Intel and Amazon – Game of numbers.

Good numbers but both have problems.

• Both Intel and Amazon reported good results, but both have some fundamental problems that are hurting current performance and will take some time to be fixed.

Intel Q3 23 – Risk imbalance

• Intel reported good results as the PC market came back to life to some degree, but the risks that Intel is increasingly facing mean that I would need a very low multiple to get involved.
• Q3 2023 revenues / Adj-EPS were $14.2bn / $0.42 compared to forecasts of $13.6bn / $0.33 as the sales mix tilted towards commercial and gaming PCs resulting in a higher average selling price and better profitability.
• However, compared to its glory days of 60%+, gross margins are woeful at just 42.5% but Intel is confident that if the recovery plan continues to go well, it can get back to these levels.
• Here, Intel is executing better than I thought it would on its plan to migrate through 5 nodes in 4 years, and it is steadily catching the competition.
• The problem is that the competition is not standing still and if Qualcomm’s Snapdragon Elite X (of which more on Monday) is as good as Qualcomm says it is, then Intel has a big problem as Apple-grade silicon is coming to Windows.
• The caveat as always is software and the last time Microsoft tried to migrate the Windows ecosystem to Arm it did a poor job and completely failed to get any traction.
• However, Apple has proven that it can be executed perfectly but this is much easier for Apple as it has complete control of the software and the hardware.
• Qualcomm and Microsoft have been working closely together, and I suspect that Intel’s fate in PCs lies in its ability to match performance by going to smaller nodes and whether Qualcomm and Microsoft make a mess of the migration to Arm.
• Intel is expected to earn $0.64 in 2023 and $1.74 in 2024 putting the shares on 55x 2023 and 20x 2024 PER.
• Things start to look interesting in 2025 and 2026 where it is trading on 14x and 5x PER respectively, but this is a long time to wait, and a lot could happen between now and then.
• By contrast, Qualcomm (the challenger) is on 12.6x 2023 and 11.4x 2024 which looks much more attractive on a risk-adjusted basis.

Amazon Q2 23 – The AI effect

• Amazon reported excellent results, but AWS was a real disappointment, and it is finally becoming clear to the market that Amazon is a laggard in AI and desperately needs to catch up.
• Q3 23 revenues / EBIT were $143.1bn / $11.2bn nicely ahead of forecasts of $141.6bn / $7.7bn driven almost entirely by Amazon’s focus on efficiency.
• The fact that Amazon was able to cut costs without hurting revenues is a sign of bloat and I suspect that there is more that Amazon and many of its peers could do to improve further.
• North America reported EBIT margins of 4.9% which are the best I have seen for some time while International broke even compared to EBIT losses of -8.8% in Q2 2023.
• However, AWS lagged by posting growth of just 12% YoY where it is clearly missing out on the AI dividend that Azure is currently enjoying.
• This is because despite having one of the best opportunities in consumer AI (see below), RFM Research has viewed Amazon as a laggard in AI for years (see here) which is now playing itself out in AWS’s underperformance.
• This will take time to fix as not only does it have to catch up, but it also has to deal with the fact that OpenAI has become a household name which Microsoft can exploit to bring customers onto Azure rather than AWS.
• This is why Amazon had a lot to say about AI on its conference call but so far it has delivered very little.
• Its best shot at this is through Alexa which with 1bn hardware devices installed across the world has the opportunity to bring Amazon’s AI directly to the consumer without having to do anything outside of its own cloud.
• The problem at the moment is that Amazon’s AI is not much good for anything other than playing music, setting timers and turning the lights on and off.
• Furthermore, its generative AI demos at its annual Alexa event were stilted and slow which is a real problem when one is trying to create a user interface based on conversation.
• To do this effectively, it is increasingly looking like one will need to deploy the model on the device itself, but this is still a little way off.
• In the meantime, if Amazon can really improve the quality of the Alexa agent in the cloud, it might see a big increase in usage through the existing devices who would then be more inclined to upgrade to another Alexa device with some of the interface deployed on the device.
• This is why I think Amazon is investing $1.25bn in Anthropic as a future version of Alexa is certain to be based on Claude which is a viable competitor to ChatGPT.
• This is going to take time to fix, but after years of doing very little, Amazon is finally on the case.
• Time will tell whether or not it is too late.