Lexra, licensees fail in MIPS takeover bid

Lexra, licensees fail in MIPS takeover bid
By Michael Santarini, EE Times
February 3, 2003 (9:47 a.m. EST)
URL: http://www.eetimes.com/story/OEG20030203S0014

SAN MATEO, Calif. — A hostile takeover of microprocessor core vendor MIPS Technologies Inc. has collapsed, and the leader of the fight — rival Lexra Inc. — is going out of business, EE Times has learned. At the same time, two other groups reportedly are still attempting a buyout of MIPS, the No. 2 RISC core vendor.

Sources with first-hand knowledge of the attempted takeover, who spoke on condition of anonymity, said that Lexra led a revolt of MIPS licensees upset at the company's technology direction. However, a recent spike in MIPS' stock value made the takeover price unaffordable, sources said.

The bid, which came together during the last half of 2002 and fell apart several weeks ago, highlights how vulnerable public companies find themselves, given depressed share prices and — in this case — turmoil within a cores business that seems to be coalescing around fewer than a handful of architectures. MIPS chairma n and chief executive officer John Bourgoin rejected repeated requests to talk about the situation but issued a statement. "MIPS is not for sale," he said. "We have strong financials and an expanding license base, which grew by 35 percent in our most recent fiscal year."

Lexra's president and CEO, Charlie Cheng, would not confirm or deny the MIPS takeover attempt. "It is Lexra's policy not to comment on rumors," said Cheng, who did confirm that Lexra is in the process of dissolving its assets. He said Lexra's 35 employees have been told the company is going out of business. Its Web address has already been taken over by an Internet Protocol hosting firm.

The takeover group consisted of Lexra, other MIPS licensees, including Integrated Device Technology Inc. (IDT), one major foundry and financiers. At the behest of a number of licensees, Lexra and a banking group from J.P. Morgan put together a takeover plan and over the last eight months set about signing up investors, sources said. When they began, MIPS' stock was trading at roughly $1.25 per share and the group was looking to raise enough money to offer MIPS and its investors more than $4 per share.

It was imperative to "present this as a friendly buyout," so the group would be able to retain MIPS technologists, one source said. The group apparently had recruited Doug Spreng, the recently retired president and chief operating officer of American Micro Circuits Corp., to head the takeover bid. Spreng was unavailable for comment.

32- vs. 64-bit discord

According to materials supplied to EE Times, the takeover group's plan was fueled by dissatisfaction among licensees of MIPS' 32-bit cores and among early investors, who have seen MIPS' stock price fall and watched the Mountain View, Calif., company lose 32-bit market share to ARM and PowerPC cores. MIPS was too focused on its next-generation 64-bit architecture at the expense of its popular, very competitive 32-bit offering, the dissenters argued.

IDT executives would n't say whether their company was involved in the revolt. "With regard to speculation that MIPS Technologies is involved in merger-and-acquisition activity, IDT does not comment on rumors or potential future activity," said Phil Bourekas, vice president of the internetworking products division at IDT.

One source said that if any licensees are dissatisfied with MIPS, it would likely be those licensing cores rather than the architecture itself. "Architecture licensees have no reason to complain about MIPS," said one licensee, speaking on condition of anonymity. "We license their architecture and do most of the development ourselves."

Sources said the takeover group wanted the new company to function more as a core provider and significantly cut back its list of licensees. That would have helped reinvigorate MIPS as a 32-bit MPU player, they believed. According to a report from industry consultant Andrew Allison, ARM Holdings leads the 32-bit market and MIPS is second, but is being threatened by the Pow erPC for the No. 2 position in RISC MPU shipments.

MIPS officials were made aware of the takeover try in October, and in meetings with the group later outlined terms and a stock purchase price that would be agreeable, sources said. But, at that time, MIPS bosses declined to comment on the takeover plan, and CEO Bourgoin called an EE Times story about it "pretty bizarre."

Late in the year, MIPS' stock essentially doubled, even as share prices for almost all other vendors of silicon intellectual property (IP) declined. Analysts said that at stock prices below $5, percentage fluctuations are common, especially if acquisition rumors are in the air.

From October to December, MIPS' shares rose steadily to roughly $3.50, which made the takeover group's $4/share bid too small for investors to OK a buyout. The group abandoned its efforts in December, said a source.

MIPS' stock price has since declined to $2.70/share, a siren call for new suitors lured by the company's low market cap of roughly $109 million and an MPU architecture patent portfolio potentially worth many times that, sources said. One source said that two other groups — the first a large competitor and the second a group of ex-MIPS executives — are now angling to buy the company. ARM vice president John Rayfield would not confirm or deny that ARM is pursuing an acquisition.

No love lost

For Lexra's Cheng, the attempted takeover apparently was a last-ditch effort to save Lexra's business. When the company was launched in January 1998 as a core vendor offering a 32-bit microprocessor core based on the MIPS architecture, the silicon IP arena was new and many thought it would become a multibillion-dollar standalone industry.

This hype prompted Silicon Graphics Inc. to spin out its MIPS group as a separate entity, which in turn went public in April 1998. The very week of the spinout, SGI and MIPS sued Lexra for copyright violations, claiming false advertising and trademark infringement. Many believed Lex ra could have eventually won that suit, but Lexra's business declined because of it and the economy made it impossible for Lexra to continue to fight in court, Cheng has said.

The case was settled in 2001 under the terms that Lexra sell its core to MIPS and become a MIPS licensee in return for funding from MIPS. While the settlement brought $10 million in cash to Lexra for the short term, it made it unlikely that the company could survive for the long haul offering a standalone MIPS-based chip, in competition with former customers offering discrete 32-bit MIPS products.

MIPS traditionally has been known as a licensor that leverages its key patents. The company went public on the back of a huge licensing deal with NEC, which developed the main engine for the Nintendo 64 game machine with its 64-bit MIPS license. After spinning off from SGI, MIPS began to create its own cores, but to this day it draws the bulk of its revenue from architecture licenses.

What about the workhorse?

Rou ghly a year ago, MIPS decided to step up efforts to develop a high-speed synthesizable 64-bit architecture and cores. Licensees have complained that MIPS focused too much on this market and was letting the workhorse 32-bit market slide. The economic slump has decreased demand for the 64-bit core, which is by most accounts harder to design into a system-on-chip (SoC) than the 32-bit offerings.

Meanwhile, ARM and PowerPC vendor Motorola Inc. have invested heavily in developing and supporting their 32-bit products. ARM is especially diligent about offering design kits that allow users to design cores into SoCs more easily. Rayfield said that ARM now draws 20 percent of its revenue from its development tools business. At the same time, ARM has also stepped up development of higher-speed processors that compete more directly with the fast MIPS offerings.

Rayfield said that the newer ARM 10 and 11 families narrow the gap in performance between ARM and MIPS, while maintaining their low-power edge. Also, he said, MIPS has seen the most success in standalone discrete products, whereas ARM cores are used in standard-cell ASICs where yield and reliability — not blistering speed — are critical.

In October, MIPS closed its Denmark design center, which was developing 64-bit synthesizable cores. The company has also refocused its efforts on the 32-bit market, most recently releasing the Pro Series family of extendible 32-bit cores.