Following a pretty good third quarter report from Intel (NASDAQ:INTC) last week, and year-long speculation that the company can regain market share from Advanced Micro Devices (NASDAQ:AMD), comes an interesting note this morning from American Technology Research analyst Doug Freedman about the particulars of those new chips.
Freedman says Intel may be able to trim some R&D spending with future chips because it will be making processors for various kinds of computers out of a single kind of silicon die, rather than having separate semiconductor die sizes and types for different chips for notebooks, servers, etc. That could help the company target lower prices for its chips
But there’s a downside: It’s more eggs in a single basket as Intel puts more of the company’s product line at risk should chip production encounter any hiccups:
The Good: […] much lower R&D spending as mask sets are multipurpose […] R&D reduced with all effort focused on new die type, enabling the new architecture every two years (i.e. a new die type) […] we would expect Intel to continue to be price aggressive in high ASP markets as it likely has die yields to premium grade server and desktop products that are above market demands.
The Bad: The base die better be good - or many products will suffer as a result. The single die approach comes with high risk from technical compromises that will impact the company’s complete product line. […] Products may find the way into unauthorized uses as the die types are the same with package and test the only differentiators. […] Customers are likely to continue to apply ASP pressure as they recognize the cost savings Intel will be enjoying with the new product roadmap.
Freedman rates Intel shares Buy and expects them to rise to $26 over the next 12 months.
Intel shares are up about 1% this morning at $21.53 and are down about 18% this year.
Addendum: And maybe AMD should pray Intel does stumble. Caris & Co. chip analyst Rick Whittington says that with vastly lower production resources than Intel, the smaller chip company could take a beating in desktop chip sales if it doesn’t move smoothly to the next chip production technology, namely, circuits with dimensions of 65 billionths of a meter:
[W]hat AMD must execute is a quick ramp to yield on 65nm while readying a fast transition to 45nm. Should this drag out like the 90nm to 65nm, then AMD will have to hope Intel fails to field a chip-to-chip interconnect until early-mid 2008 (as today appears the case) and does not expedite its own 45nm ramp, now slated to commence mid-2007.
In desktop chips, says Whittington, the price war prices will continue:
That price war may be taking a breather in seasonally strong periods, such as the present (and perhaps even the Vista launched Q107) but what is bothersome is that the ASPs are not coming back up, just settling at the lows. The upside surprise AMD enjoyed in Q2 with notebooks (Turion64) came at a time when Intel only had the 32-bit Yonah to sell. Now that 64bit Merom is starting full volume production, with a sharp ramp slated in 2007, the period in which AMD might be able to protect Turion64 [average selling prices] wouldn’t appear likely to extend beyond Q1.