4DS - Anything but Charting, page-23514

You must be losing your intellect in your old age.

Tens of millions is 10^7 as a starting point.

As its tens and not ten, let's assume its 3.4 x 10^7.

Then, they tripled it (maybe more, but let's be conservative) i.e. 10.2 x 10^7 or pretty much 10^8.

Enough to last over 3 years cycled every second.

What else do we know?

We know that current spikes that degrade endurance are unavoidable in the testing of non platform lots. This should theoretically improve endurance when currents are controlled by a transistor in a platform lot and Wilbert has high confidence that the modest degradation will be reversed.

We also know that interface switching reram is a less violent process at the physics level than filamentary nand flash. Current can be scaled down with the size of the cell. This all bodes well for endurance when cells are arranged in an array next to each other, as there is less chance of a cell degrading its neighbouring cell.

Finally, we know that commercial test equipment can actually definitively test the upper limit of our array.

10^8 and we are home.