I wouldn't mind identifying any of the really high risks we see with 4DS. I am LTH on 4DS and I think the upsides outweigh the downsides at this late stage but that said this is still a speculative stock.
So here is how I see the key risks for 4DS.
1. The metrics aren't good enough given the expected $/GB manufacturing cost. We already know the tech works, but in my view if
A. DRAM-like speeds, with
B. Data retention orders of magnitude longer, with
C. Product endurance to at least match hard drives, isn't also matched by
D. Manufacturing costs at least as cheap as DRAM now,
Then - it isn't worth as much as we all hope. The loss of any one of those four factors makes the product more niche and/or more risky. Still valuable, but maybe requiring more research and tuning.
In the advantage of 4DS in this space is imec is manufacturing it on current SMC foundry equipment so it won't require major retooling to make - that is a huge plus. It is also 3d, like NAND and unlike DRAM so another plus.
2. However, there has been little discussion about another key risk, which will likely be resolved at the megabit chip - defect rates. The test pieces so far have been 8kB or 16kB I think? We know how CPU and graphics card manufacturers have managed to get around chip defects often by slowing down CPU speeds or disabling certain cores and repackaging these units as slower models. I don't know how this works with memory manufacturers and would like your thoughts on this.
However it is safe to assume with any new manufacturing process there will be defects at first which will increase initial costs. If this is very high, expect multiple iterations of the Mb chip, or a correspondingly high $/GB manufacturing cost, and lower take over value.
In 4ds's favour is again imec and the WD agreement to assist with these challenges, as well as the manufacturing iterations already completed - no doubt some healthy data on defect rates which we can only hope has been manufactured out. (As an aside I would expect at least some wafers failed, perhaps completely, else the quarterly updates would have said they all worked). I can only imagine the millions of cycle times mentioned is also favourable to a long expected lifetime, if the manufacturing conditions that create those can be easily replicated for mass production.
3. Too late to the market. There are many competitors already and if it takes another 2-3 years to even commercialise, let alone achieve some market penetration, then it is not worth as much. There have been a good few articles expanding on the market in the already memory space posted here today so I won't repeat them. It's a $100b market so far from small, and there is space in the market to fit many types and suppliers, but 4DS has to get there soon to get first mover in storage class memory. Next year really is the litmus test for commercial viability and if the integration with imec's memory chip goes well then who knows how far it could go.
To conclude - my view is 4DS has derisked a few of the above concerns - enough for me to take the risk and hope for the big payout. But DYOR and good luck all.
Expand
