Next Generation Semiconductor Design

The structure of the semiconductor market is going vertical due to the dramatic increase in complexity requiring semiconductor design integration.  End product companies are looking to integrated development platforms not the historical process of attempting to integrate various vendor products and designing components using multiple, disconnected design tools.  There is an acceleration of designers using Xilinx (XLNX) or Altera’s (ALTR) integrated platform for the next generation semiconductor products; creating positive revenue and margin tailwinds for these two companies. 

Xilinx

Every year for the past decade Xilinx has generated positive free cash flow, including the downturn in 2001 and 2002.  It has a global customer base of over 21,000 customers (the top 15 customers account for 33% of sales).  Xilinx has a little over 50% share of the PLD market.  It is the 3rd largest company in the overall logic market.  In 2007, a survey of semiconductor customers ranked Xilinx, for the first time, as the most important vendor, ahead of IBM or Intel.  Over the past decade it has slowly been moving up the list on this survey.

Altera

Altera was founded in 1983 and has been public since 1989.  It has a global customer base of over 14,000 electronics equipment makers. 

PLDs replacing ASICs

Software is now the bottleneck in semiconductor product design.  Historically, various unconnected software development tools from the EDA (Electronic Design Automation) industry were used to design semiconductor devices.   However, due to the lack of innovation and integration in the EDA market designers are looking for other integrated software development options. 

The two largest EDA firms Cadence (CDNS) and Synopsys (SNPS) are in the process of integrating their various development tools, many obtained through acquisition, into integrated software development platforms.  As IC designers’ focus more on embedded systems they are requiring an end-to-end design platform.  However Synopsys and Cadence tools are still fragmented because the tools were independently designed and built; often from different firms with dramatically different software architectures.   The time consuming risky process to re-design and integrate these tools are costing these firms valuable time in the competitive, fast moving semiconductor market.  Both Xilinx and Altera currently have powerful platforms built from the ground up over more than a decade.

The PLD market is growing by encroaching into new end adjacent markets - such as industrial, automotive, and consumer devices - that have historically used solutions built around ASICs (Application Specific Integrated Circuits) or generic processors.  The $4 billion PLD market is growing within the $72 billion semiconductor logic industry which includes ASICs and ASSPs (Application Specific Standard Products).  The number of new ASIC starts has decreased from 12,000 in 1995 to 2,000 in 2002 and may not exceed 1,000 in 2008.  Many of the new semiconductor solutions have been and are still migrating to the PLD vendors. 

A complete ASIC flow includes coordinating with multiple suppliers of services, silicon, tools, and IP; while using a PLD platform is a one-stop shopping experience.  By vertically integrating the chip’s hardware design and offering a software design platform to the end product designer that includes framework IP (intellectual property), the PLD vendors have moved the level of abstraction up a layer; increasing the efficiency of designing digital products. 

Since the complexity of designing a semiconductor device has increased dramatically many companies have outsourced design to third party firms, complicating IP ownership and potentially outsourcing what should be a core competency.  Using a PLD eases a company’s ability to in-source the entire product’s design while also shrinking the time spent on design and testing.

Every year, PLDs make more economic sense as designing and manufacturing semiconductor devices become more time consuming and more expensive.  Initially just a prototyping tool, PLDs can now cost effectively ramp up to 100,000s of units up from 10,000 units five years ago.  Some low cost PLDs such as Altera’s FPGA Cyclone® series are now in devices selling in the low millions of units a year. 

Benefitting from a Strong Platform

The PLD (programmable logic device) market is a duopoly with Xilinx and Altera controlling 85% of the market followed by Lattice (LSCC) with 7% market share, Actel (ACTL) with 6%, and Quicklogic with 1%.  The reason why Xilinx and Altera have consistently grown market share over the years while having industry leading margins is due to the power of their platforms.  Each platform has built a strong moat based on the number of integrated circuit design engineers that understand it (and whose career’s success are largely based on this knowledge) and the amount of intellectual property available. 

Switching costs are high for engineers to learn new design frameworks and every year thousands more engineers learn Xilinx and Altera's development tools in colleges throughout the USA, India, China, and the rest of the world. 

Xilinx and Altera’s development platform each offer thousands of IP cores (re-usable intellectual property) built by Xilinx or Altera (depending on whose platform you are using) and offered for free (or nearly free).  Their platforms also support 3rd part IP cores.  Leading IP companies such as MIPS Technologies (MIPS) and ARM Holdings (ARMH) have made recent moves to offer IP on the PLD vendor’s platforms. 

