UBS Global Technology Conference


Sanjay Mehrotra (CEO) and Mark Murphy (CFO) provided an update on the current quarter, along with new guidance for FQ1-24.

  • · They took guidance up by around $250M, from a midpoint of $4.4B to “approaching $4.7B.” This is an example of the executive group being too cute in their language. Just raise the midpoint and, if necessary, change the range. Don’t reframe it to make the revenue guidance seem bigger than it is. Investors are smart enough to see through that and will realize the executives care more about appearance than substance. They took gross margin 375 bps more positive, to a midpoint of (2.25%). The Market reacted negatively to raising operating expenses from $1.01B to $1.1B. The explanation is given below. EPS was taken to ($1.17) from ($1.24).
  • · Higher operating expenses guidance was from timing of some R&D expenses, timing of asset transactions, and higher accruals for employee compensation, based on the improved outlook. Increasing operating expenses by almost 10% is a lot and may indicate either a lack of focus on cost controls or an overemphasis on temporary cost reductions. Specifically, the lower cost of employee compensation from a smaller bonus pool (the result of the memory market being worse than forecasted) is a temporary expense reduction.
  • · The current fiscal quarter ends this Thursday (November 30th).
  • · Gross margin is expected to be positive in the second fiscal quarter, ending at the end of February 2024.
  • · 2024 will be a year of pricing recovery for the memory industry. 2025 will be a record year. Nominal pricing (whatever that means) is all that is needed for 2025 to be a record year. Mehrotra’s point is they are not modeling an extraordinary recovery in pricing for 2025 to be a record year.
  • · HBM requires 2x the number of wafers to produce an equivalent number of bits as normal memory requires.
  • · By the end of fiscal 2024 (late August 2024,) the company expects most of their excess capacity to be back online. There will still be underutilization in legacy capacity.
  • · Their target inventory level is 120 days. By the second half of FY-24, they expect to be “within a few weeks” of their target inventory levels. This statement is vague to the point of meaninglessness.
  • · According to the company, Micron’s NAND market share is 12%.
  • · The restrictions on access to the Chinese markets put in place by the CCP have particularly affected the company’s sales into the data center and networking segments.
  • · High levels of investment into memory, particularly DRAM, by indigenous Chinese companies are threatening to products from the legacy nodes rather than leading nodes. This puts more pressure on Micron’s legacy nodes, which are already under stress from the weak market conditions.
  • · Long-term cost reductions are expected to be high-single-digits % in DRAM. Cost downs will be lower in 2024 because of slowed technology migrations and the conversion of capacity to HBM and DDR5, both of which have a higher cost per bit because they have lower die efficiency.


Having Mehrotra on with one of his staff brings into stark relief how long-winded he is. The CEO’s opening statements went on and on while the CFO spoke briefly. This would be a feature if Mehrotra had interesting things to say. He does not. He repeats the same canned points about how memory content is growing, how the latest innovation will consume more memory, etc. He may not have any interesting insights to share, or he may not want to say anything remotely controversial. Whatever the case, every time I listen to Mehrotra speak, it is mostly a waste of my time; a small number of useful points drowning in a sea of words only an IR lawyer could love. As for the guidance change, the memory market is improving more quickly than forecasted two months ago, but not by as much as I hoped. So far, the slope of ASP increases is shallower than I thought it would be at this point in the cycle. I suspect the slow release of inventory is buffering a larger supply-demand imbalance.

  • S.H. (cyclical long MU)
1 Like