Analyst Opinion – On the surface, the cooperation between Intel and TSMC seems like a routine cooperative partnership announcement. Yet with this one action, if managed effectively, Intel has almost assured Atom’s success as a major player in the growing world of consumer electronics.
Intel and Taiwan Semiconductor Manufacturing Company (TSMC), one of the world’s largest chip foundries, just announced a marketing collaboration involving Intel’s Atom processor. Atom is Intel’s effort to downsize its processor chips to fit into the realm of emerging smart devices below the Personal Computer space. TSMC will work closely with Intel to port some of the Atom processors to its own process and design flows. TSMC will also have the ability to do engineering on the chip to build customized versions for the large number of existing TSMC customers. However, Intel will have ownership of the final device and the customer, as Intel will be selling the custom designed chips that TSMC designs and builds in its foundry.
As PC sales wane, and their chip revenues along with them, Intel looks to additional sources for revenues. Consumer products represent a massive potential market, though at clearly lower margins and price points. But, Intel’s cost of operations makes it a supplier at too high a price to go after the cut-throat and highly price sensitive consumer market. And Intel is not set up for customized, System On Chip (SOC) solutions the market demands. Enter a partner that can bring all of this capability to Intel – TSMC.
This is a direct attack by Intel on competing processors, especially the ARM processor, which is trying to move upstream from the smart phone and embedded gadgets market it currently dominates, while Intel is trying to move downstream with Atom into this overlapping space. The battleground in the middle will be aggressive and likely bloody, with huge potential returns. And while Intel’s attack is primarily on ARM, it also has profound effect on other players – AMD, Qualcomm (Snapdragon), Nvidia, TI, and even Marvel to whom Intel sold off its own ARM-based processor (XScale).
A win-win for Intel and TSMC
Intel gets a vast new market potential for Atom since TSMC has connections to many consumer and lower end PC-type products (e.g., MIDs, webtop devices, netbooks, media servers/set-top boxes, etc.), especially in the important Far East markets (Taiwan, Japan). TSMC gets to offer a high performance processor it did not have to design, but that it can customize for the clients who will take volume products. It also adds the ability to merge the work TSMC is doing on WiMax enabled devices and couple it with Atom processors.
Intel gets to have customized designs done by TSMC for a number of volume customers, a service which Intel does not generally do all that well, and is not set up to do efficiently. What Intel gets is a large potential for embedding chips in products they might not otherwise be able to reach, and generating revenues from the chip sales.
This is a very symbiotic relationship. Intel brings the core Intel Architecture-based Atom. TSMC brings knowledge of SOC creation and customization for specific customers, high volume production at low cost, and a customer base of many consumer oriented vendors. It’s a win-win. One thing to note is that this is not an outsource of the Atom per se. Intel will continue to build standard Atom chips in its own fabs for its mainstream customers while TSMC will only build customized silicon in high volume.
Extending Intel’s chip business while staying in control
Intel gets to own all of the customers that come out of this relationship, requiring that all chips sold come from Intel. That means Intel gets to set price, choose who the clients are and what they can build with Atom, and make the margins it wants and needs. TSMC brings customers to Intel from its vast installed base, while giving its customers potential new and powerful chips for the increasingly complex consumer electronics devices coming to market. However, there could be some potential conflict if TSMC brings a customer and Intel says no, or wants to charge too much, since Intel has ultimate control. Further, there may be some conflict over ultimate ownership of the customization components. And there is a small risk that Intel may not easily be able to combine what TSMC builds in its fabs as the processes are not identical. Overall, however, these risks seem minimal and manageable.
This is a smart move by Intel, and a win for TSMC as well. Unless something is inherently flawed with the architecture or Intel is too controlling of the relationship (or tries to charge too much for the finished goods), it is almost certain that Atom will be successful and permeate a large swath of consumer devices territory – an area in which Intel has previously not been a major player.
The opinions expressed in this commentary are solely those of the author.