Micron FQ3 26 – Peak Power

Micron trades for stability.

• A monster set of results underlines Micron’s current market power, which it is wisely using to trade further price upside for longer-term stability, which will shield it when the notoriously cyclical memory market finally rolls over.
• FQ3 26 revenues / Adj-EPS were $41.5bn / $25.11, crushing estimates of $35.7bn and $20.49, and there was more to come.
• FQ4 26 guidance also handsomely beat expectations with FQ4 26 revenue / Adj-EPS $49bn – $51bn / $29.73 – $31.73 far above the $43.2bn / $25.31 consensus estimate.
• The key to these numbers was the gross margin, which has now reached an all-time high of 84.5%, with 86% forecast in the coming three months.
• This is confirmation that Micron has far more demand that it can handle at the moment, which, as far as it can tell, will continue beyond the end of calendar 2027.
• It is in 2028 that it expects the shortage will be somewhat addressed with new fabs coming online, but delays and overruns could push that out.
• While Micron insists that the days of cyclicality in the memory market are over, its actions indicate precisely the opposite.
• Here, it has closed 16 long-term contracts with customers, which in the data centre run for 5 years and in the vehicle run for 3 years.
• These are what it refers to as strategic customer agreements (SCA), the core terms of which vary from one customer to another.
• These contracts guarantee a certain level of volume and in some cases have fixed prices, others variable prices and in still others, ceilings and floors.
• It is the SCAs with the ceilings and floors that are the most relevant, as the biggest contracts are of this type, and the ceiling is set at current prices and the floors at an unspecified level, but I would guess at something like a 20-30% discount to current levels.
• This is a classic strategy to trade further upside for downside protection against a collapse in pricing, and it will bring greater stability to Micron’s earnings and substantially delay the fallout of a major correction in memory pricing.
• When all of the 16 SCA’s are completed, Micron thinks that they will cover 40% of its revenue, and I suspect that Micron will continue to push for more of these deals to get a higher percentage of revenue covered.
• This is the first concrete indication of a memory maker hinting at the top of the cycle as the large investments in capacity begin to come online, and Micron is very wisely leveraging the power it has now to cushion it when the memory market goes into reverse.
• SK Hynix and Samsung are doing exactly the same thing, and as a result, when the top of the cycle in memory comes, it won’t be a disorderly and rapid collapse but a slow and steady degradation over the course of 5 years or so.
• This is a much more preferable state of affairs, and as long as the clients can’t cancel their contracts, Micron and the memory industry will be less cyclical from here on.
• This means that the PER that is afforded to the memory makers over the course of the cycle will be higher, as their earnings will be less volatile and more predictable.
• This doesn’t necessarily mean that the shares of these companies will rally orders of magnitude again from here, but it does mean that the downside will be much less than before.
• I think that this means that the peak of the cycle is appearing on the horizon in 2028, and the fact that some of the long-term sceptics on AI and the memory trade are beginning to question their conviction means that we are closer to the end than the beginning.
• This moves me closer to thinking about selling my position in Samsung Electronics, but with 18 months to run before any tightness in supply is lifted, the ride is not completely over.