Note: This is part 2, part 1 can be found here.
This is where the second part of Intel’s spending comes to bear, they bought or promised to buy 15% of ASML. For the moment, it looks like this investment was offered by ASML rather than asked for by Intel, and there is another 10% up for grabs. The intent is to have all the players, or at least the major ones have a chunk of ASML.
Why? Some say it was done to ensure future order timing from ASML, others point the finger to fabs trying to influence development directions. In either case, it says in very clear terms that vendors can no longer develop the next generation of tools without knowing who will buy, and how much they will spend. Worse yet, they can no longer afford to do it on their own, the costs are too high even with guaranteed orders from a single big customer.
If this sounds like a dire problem, basically that the industry can no longer afford to make the next generation tools, it just means you understand the problem. The industry really can’t afford the next gen tools, and so they are telling the customers to pay for R&D up front. And ASML is forcing them to put some skin in the game too.
In Intel’s case, this amounts to $3.1 billion for 15% of ASML, with a further $2 billion in stock being shopped around to the other players. If the next gen tools don’t sell, the big players are collectively out $5 billion no matter who’s fault it is. That is a lot of incentive to buy the tools ASML makes, and not press to hard for discounts while you are at it.
Litho equipment tends to be the leading edge of the tool chain, and if other vendors are not in a similar bind, they will be soon. If ASML and Intel are willing to make these rather unprecedented financial moves, others are sure to attempt it in short order. Please note that while there have been much smaller incentives and collaborations in the past, there has never been anything on this scale that SemiAccurate can recall.
Once the major tools vendors have moved to this model, you have to ask yourself if there is a tools industry left at that point. If all the customers have large stakes in the companies making the tools, is it an industry or a joint effort among the foundries that just happens to have a stock ticker? In any case, the line between an incentive and the end of an industry is mighty fine.
To complicate an already messy situation, there is a bigger concern, the semiconductor companies themselves. If you didn’t notice, there isn’t a lot of love between competitors on this level, and the few cooperative efforts that exist tend to be more forced marriages than anything else. If the players are going to all get together and jointly fund the next set of tools, does anyone else see a royal mess developing? If one fab wants to go one way, and one or more another, does it help or hinder things if they all have board seats with the tool vendor?
Worse yet, can a vendor develop a tool for a bleeding edge player without tipping off the others to some very sensitive time tables and industrial secrets? If ASML decides to push for tool XYZ to be ready in 18 months, and that tool can only be used for ABC, can they keep that very likely material information from Samsung and TSMC if those two have board seats?
As things stand, there are few secrets in the industry left, and this type of investment deal could very likely destroy any chance of there being any from now on. I don’t think any player wants the line between competitive intelligence and quarterly reports to blur to that degree, so what can the industry do? Messy is far too kind of a description.
ASML is also being quite public about courting Samsung and TSMC to join up and buy the remaining 10% of the shares offered. From their perspective, having the three biggest players in the industry on board is a good thing. For Samsung and TSMC however, they are precluded from reaching the same stake that Intel has, so even together, they will not have the same level of influence.
In their conference call on the subject, Intel was very clear that this investment would not lead to them getting any type of preferential access to new tools and nodes, but given their already sizable lead on the industry, the real question should be if anyone foresees another player wanting the tools in the same time frame. That said, one thing that was not asked, or for that matter preemptively denied, was whether Intel would get preferential influence over development directions.
The $4.1 billion Intel is laying out says yes. Even if one player stands up and buys the remaining 10% of ASML, they won’t have the same stake as Intel, a fact probably not lost on anyone considering investment. Is a perpetual #2 position worth a billion or two? Is that #2 position still worth it if Intel has no choice but to continue the funding or effectively idle themselves until the industry catches up?
So now the real question is, will the other industry players join up and consign themselves to perpetually picking up the scraps from ASMInteL, or will they all step up and back the other players, Nikon and Canon? ASML is far and away the leader in litho tools, so would the rest of the industry be wise to bet against them? In any case, it has been nearly two weeks, and no players seem to be in a hurry to jump on board.
To take the broader view, will everyone join together and set the ball rolling for the ‘not a consortium’ model, or will they fight it hoping to hang Intel out to dry? Intel can fund the development of the tools on their own, painful though that may be, but will they? Will there be a tools industry at all in a few years, or will they just be internal R&D arms of the remaining semiconductor houses? Who knows. The one thing you can say is that it may not be clean, easy, or straightforward, but it will definitely be expensive. Very very expensive.S|A
Latest posts by Charlie Demerjian (see all)
- SemiAccurate has Skylake-SP die shots, sizes, and more - Jul 21, 2017
- Skylake-SP has a diverged core - Jul 19, 2017
- Intel’s Purley platform architecture is a step forward - Jul 13, 2017
- Intel launches Purley aka metal Xeons - Jul 11, 2017
- AMD’s Epyc has lots of connections - Jun 26, 2017