has long classified Amazon as a laggard in AI, and very little has changed despite a lot of talk and promises about how Amazon is investing in generative AI and LLMs.
• I still get endless advertisements for products I have already purchased, Alexa remains as dumb as ever, and merchants continue to struggle with AI-related errors when it comes to the merchandise they sell on Amazon.
• Simply throwing billions of dollars in the direction of Anthropic is not going to fix the problem, and the market knows it.
• In theory, it should not make any difference if one’s in-house AI is weak when it comes to hosting the AI of others, but the revenue growth tells a different story.
• It is almost as if the market looks at Microsoft with its position with OpenAI and Google with its in-house AI expertise and decides that these two will offer a better hosting of AI than Amazon.
• Consequently, I think that Amazon’s ongoing weakness in AI is hurting its cloud business, and unless something changes, it is not inconceivable that Microsoft takes over market leadership.
• Microsoft clearly smells blood in the water as it disclosed at its FY 2025 results in July that FY 2025 revenues of Azure had surpassed $75bn, meaning that its ARR is probably closing on $100bn.
• Amazon’s ARR is currently $123bn, meaning that for the first time, Microsoft is within striking distance.
• It was not long ago that Amazon was 2 or 3 times the size of its nearest rival, which is an indication of just how much ground Amazon has lost.
• This is a problem of execution as Meta Platforms was in a similar position several years ago and has managed to bring itself back to within striking distance of the leaders.
• Consequently, Amazon needs to do more than have an option to acquire Anthropic when it runs out of money to be seen as a leader in AI.
• It needs to fix Alexa, which now looks prehistoric compared to Gemini, ChatGPT and so on and generally walk the AI walk rather than talk the AI talk.
• Failure will mean that it continues to cede market share and potentially even market leadership.
• Amazon is currently trading on 34.3x 2025 PER, putting it broadly in line with Microsoft but much more expensive than Google.
• Given that Google is doing reasonably well in the AI space and is still posting good revenue growth, I would own Google over Amazon any day of the week.