SAN JOSE, Calif. A distinguished panel of semiconductor and IP executives predicted on Monday (March 22) that the licensing of intellectual property and collection of royalties on its use is the only profitable IP business model, but that it only shows positive results over time.
While members of the IP99 panel differed in their views on the best business model for a provider of IP, they agreed that being such a provider today is not for the meek. "We've been licensing the ARM architecture for some eight years and we have seen return on the licensing and royalty fees only in the last two," said Timothy O'Donnell, president of ARM Inc. (Cambridge, United Kingdom).
John Bourgoin, chairman and chief executive officer of MIPS Technologies Inc. (Mountain View, Calif.), concurred. "We've been at it for a while and I can tell you three elements define the IP business model: volu me business; value-add; and protecting your assets," he said. "If you can't do all three, you shouldn't be in the IP business." MIPS gained an edge by releasing the instruction set of the MIPS architecture, Bourgoin said. "We have 20 design teams from our partners who add value to the MIPS architecture by making changes to the set, which is then available to all."
Paul Nahi, general manager of the Segment Engineering Division of NEC Corp. (Tokyo), said NEC runs its IP business as a customer service organization. "Outsourcing of IP services will be growing in the future, because the success of R&D in IP is inconsistent," Nahi said. The software and system expertise dedicated to those efforts will be sorely needed in the next few years, and is sorely lacking at this point, he said. Volume production is the only way to recoup on IP investments, said Nahi, who cited one current NEC design with a production rate of 400,000 units/month.
Longtime EDA industry luminary K.C. M urphy, now president and chief executive officer of Pivotal Technologies Corp. (Pasadena, Calif.), agreed that any successful IP provider requires a licensing stream, the volume proliferation of an end product and a service strategy. Noting four basic IP modules that a provider can chase control; logic and memory; I/O; and analog/mixed-signal Murphy said Pivotal will attempt to carve out IP cores from a $25 billion total available analog market. "We know analog/mixed-signal IP is a challenge, but we expect to use a standard 0.25-micron CMOS process and on-chip test capability and go for the volume market," he said. "I firmly believe that, while anybody can carve out a Ferrari, we need to aim for a fleet of Ford Tauruses."
The fifth panelist, Jim Ballingall, vice president of worldwide marketing for UMC Group (Hsinchu, Taiwan), preached a mantra of partnerships with outstanding IP and design service providers. " A foundry should be thought of as really being a marketing channel for IP providers to reach hundreds of customers," he said. To allow customers a way of dealing directly with a foundry, UMC recently began to offer free 0.25-micron standard cell libraries to its customers; the libraries are distributed and supported by third-party library providers. Prodded by panel moderator Robert Chaplinsky, general partner at Mohr, Davidow Ventures (Menlo Park, Calif.) to distinguish between libraries and "real" IP, Ballingall said that IP is application-sensitive, while libraries address generic designs.
The dinner panel culminated the EE Times-sponsored IP Business Forum day at IP99, a three-day technology business conference now in its third year and schedule to run here through March 24.