Samsung Q2 – Party On!

Samsung confirms the AI party is still on.

• Samsung has reported another set of excellent results confirming that the spending on AI is continuing apace and that it may be a few quarters before reality catches up with the hype.
• Q2 revenues and operating profit will be KRW73tn-KRW75tn / KRW10.3tn – KRW10.5tn in line with the revenue estimate of KRW74tn but way ahead of the profit estimate of KRW8.8bn.
• Somewhat unusually, the fact that profits would be better than expected had not leaked into the market before the numbers which I think points to the fact that this has been mostly driven by pricing which in turn led to a bounce in the shares. .
• DRAM pricing has risen by 13%-18% during Q2 2024 while pricing of NAND flash memory increased by 15%-20%.
• For a commodity product like memory, this indicates very robust demand almost certainly driven by AI which flows directly to the bottom line giving the higher-than-expected profitability.
• This looks like it will continue in Q3 2024 as Samsung is expected to qualify as a supplier of high bandwidth memory (HBM) chips to Nvidia putting it back into contention with SK Hynix and Micron.
• These results underpin what Micron reported a couple of weeks ago where expectations arguably got ahead of themselves causing Micron not to beat and raise as was widely expected.
• However, when one looks at Micron’s AI business, its high-end HBM is sold out until the end of calendar 2025 and it is seeing strong AI demand elsewhere also driven by AI.
• Consequently, this re-affirms that the AI party is still on which I think we will see repeated when Nvidia reports its results in August.
• We will also see it in the capital expenditure budgets of the large cloud providers such as Amazon, Microsoft and Google which look set to continue expanding.
• The key beneficiary of this is Nvidia with a $100bn and growing AI data centre business that delivers gross margins of 80%.
• This is why despite its spectacular rally it is one of the cheapest AI opportunities available in the market today (Meta, Microsoft and Google are not AI investments) and so if I had to have a position in generative AI, Nvidia would be it.
• The problem that Nvidia faces is that it has 85% market share in a market that is currently growing like crazy and may grind to a halt at any moment although there is no immediate sign of this.
• When the hype bubble bursts and the excitement dissipates and gives way to reality, there will be a reset and there will be nothing that Nvidia can do to avoid that.
• That being said, Nvidia will suffer much less than the rest given it has real revenues and cash flow but it will suffer nonetheless.
• This is why I continue to look more laterally at inference at the edge or nuclear power to run all of these new data centres that are springing up.
• I have positions in both of these themes.