Top 10 Embedded Trends for '10 - Part 2

This is the second part of a two part blog post on the 10 most important trends that VDC is following for the embedded systems market in 2010. Part 1 can be found in the post directly below.

6. Further Consolidation of Semiconductor Suppliers

The prohibitive economics of the semiconductor market are driving consolidation all around the market. It is likely that further industry consolidation will also touch the embedded market. Some of the potential scenarios are outlined here.

A large microcontroller supplier, like say Renesas, could acquire an x86 license that could then someday be used as the core architecture in its microcontrollers instead of RISC/Power, by picking up the smallest of the three x86 licensees – VIA. This would allow MCUs to take advantage of the huge ecosystem of programmers familiar with x86 and tools built for x86. Perhaps a large multi-faceted silicon vendor, Freescale, would then also be rolled into that aggregation creating a semiconductor behemoth with dollar volume sales approaching that of Intel, with the vast majority of it based entirely in the embedded market.

Intel itself is flush with cash and has stated that the embedded market is one of four markets targeted to drive corporate growth strategy over the next five or so years. This could very well spell out an Intel acquisition with implications for the embedded market. As programmable logic proliferates in the embedded market, Intel might decide that the programmable logic market is worth a second try to help its stated goal of improving its footprint in the embedded market through the acquisition of either Altera or Xilinx.

There are also a handful of many-core processor startups that could soon be acquisition targets if their technology is proven to be worthwhile. Quanta, the Taiwanese ODM, has already taken a stake in the many-core startup, Tilera, as of late 2009. Although it may be hard to predict exactly what deals will get done, it seems virtually assured that some consolidation will take place.

7. The Foundation for the Many-Core Market is Laid

The numerous massively parallel/many-core processor startups continue to advance their technology. For the past three or four years this has been emerging technology that provided workable silicon solutions, but has lacked software tools and programming experience to match the silicon. These vendors are beginning to make some real progress in building the software infrastructure around their silicon that will be required for their solutions to be usable. While, many-core processor technology will be an on-going work-in progress that is likely to take another 5-10 years, 2010 is likely to be a year in which some significant advancement is made and more design wins are gained with early adopters. The many-core/massively parallel segment will always be a niche high performance market and is unlikely to ever become the mainstream in embedded applications, but VDC currently projects that this could prove to be a very healthy niche for a few vendors.  The most likely scenario is that the many-core processor market will someday resemble the current FPGA market – large enough to support two sizeable vendors and two or three small players. The foundation for those two leading vendors could be laid in 2010.

8. Parallel Programming Progress

Closely related to the above trend, progress continues for parallel programming, which is required for many-core processors to operate near their full potential. This is another long-term trend that will continue to unwind over many years, but is likely to experience some major gains in 2010. A number of firms have already done a great job of enabling parallel programming on their many-core processors. Cavium Networks has done a great job of making parallel programming easy on its processors for communications and networking users. Meanwhile Nvidia’s CUDA environment has made parallel programming possible in C for users in a number of embedded vertical markets. In 2010 most of the parallel programming progress is likely to be in the same vein as what these two vendors have already done – enabling current programming techniques to work in a parallel processing environment. Especially enabling C, which is ubiquitous in embedded applications and will not be going anywhere in the next 20 years, to work with parallel processors.

9. Disaggregation of Embedded Computing Networks

As embedded computers are increasingly able to help end users drive agility and intelligence for their organization by providing information from the edge of networks it is causing a disaggregation of embedded computing infrastructure/networks and the enterprise information systems that they are connected to. By disaggregation we mean the movement of functionality and connectivity from the core of networks out to the edge nodes of networks. A new class of embedded computer system, the edge node or scalable edge server, is being created. These are low cost, modular, and scalable platforms that provide functionality and connectivity in the field directly at the edge of networks. The advent of cloud computing will increase the demand for such edge node computers as this class of embedded computer will enable the deployment of cloud services in almost all of the industries served by embedded systems. Without information collected at the edge of networks by embedded computer infrastructure, IT has nothing to do. With the deployment of the cloud this will become even truer. Without a scalable edge infrastructure it is impossible to get useful information into the cloud. The segment of scalable edge node computers will be growing over the next five years, starting today. Many embedded suppliers will be able to capitalize on this trend.

10. Greater Integration for Merchant Embedded Vendors

Merchant embedded computer suppliers have been moving higher up the food chain for some years now. The industry began with embedded hardware vendors selling computer boards to their OEM customers whom were doing most integration in-house. Vendors were then able to move into integrated systems (hardware components such as chassis/enclosure and boards integrated together) and more recently have begun to sell a lot of turn-key solutions (fully integrated systems complete with OS/RTOS, middleware, and even application software). The number of integrated systems and solutions sold by merchant embedded suppliers has been increasing for the past five years. This growth will continue as OEM customers continue to realize that embedded compute platforms are no longer differentiating value add for them. This will allow merchant suppliers to not only to convert more previously captive accounts and grow their total served available market (SAM), but to also sell higher margin products and services. Coming off of the large downturn of 2009 which has caused many large OEMs to slash their internal development budgets, it is likely that the growth of integrated systems and services revenues at merchant embedded vendors will accelerate in 2010 as customers still doing some integration look to outsource and preserve their precious development dollars for higher value activities.