02.26.2010 SanDisk Investor Day Q&A
[This transcription is not complete. I transcribed the portions that seemed most interesting.]
Eli Harari, Chairman of the Board, and CEO, etc
Q: One last question on SSDs. Your company has gone through 3 iterations of SSDs. There are some companies that are having some success like Intel is working with Hitachi on SSD. My question is the long term success model for SSD. Is there a piece that you guys are missing? You guys have the know-how in NAND. You have the controller technology, but is there a piece that a JV can do an HDD?? company that can make you successful in SSD in the long run?
Yoram: I don’t think we are missing any component.. I think that we have all of the components as you said- controller, memory and the system knowledge, to do that.
I think that it is a learning process. It is a different market than we were used to in the imaging and the handset, but I think that in this drive that we just released, it is what we call the balanced drive. We have really optimized for the user experience.
We don’t think necessarily that the highest performance is the most important. Just measuring to some PC mark or some other benefit ??? I think that the real measure in doing a balanced drive is very good endurance, has very low write amplification ?? it is a term that we use for increasing the endurance. And essentially it is already 3 to 4x faster than a hard drive. The system itself is sufficient. You will not see any difference any more.
So it is really giving a balanced drive that is good, that is reliable, and that gives a good user experience in the PC.
Eli: I think that your question is are we planning to be in the enterprise for example? I think that is really the question. I think that today of course we are not addressing the enterprise. We are addressing more the biggest part of the market opportunity, which is the notebook and netbook PCs.
There is nothing to stop us from in the future, either being directly in the enterprise through a partnership or supplying components to the enterprise space…
[relatively early in the Q&A]
Eli: Let me add to that, I think, if you do the math, and some of you, I know do the math, to go from an industry supply of 10- 11,000 PB in 2010, most of it is at 32 nm, to 70,000 PB in three years. Its tough to see how the industry can achieve that.
Now that 70,000 PB may not happen. But we do think that the growth engines that we discussed today are real and that we are very well positioned to play a significant role in those. So its our optimism about growth outpacing supply or matching supply- more or less.
This is not saying that there won’t be any down factors. I’ve said that before, but in general, looking over the next decade and certainly over the next three years, we think that the growth trajectory is hockey-stick- is going to be real and that’s what is going to moderate the pricing situation…
Q: The second question is in one of the slides you were comparing SSD with HDDs, laptops and the price delta is still huge- like 1 GB of HDD is about lets say $2 while 1 GB of SSD is more like $20, so almost 10x or 12x type of delta- so if you flash forward to 3 years- where do you see that price premium going to? And in your view of demand, how much premium can SSD command and still be deployed in mass volumes? Is it 2 times? 3 times? Or do you think it could even be higher premium than SSD can demand?
Yoram?: I think again, we showed [pause] I think the premium is going to be where they are using or the application or the device is using the intrinsic value of flash. And when the intrinsic value of flash comes into play and plays a central role in the consumer experience of that device, then you have a premium.
When you have a situation where you have a device where a HDD or flash can play into that equipment almost at the same point, but flash has some small advantages, then the price premium is very small.
So I think what we feel is that based on some trends that we see right now in the market, that the trend is going to be having much better consumer experience devices that are very thin, that are low power consumption, high performance, and this is what is going to be the trend.
If that is going to be the trend, then flash is going to sell at a premium relative to hard drives in those types of applications.
I don’t know its hard to say. As volume grows, probably the prices, the ASP is going to be probably somewhat lower than what we showed in the slide, but it could be, I don’t know, its hard to estimate. But it could be 2x, maybe 3.
Eli: Let me put it maybe another way. You have seen in the last five years we have cut our costs by more than 50% each year. The price came down by more than 60% each year- almost. Where we are today- with SSDs price wise or cost wise- if you cut that by another factor of two, definitely it is going to have a huge impact on adoption.
If you cut it by another factor of two- it is done- you have reached the tipping point for most people. So can now we do it- with 1x nm? And can we make money on it? That is the challenge for the industry. But you know, its moving in the right direction and SSD will have to compete with other products as far as margins.
And if the margin is not there- for us- then we will allocate less of our captive supply to it in favor of other applications.
This is the beauty of diversification. You have diversified channels, both in terms of product, it terms of markets, and in terms of channels. You don’t want to starve to death any one of those, but you certainly can dial in and out the amount of capacity that is allocated and the gross margin contribution is going to play a role in that…
[middle-ish in the Q&A]
Eli: I tried and I think you got from the other people, [missed words] very strong belief that the market demand is going to grow at a very rapid rate that will require additional new production capacity in the industry.
You cannot grow from 10,000 PBs [missed words] to 60, or 70 or even 40,000 PB without new fabs. The question is not if, the question is when.
Q: And magnitude
Eli: You have to build a mega-fab because otherwise you are not going to make much difference. [laughs] First of all, but more importantly, you are not going to be competitive.
Small fabs don’t cut it. OK, its really a mega-fab or stay at home.
That the magnitude. But that by itself has an implication on when, because if you could do it in tiny little increments, or something [you can’t do it in small pieces]. Now I would say that those mega-fabs don’t necessarily ramp overnight. This is not like a 6 month exercise.
