Memory is a commodity and everyone keeps forgetting it

Memory traded at 1014.74 this morning, +4.46% on the session. I have watched this industry through enough cycles to be suspicious of anyone, myself included, who claims this one is structurally different.

The thing that is always true

Memory is a commodity. Not in the loose sense that people use the word, but properly: an undifferentiated product where the marginal producer sets the price and every participant's incentive is to build more capacity than the industry needs.

Every upcycle produces the argument that discipline has arrived and this time supply will stay rational. It is a good argument. It has also been wrong, in some form, in every cycle I have traded.

Where the cycle actually turns

The sequence rarely varies:

• Pricing bottoms while the commentary is still bearish
• Utilisation climbs before anyone announces anything
• Capex announcements arrive after the pricing recovery is well established
• Supply lands twelve to eighteen months later, into a market that no longer needs it

The interesting window is between the second and third steps. That is where the earnings improve, the announcements have not yet spooked anyone, and the eventual oversupply is far enough away to be somebody else's problem.

Where I think we are

In that window, though nearer the middle than the beginning. Pricing has clearly turned. The capex commentary has started but has not yet become a stampede.

The trade is not owning memory. The trade is owning memory for the specific stretch when improving economics and unannounced capacity overlap, and having a plan for leaving.

The exit is the whole trade

My view is long, and it comes with a leaving condition rather than a target. When capex announcements start clustering, particularly if more than one large producer moves in the same quarter, the clock has started and I will be stepping aside regardless of how the price action feels.

Being right on the cycle and wrong on the exit is how people lose money in semis while telling a correct story. The story keeps being correct for a while after the trade has stopped working.

What would make me wrong sooner

A demand shock, obviously. Less obviously: a capacity announcement arriving unusually early. That would compress the window I am relying on and I would rather step aside early than argue with it.