October 12, 1998, Issue: 1030
Section: IP Trends
--------------------------------------------------------------------------------
'Cottage industry' becomes a big-company game
Michael Santarini
Just two years ago, experts predicted that the intellectual property (IP) business would turn into a gold-mine cottage industry. They maintained that anyone, anywhere in the world, with a laptop computer and a slab of design in Verilog or VHDL would be able to quickly become a successful IP provider, go public or get purchased by a larger company.
Even though many small third-party providers of design services jumped head first into IP creation, the IP business today is no safe haven for the little guy. Instead, it is an industry with many barriers to success and sustainability, and it requires sophisticated design and marketing capabilities as well as highly valued IP.
"I think we are clearly moving away from being a cottage industry," said Raj Gollamudi, industry analyst at investment-banking firm Wessels Arnold and Henderson (Minneapolis). "I think we are moving toward an industry structure where you have a small number of players who have good relationships with people and who can offer a complete package."
Unique IP, good market timing, technology that is silicon-proven, legal responsibility and worldwide sales and support are all becoming key elements of a successful IP business. Some vendors lacking one or more of these elements have had to reassess their business models; for others it has apparently led to an agitated exit.
Bernd Braune, senior vice president of Mentor Graphics Corp. (Wilsonville, Ore.), said that the company's IP division is in ongoing talks with 15 to 20 small IP companies seeking to have Mentor source their cores or buy them out of the business altogether.
"The idea that a one- or two-man company with a piece of IP on the Internet can be a success is just not real," he said. "Those outfits may be able to support one customer, but not two or more companies at one time, and customers are terrified about patent infringement. In the beginning, some people tried out cores from some of the garage shops and later in the design cycle they found bugs. The customers didn't have any leverage to go after the little company, work with them or even sue them."
Mentor Graphics' IP division plays in the ever-competitive general-purpose cores arena and has leveraged its size to its advantage. Braune said that Mentor's IP business and its core store business model is thriving, doubling bookings and revenues every quarter.
Mentor's strategy is to provide customers with the opportunity to sign a business and legal agreement up front that covers all of the 75 internally created cores in Mentor's virtual library. Mentor also works with its library customers to obtain cores from third parties.
The company mainly offers telecom, datacom and 8- and 16-bit embedded processors and has attracted large customers like Siemens. Earlier this year, Mentor initiated a program in which it offers all IP creators the chance to create IP to Mentor's standards so that it will work in the company's pipeline.
Size does matter
"In general, what we have figured out is that size matters in this business," said Braune. "Either you are a significant player or you have something wrong with your business model. To make money, you have to have a very good sales, distribution and execution infrastructure."
CoreEl Microsystems Inc. (Fremont, Calif.) is one such small IP company that is currently negotiating with Mentor. Bidyut Parruck, vice president of strategic products at CoreEl, said that in doing so his company hopes to gain larger marketing muscle, but even now the privately held IP creator is by no means a small fry. With 10 people in the United States and 75 more in its R&D and maintenance operation in India, it fields an OC12 Sonet core in addition to offering ATM and gigabit Ethernet cores.
"We are one of the largest independent IP providers in Sonet, ATM and Ethernet, but we can't reach the European market like Mentor can," said Parruck.
Parruck said that CoreEl was bidding for a contract to license its core to 3Soft and lost to IBM, which offers the Sonet core free. "That's the kind of stuff that happens-it's not possible for a small company to give that away for free."
Parruck also said that increasingly, CoreEl has to bid against companies that bundle cores with something else like tools or foundry services. "What we do is form partnerships with semiconductor companies and people like Mentor; that allows a wider distribution of our cores," he said.
Although smaller companies in the market are struggling to nab a piece of the IP market, companies like Sand Microelectronics Inc. (San Jose, Calif.) and the Virtual Chips group of Phoenix Technologies (San Jose) are apparently thriving despite an onslaught of competition in the standards-core arena, and reports that competition has made PCI, USB and 1394 cores into commodities.
Both companies pointed out that although they are seeing more organizations enter the competition, their early entry into the market has given them an advantage and has allowed them to develop a reputation for reliability.
"In this space the value to the customer is the product you can bring to them," said Robert Nalesnic, vice president of marketing at Sand. "What kind of features and capabilities does it have and what kind of reputation does the vendor have? Can they get the job done?"
In particular, Sand offers PCI, USB and 1394 cores and packages with customized verification environments. Recently, the company moved into the design-services side of the business to accommodate any additional work by its competitors.
"You are going to find plenty of vendors that have architected a reasonable product, but when they go to do the design they have to throw all their engineers at the design to get it done, and they will not have enough resources to throw at the next design," said Nalesnic. "So it's fairly easy to get one design, where it is difficult to do 10 or 20 designs in parallel. It's a combination of the use of structure you have in place, the level of support you need for the product and types of customers you need in terms of support requirements."
