Amazon & Apple – AI opposites

Amazon Q2 26 – Reassuring commentary

• Amazon has confirmed that the AI rollout is still on by stating that it has more demand than it can meet, even when spending $220bn in capex for this year, which it expects to make a good return on.  
• Q2 26 revenues / adj-EPS were $200.6bn / $5.75, well ahead of expectations of $197.0bn / $1.82, but a lot of this was due to a write-up in its investment in Anthropic.
• If I take the unrealised gain out and adjust for tax, then Q2 26 EPS would have been $2.01, a more modest beat of expectations.
• The star of the show was AWS, where growth has reaccelerated to 37% YoY, which is crucial if AWS is to stay ahead of its rivals.
• AWS is now a $169bn run-rate business with excellent 39% operating margins.
• Amazon said that it would be increasing capex to $220bn as a result of higher prices, but in contrast to Google and Microsoft, this was well received.
• I suspect that this was due to a combination of very strong AWS performance and an explanation of the data centre business model on the conference call.
• Here Mr Jassy pretty much reiterated the business model that RFM Research has previously published, but he omitted two key elements.
• First is pricing, which currently is extraordinarily good but as supply of compute comes into balance with demand, this will normalise and bring margins back down.
• Second is the fact that much of the equipment that Amazon says it can buy only a few months in advance is currently in short supply and requires commitments of 12 months or more ahead to be able to buy.
• Hence AWS’s ability to ratchet down spending suddenly if demand evaporates is not as great as it would lead us to believe.
• That being said, Amazon is in an increasingly good position to provide compute to the supply-constrained market.
• Hence, I think that margins could go up again from here and remain robust for at least 12 months or so.
• However, the shares are trading on 24.9x 2026 PER, meaning that either Meta (more speculative) or Google (also exposed to this theme) would make a better position to take.

Apple FQ3 26 – Dent from memory and supply

• The memory and supply shortage has also hit Apple, as its iPhone launch quarter will not be as good as usual as a result of higher prices causing weakness in the overall consumer electronics market.
• FQ3 26 revenues / EPS were $109.4bn / $2.02 ahead of estimates of $109.0bn / $1.89, but Services disappointed and guidance was not as good as hoped.
• Services revenues were $30.7bn, up 12% YoY, behind expectations of $31.4bn, which is also concerning as this has been a major factor in helping Apple keep growing despite its maturing hardware businesses.
• FQ4 26 revenues are expected to grow by 9% to 11%, below expectations of 12% YoY, as Apple is having difficulty securing capacity to make its silicon chips and is also dealing with the memory pricing issue that has hit everybody.
• Macs are a bright spot as the Mac Mini with the M5 chip has proven to be a very cost-effective way of running AI models at the edge of the network, and Apple’s Mac numbers reflect this.
• However, the size of this business is not large enough to materially move the needle, meaning that Macs alone are not going to bring the company back to high growth.
• The iPhone business has been far more resilient than one would have expected given the conditions in which it is operating, and this is due to the slight shortening of the replacement cycle where a new form factor has encouraged users to upgrade more frequently than they otherwise would have done.
• Apple has been forced to increase prices on many of its products, and this too will have an impact on demand, albeit less than anyone else given the strength of the Apple brand.
• With its AI strategy still effectively in development, Apple could be seen as a hedge against AI volatility, which has been extreme over the last month or so.
• However, with the shares still on over 30x 12-month forward PER, this is an expensive way to do it when one can have some of the best software names at a 30% to 50% discount.
• These are trading with almost perfect inverse correlation to the AI names, with some having the benefit of also having 70% to 100% fundamental upside when the market realises that they will not be killed by AI.
• Hence, I still don’t see any real reason to own Apple here and think it will underperform both the AI trade and the good quality software names.