SANTA CLARA, Calif. Intellectual property cores have real value, but determining that value is no simple matter and may involve a host of business models, according to a panel of industry experts who debated the topic Tuesday (March 23) at IP99.
Only a handful of "star IP" companies have managed to translate the value of their cores into real profits to date, the panel agreed, but its members were divided over how broad or narrow a group of players would ultimately make a real business out of IP trading.
"So far we've seen two markets for system-on-chip emerge digital GSM handsets and videogame consoles," said Erach Desai, vice president of equity research in the technology group at Credit Suisse First Boston, who chaired the panel. "There will be a lot more drivers like digital set-top boxes and non-traditional computing devices. We are just at the beginning of that growth."
Desai split the market for IP into three tiers 151; star IP, such as high end microprocessor or memory cores that could command per-use royalties; standard IP, such as USB, 1394 and Dolby AC-3 decoder cores; and commodity IP or generic libraries.
Only star IP has "transactional value," said Paul Lippe, senior vice president for business and market development at Synopsys Inc. (Mountain View, Calif.), which splits the IP market up into roughly similar thirds. "A substantial amount of commodity IP will be a loss leader for the services business. The likelihood that anyone will pay royalties for IP in perpetuity as they do for Microsoft Windows is exceptionally low," Lippe said.
Lloyd Nirenberg, chief executive officer of IP Valuation Inc. (Palo Alto), which hopes to carve a business out of helping to negotiate IP deals, took a decidedly different view. "It turns out there are an enormous number of pricing models that are viable," Nirenberg said.
Nirenberg showed a chart detailing at least 20 conpensation models for a given piece of intellectual property. His company is attempting to break new ground in developing models that include charging a licensee for the right to market a core and for the extent a core is reused. "The value of IP depends on the amount of incremental profit the licensee can get for it," Nirenberg said. "If you don't reuse it, it's a widget."
Various models might generate the same net amount of money but articulate widely different risk profiles, he said. "It's all a question of how you want to balance the risks, and we have software to help you determine that," he said.
Indeed the value of IP is often relative to what the licensee can do with it, said Reynette Au, vice president of marketing for ARM Inc. (Los Gatos, Calif.). A small fabless company that uses an ARM core in one or two designs might pay a much lower rate for the same core that a large conglomerate such as Toshiba might use across a vastly broader range of designs, she suggested.
And the value of IP goes up and down over time, based on a range of fac tors, she added. "In the very early days, ARM couldn't give away its architecture we tried.," said Au. "As time went on, third parties developed new tools and the use of the chip began to proliferate, and the value of the IP increased."
The panel repeatedly cited ARM, MIPS Technologies and Rambus as the market's current "star IP" companies that have carved out a business in intellectual property. Desai and others said the common thread to their success was years of market development and investments from parent and partner companies, a fact that formed a warning for new entrants to the nascent IP market.
"My hats off to ARM, MIPS and Rambus, but that doesn't mean it's an 18-month product cycle for anyone else," said Lippe.
Nevertheless, the Synposys manager held out hopes that some startups could still carve out a niche here. He specifically cited potential in 3-D graphics, communications and low-power embedded systems. But he added one note of caution for such startups. "They need signif icant technology and market insight to get out of the way of existing players, and there is not a lot of venture-capital money chasing these deals."
In addition, there may be room for more memory core players, he suggested. "So far the Japanese companies haven't been able to leverage their expertise in this area, so there is potential for new entrants," Lippe said.
Desai noted structural road blocks that new players may also face. "Wall Street loves a simple model we can extrapolate and build upon," he said. "IP is a sexy model, but there are problems in [a lack of IP] tools, models and design services we need to go with it."
Panelists largely agreed that the model of "free IP," which has been tied in some cases to a services business, would not stand the test of time. "Cadence put a stake in the ground that all IP would be free with design services," Desai said. (Cadence did not have an executive attending the panel session.)
"Why would you want to put yourself in a labor market?" asked Nire nberg.
"Some people think the future of the industry is in design services, but what that has to do with design reuse, I don't know," added Lippe.
"There is no free lunch," said Michael Moursi, director of the IP division of Mentor Graphics Corp. "Free IP is not a business proposition, and at the end of the day the user must pay and the designer needs a return."
"A statement like 'IP for free' is scary," added Marco Carilli, manager of corporate IP reuse strategy for STMicroelectronics. Simple, common sense metrics like quality and time-to-market will dictate the value of cores, he said. "People will buy IP to buy time, but that value will be diminshed if it takes time to rework that IP for their designs."