Barry Hoberman, senior director of Virtual Chips products at Phoenix Technologies, said that the company has had to offer a complete solution to be successful in the market. It is a solution, he said, "which comprehends both the silicon IP and system firmware. A small shop doesn't have as many things in the basket to provide that complete solution. From that perspective, it's hard for little guys to come straight in from the service business."
Hoberman also said that success in the space requires some structure of services capability. "Like any other business, there is some inertia created by being a leader or by being first. A newcomer who is small and doesn't really have much in the way of established credibility or proven designs can't give users a warm feeling that [it is] going to be able to help [the users] get their designs done. There are real barriers to small fries going in. If you come in late and you come in weak in terms of commitment, you are not going to succeed in terms of market share."
Highly valued IP
The current market value of the core is increasingly becoming the name of the IP game and is determining the success of companies.
"High value is very important in the IP world," said Jeff Tate, president and chief executive officer of Rambus Inc. (Mountain View, Calif.). "To be high-value means something has to be very important in the system and cannot be created easily," he said. "If anyone with a laptop anywhere in the world can do it then it's probably something that does not have high value and [is] something the customer could create on his own."
Rambus, along with ARM, MIPS Technologies and the DSP Group, are arguably the most successful IP creators in the industry today. They have differentiated their technologies within the cores arena while Rambus stands alone in its own sector of IP, offering a unique memory-interface technology that bridges communications between ICs.
Each company has been around for five or more years and each anticipated a need in the market and then filled it with technologies that could not be easily copied or reverse-engineered by competitors. All are popularly known as Star IP or Differentiated IP companies. The question is, how long will the differentiation last?
Until recently, ARM Ltd. (Cambridge, England) has enjoyed a market that has been fairly free of competition so is less fearful that its products would become commodities. However, over the last year the company has seen indirect competition in the embedded processor business from the rebirth of MIPS and more direct competition in the low-power microprocessor business from companies like Argonaut RISC Cores.
To expand its foothold in the market, ARM has tried to marshal its efforts not behind a single, generic CPU core but behind several different ones, said Pete Magowan, ARM's European vice president.
"In one way, ARM is already a commodity," said Magowan. "We want to proliferate the architecture everywhere and have licensed about 30 semiconductor partners to do that. We will add new partners where we see it expanding market penetration." Magowan said that while ARM's Mips-per-watt performance was vital to its early success, other, complementary, factors help with sales.
"The strategic thrusts are lowering systems costs. There's Mips per watt but also making the architecture the easiest to use-adding peripheral blocks and use of the AMBA bus, for example." AMBA stands for ARM Modular Bus Architecture; it is ARM's on-chip bus, used to allow peripherals to connected ARM cores.
"We're getting more and more involved in applications design," he said. "In software we can provide soft modems [and] Chinese handwriting recognition. All these elements contribute to the added value [of the architecture]."
MIPS Technologies Inc. (Mountain View) plans to protect itself against direct competition in high-performance cores with legal muscle, said John Bourgoin, president and chief executive officer. For example, the company is currently suing core provider Lexra Inc., alleging Lexra made false claims of MIPS I compatibility. Lexra has denied MIPS' allegations.
"You have to have a strong patent portfolio in this business or what you find is that people, without incurring the expense of investment, will simply knock off the things you did that are successful and put it on the marketplace and charge less," Bourgoin said. "If you are going to generate the kinds of funds you need to sustain your investment, you have to be able to get a fair market value, which includes the amortization of the investment. And you have to be able to protect yourself. Otherwise, after you release your first product and everyone copies it, you find you don't have any basis on which to sustain your business."
DSP Group Inc. (Santa Clara, Calif.) so far does not face competition in cores, but faces strict demands of customers, said Gideon Wertheizer, vice president of marketing.
"Customers are starting to compare our core performance with performance of standard products from companies like [Texas Instruments Inc.]," he said. "They are expecting to see 30 percent or 40 percent performance improvements after one or two years." DSP's first core was Pine; two and a half years later came Oak, which boasted a three-times performance improvement.
Meanwhile, Rambus faces somewhat more challenging competition. It is competing against industry memory standards rather than other IP vendors, and must persuade chip makers to replace SDRAM with Rambus. Tate said that thus far it has only scratched its way into 1 percent of the DRAM interface market share.
Although companies in the Star-IP business are safe for now, companies in the general-purpose and standards market have to adapt different business techniques to deal with increased competition and threats of core devaluation.
"It is kind of like in the case of ARM or MIPS," Tate said. "Once people have adopted their software it makes it harder to switch. With things like PCI, SRAM or DRAM core, it is much easier to switch from one vendor to another."
-Additional reporting by Peter Clarke
Copyright ® 1998 CMP Media Inc.
|