A mega-fab could easily take two to three years to ramp- and the investments would be spread accordingly. I said in my presentation, that we believe that for us, SanDisk, regardless of what anybody else says??- it is premature for us to make that investment decision at this time.
I think that the industry itself, definitely will hinge on such decisions from other companies.
Q: Thank you
Q: What is the current installed capacity in fab 4? With the 10% increase in the JV capacity, what will the capacity be at the end of this year? What is the maximum capacity in fab 4?
Sanjay: I think I mentioned that or maybe had mentioned that. With the approximate 10% increase in capacity that we are looking at, we will go from 1.5 million wafers last year to around 2 million wafers per year for the SanDisk portion- that’s the 10% plus the remaining portion, the 10% we are looking at doing this year. There is additional clean room space available, that will all take it to 2 million wafers per year.
And we also said that decision regarding the remaining space we have not made that yet. That expansion would take place sometime in the 2010/2011 timeframe.
The 10%- we are making that expansion in 2010. Even when we made the decision to do the rest of the expansion, it takes three months of lead time. It really is realistically a 2011 expansion.
Q: All I am trying to find out is at the end of the 10% expansion, how much space is there? Is there enough space to continue expanding in 2011?
Sanjay: Yes, there is space available, there will be space available, in the clean room after the approximately 10% expansion in the second half of this year.
Eli: Perhaps another way to look at it. fab 4 today, is definitely more than 50% utilized. Really a relatively small part of fab 4 will be available for expansion…
Q: The second question is regarding the future fab plans. There is some mumbling about Toshiba looking at a future fab 5 and what is SanDisk’s position in alignment with Toshiba on a future fab 5?
Eli: Of course we are very strong strategic partners with Toshiba. We are in constant discussions with Toshiba about a lot of things regarding the joint venture that we have between us. But the two companies of course have the freedom to make their own decisions based on their own market requirements and their own, basically, circumstances.
So as I said again in my presentation, we will continuously evaluate the market situation and we will proceed with caution.
[near the end of the Q&A]
Q: First of all I’d like to congratulate you on a great investor day.
Eli: Thank you.
Q: I have a question. Going at it a little different way. Right now you said that the industry capacity is about 10 petabytes- about 100 million wafers. So roughly 1 terabyte a wafer. We are somewhere in the 34 to 43 node, nanometer node. Somewhere in between there on average. How much capacity from the existing 110 million wafers would we get to in 2012, 2013, once we reach the 1x node?
Eli: Actually I said about 10 million wafers annually for the whole industry and about 10 to 11,000 petabytes [PB] which gives you- you’re right- around a terabyte [TB] per wafer. That is more for the 40 nm technology on two bits/cell. [1 PB = 1000 TB = 1,000,000 GBs]
For the 32-34 nm its more in the 1 1/2 TB or so. A little less if you don’t use three bits and more if you use three bits per cell.
You know, you can do the math. Take a 32 nm multiply it by itself- 4f squared and the next generation may be 25 or 24- do the same- that gives you at the rawest level how many more bits you can cram into that same wafer. And basically you will see that its slowing down.
Q: So that is the 30 to 40% cost reduction.
Eli: Yes. So lets assume that you are getting 30% more bits per wafer going to 20x and let’s say another 30% going to the 1x.
Q: OK very good. And how many new fabs would need to be built then to reach your estimated demand in the out years? At one point a few years back you were talking about 6 new fabs.
Eli: OK, and don’t hold me to it. To do that math.
Lets say that 1x generation product is between lets say 2.5 and 3 TB per wafer, approximately that- lets say 3 TB per wafer. You have to be really really good to get 3 TB per wafer, and probably use 2 to 3 bits per cell.
That means 30,000 PB in 2013, an incremental 30,000 PB would require another 10 million wafers which is 5 mega-fabs, assuming 1x @ pretty good yields @ 3 bits per cell.
This is the point I was making earlier- that the projected growth in demand requires very substantial new capacity.
The existing capacity is already generating 10,000 PBs- 11,000. So lets say it can go to 30,000 PB. If you want to build another 30,000 you need to have another 5 or so fabs- mega-fabs.
Q: On an average of $8B per fab?
Q: Roughly, plus maintenance capex. That’s a big investment- a big risk.
Eli: Yes. Somebody told me that a nuclear aircraft carrier cost about $5B. That gives you an idea how big $8 billion is, but as I said, it’s spread over a three year period- investment wise. In any kind of new capacity it would be very likely with Toshiba so our number would be less than that number.
Q: I hope so.
Eli: Me too. [laughter]
?: A question about the future technology and 3D. Is there a difference in the capital intensity between the two technologies? Is one an easier wafer to build than the other?
Eli: 3D would require greater intensity of lithography because you are going say up to eight levels, you have the opportunity to generate 8 levels of memory and each level has a certain number of critical stages. So yes you would need more lithography and therefore the balance in that fab would be much more lithography and etch than furnaces lets say.
But the idea is that after you account for all of this- you take all of this- more lithography- it still comes out cheaper per GB and hopefully it starts a new scaling path. It reenergizes Moore’s law.
We can probably take one more question.