WHAT’S DRIVING THE EXPANDING MARKET FOR SILICON IP?
The fast pace changes occurring in pop culture is having a profound effect on the consumer electronics industry, especially handheld devices that have become the status symbol of socially networked hip 20-somethings and younger. A quick look at the numbers supports this assertion. Smart devices are being replaced every 11.5 months, according to Tomi T. Ahonen and Alan Moore in their book “Communities Dominate Brands: Business and marketing challenges for the 21st century.” Apple alone had 435 million as of last June and are is activating 12 million new iTunes accounts every month, according to market research firm Asymco. And if market share data is correct, Google—its OS’s adoption is growing faster than Apple’s—is doing even more activations and is growing new activations at a greater monthly rate. All this means enormous pressure on smart device manufacturers and the chip companies that supply them.
There are two critical elements in every smart device: the CPU and the memory that serves this computing engine. The most obvious memory is flash and SRAM as the former provides the massive non-volatile storage medium and the SRAM the high-speed cache that compensates for the slower access of the larger non-volatile data store. What gets overlooked is the small amounts of non-volatile memory that provide security to smart devices that are easily lost, stolen, or hacked without the owner’s knowledge and conditional access to content providers and retailers wanting to serve these hundreds of millions of affluent consumers.
The business fortunes of silicon intellectual property (IP) vendors in general and non-volatile memory IP vendors in particular are intricately linked to the rapid pace of device replacement and the security concerns cited above. And the fortunes of third-party silicon IP vendors are improving as a result. Economics and fear are at the base of this trend. The success of ARM and Imagination Technology demonstrates the impact of economics. No major fabless semiconductor vendor hoping to supply chips into a major smart phone design is going to devote manpower and money to create a new CPU or graphics processor architecture. Both can be purchased from third-party vendors with the scarce low-power and high-performance CPU know-how for today and a roadmap for future products.
Engineering skills that were once plentiful thanks to engineering schools producing large numbers of graduates are becoming increasingly scarce. And the talented engineers that once existed are now beginning to retire. Nowhere is this reality more apparent than in the number of engineers skilled in memory design: flash, DRAM, SRAM, one-time programmable (OTP), and ROM. Buying this expertise in the form of silicon IP is the only recourse left to fabless chip companies faced with shrinking design windows and rapid migration to smaller process geometries. Flash and DRAM are integrated in the form of dice integrated in multichip modules. SRAM and some amount of flash are embedded on-chip for design teams able to find flash processes at the process nodes required for their SoC design. Most of this memory integrated on-chip or in package is acquired from third party IP vendors.
Consider the simple fuse, once a commodity provided by the foundry as part of any contract. Today, as the amount of one-time programmable memory has grown, the physical size of the fuse in silicon has sent designers looking for alternatives, especially at smaller process geometries. And fabless companies with their own internal OTP design teams have had to ask should these designers be better utilize on higher value elements of the larger SoC. These factors are contributing to increase demand for OTP memory vendors such as Kilopass.
Finally, the one compelling factor driving demand for anti-fuse OTP vendors is security. Most smart devices are accessing online content from commercial vendors that have relied on the conditional access protection afforded their content by cable TV providers. Transferring that protection mechanism to the portable device gets more difficult. Putting the keys in the SIM card used to access wireless service provider networks leaves the keys open to hacking from anyone that can access the SIM card. Burying the key inside field programmable OTP memory on the SoC makes hacking much more difficult. Storing the keys in anti-fuse OTP makes the keys as safe as they are in cable TV set-top boxes.
In the future, as smart devices finally become accepted by consumers and financial institutions as digital wallets, anti-fuse OTP will bring the same security from hacking to financial transactions that now exists for multimedia content access.
For these reasons, the fortunes of anti-fuse OTP vendors are improving and will expand the total available market that will ultimately grow the business.