Missed this announcement yesterday, congrats to all holders. A...

Missed this announcement yesterday, congrats to all holders.

A bit of a reality check is in order though.
1. They have produced on 300mm wafers. This is a really good thing. I am assuming at 40nm since that was talked about previously. However, 300mm or 200mm is largely irrelevant, its just the size of the wafer. Lots of commercial prod is still done on 200mm. As to the commercial quality equipment bit, do you think there are any "non-commercial grade" 300mm fab equipment out there? Developing fab kit is an extremely expensive exercise to get to production in its own right, anything you are using is "commercial standard".

2. The announcement is a bit scant even if you trumpet that short and succint is good. Plainly they believe they have functional chips or cells off the wafer, since nobody puts a wafer in their machine, they are cut up into chips. Why didnt they just say that? "Our cells work on commercial wafers". So some basic testing must have been done to ascertain there are working memory cells there, which wasnt clear on the first run.

3. There is a view that you just need to get working cells off the wafer and everything is a done deal. That is not the case. There is the question of them "working" (tick), then there is the question of them "working well" (to be proven), then there is the question of getting enough of those working well chips off a given wafer (yield). There is no point having cells that just barely work or wont last or that work in such small numbers off the wafer that you cant get a commercial yield. 300mm is there for higher yield, it better yield high. None of that has been answered yet except they have some working cells or chips.

4. And, they havent told us if the quality off the wafer is consistent to put full chips together. Its fine to have working cells, but those working cells have to be sufficiently clustered together relative to the non-working cells to get a working chip off. By this I mean for a 1MB array, you've got to have 1MB worth of working cells in the array. I'd give it the benefit of the doubt here though.

5. Optimisation. This phase is crucial and skirted over here in impatience that is retail shareholder land. We know from other chip developments and probably a good example is WBT. They factored 12 months into their optimisation process. This seems to be the norm from what I can see. I wouldnt be believing any story that says the chips will be production ready in April since reram relies on very careful material effects and 4DS wouldnt want to design it so it cant scale down size wise later on as well etc. They would hasten slowly from now on since any potential buyer would demand a quality product. In addition, they would probably also want to package some chips up for potential buyers to test off this wafer if its as high quality as we'd hope.

Anyway, a really good step forward for 4DS in my book. They now have something to work with that is physical and tangible at a commercial size.