For ASICs, the largest impediment to IP reuse is verification, which can consume 70% of the development cycle.  With FPGAs reuse is easier as the FPGA vendors pre-verify 3rd party vendors IP.  PLDs allow IP cores to be embedded at the software level, not in the hardware.  This simplifies the process so IP can be tested and verified by the supplier on a finite number of chips since PLDs are a common single chip manufactured in volume by the fab; as opposed to ASIC or ASSPs whose physical design is dictated by the IP cores themselves.  Another huge advantage that the PLD firms have over the EDA and ASIC firms is a much easier migrating path from one manufacturing process to the next.  This allows IP cores to be easily rolled into the next generation architecture. 

It would be hard for new entrants to overcome the platform network effects that have been built by Xilinx or Altera.  The ability of Xilinx and Altera to build developer ecosystems may eventually trump all but the highest end digital and mixed signal chips. 

An Integrated Solution

With Xilinx and Altera offering end-to-end design capabilities from design tools to IP cores, including free IP cores and a marketplace for 3^rd party IP cores, they are creating an ecosystem where the customers have one neck to choke when problems arise.  Creating an environment similar to IT departments' migration to a single primary provider (Microsoft (MSFT), Oracle (ORCL), IBM (IBM) or SAP (SAP)) so they can have a primary partner to lean on when issues arise.

In fact, the primary reason for Microsoft’s victory in the desktop operating system war was because it offered engineers the most straightforward, integrated software development environment with the largest ecosystem of partners.  Xilinx and Altera are in a similar competitive position.  A behavioral change is needed to design embedded systems and programmable chips, specifically platform FPGAs, offer a different development paradigm.  The Internet and web programming could never have progressed as quickly if websites had to be absolutely bug-free when first released, a requirement for ASICs and ASSPs.

Having IP that is available, certified, and supported is a necessity.  Programmers on the primary enterprise development platforms (such as Java EE and .Net) are offered thousands of tested and supported libraries.  This is a basic requirement of any enterprise development platform.  IP sub-systems (objects) must offer easily understandable interfaces (APIs) that allow the designer to focus on their designs differentiating factor.

Battles for platform supremacy often have winner take all outcomes with all other competitors left as niche players.  The next decade will see further consolidation around very few design platforms.  Similar to a Microsoft (MSFT) or Cisco (CSCO), the platform aspects of Xilinx and Altera’s business model do not require that the companies offer the best of breed technology; although they often do. 

Disclosure: Author holds positions in the above-mentioned securities

Comments

[daniela:]

Thanks for information.
Now I only would wish if I could understand it:-)
I looked at their websites -- to try to understand what they are doing. Both have pretty horrible websites for technology companies.
I don't understand how they are making money by providing platform almost for free?
It would be informative if author would know which one has a better customer/partner service and web site help for downloads and problem solving. Microsoft is still bad at it, so is IBM and SAP. Ease of use and service is key for technology company. It does not matter whether you are engineer or layman.

2008 Jun 20 11:35 AM

[ebndr:]

very good work

2008 Jun 20 02:19 PM

[puerhfan:]

Excellent article. Really good. Perhaps the best I've read in over a decade of owning XLNX shares.

2008 Jun 20 11:12 PM

[Brian Bober:]

Thanks for the kind words. To help clarify, Xilinx and Altera have platforms that their clients develop products on top of and Xilinx and Altera make money by selling chips that go into these end products. While it is true that Microsoft needs to care about the layman’s needs, Xilinx and Altera only need to support engineers due to the nature of their products.

As for whether Xilinx or Altera is in a better position the simple answer would be Xilinx since they have a larger market share and currently appear to have more mindshare. Of course, things are rarely that simple. Altera has better margins and has made some great moves the last couple years such as: partnering with ARM, designing for power management, and having the best path to structured ASICs.

Both Xilinx and Altera have independently discussed that future growth is in adjacent markets not fighting a battle of attrition between each other; such is the benefit of a duopoly. It typically takes two years from design win to volume production and I don’t expect meaningful revenue growth in the PLD industry for about a year and a half. Right now I don’t see a major difference in design wins between the two companies, but that could change.

I think both companies are a great value at the current price and I intend, as best as I can, to track the design wins for these two firms. That will be the leading indicator of who will benefit from the next wave of end product innovation. For the foreseeable future the market can support two major PLD companies and the first losers will be those in adjacent markets.

2008 Jun 22 11:32 PM

[puerhfan:]

Re: design wins. XLNX claims they are dominating design wins at 65nm since ALTR was late to the node. It is interesting that you are not seeing this. ALTR has since announced 40nm products, and claim they will ship engineering samples at the end of 2008. I agree--both companies emphasize the competition between PLD and ASIC and downplay the competition between XLNX and ALTR.

2008 Jun 23 11:00 PM