Maybe the best way to answer is to rephrase the question...

Maybe the best way to answer is to rephrase the question slightly … "What are they trying to achieve with these chips?"

I asked David at the presentation early this year, and he was a little circuitous in his answer (like I didn't realise that he didn't answer until about an hour later. Must have been the cabernet merlot ….).

The obvious seeming route is to be producing chips which will drop straight into a server, in a commercial package ready for demonstration and sample distribution to prospective clients.

The less obvious route is that the chip they are laying down is for testing purposes only, so it is not configured to go into a commercial package, but made to go into a test jig to run limit tests that you often can't do in the normal configuration. This is not uncommon practice with new chip types, as you can use the iteration to test parts of the chip individually in the jig rather than aborting the whole thing (and the eight weeks of work in manufacturing it) because one critical point fails stopping you dead in your tracks. In that case, you often see the results in terms of "We have a damn good idea on how to progress from here" etc.

To be honest, I can't even guess how the end game will play out, and what Jim's tactics are. You may be spot on about the early sale of the IP if they have test results which prove the feasibility on commercial production without actually going to the stage of producing a commercial plug in chip. In that case, the number of iterations would be limited to getting the deposition layers working perfectly across a suitable majority of the wafer and being able to prove that success in a test jig.

This may give them the opportunity to produce chips with slightly different characteristics to prove the ability to manufacture cells with different retention, endurance, speed etc if the differences are more in materials than architecture - which provides a prospective vender with more information from which to make decisions, knowing the cells can be produced in a commercial production environment.

If that happens, you may find a vendor wants to buy the IP but retain the team to continue development to meet their requirements - so maybe revenue to take the project through to the point of development (and in the direction) that they desire. Who knows how much kudos the 4DS team gets for its work?

To go out on a limb … I wouldn't expect more than two more iterations (although it is hard to say without knowing exactly what they are trying to achieve). Whilst it is complex, these guys know what they are doing, and it is really just doing what they have done before in a bulk manufacturing system. As long as they can make the cell work ….. the rest would be a cinch for IMEC.