Supreme Court Declines to Hear DRAM Price Fixing Case

By Scott M. Fulton, III, BetaNews

October 2, 2007, 4:18 PM

The US Supreme Court this morning decided to leave it to California courts to decide the fate of what might have been a precedent-setting civil suit against a handful of the world's leading memory manufacturers.

They had already pled guilty in federal proceedings and settled with the government, though Rambus believes it was the memory firm they were conspiring against. Today, in a move that could benefit Rambus in the end, the high court signaled its belief that such a claim is for a lower court to decide.

Back in April 2005, South Korea-based Hynix Semiconductor settled with the US Dept. of Justice, paying $185 million and pleading guilty to price fixing. A wave of settlements followed like toppling dominoes, with Elpida - the joint venture of NEC and Hitachi - in January 2006, and Samsung and Infineon the following May. That certain of their key executives jointly conspired to keep DRAM prices artificially low is a matter of record.

But who were they conspiring against, if anyone? Rambus has long believed itself to be the victim.

As early as May 2004, Rambus launched the initial California court antitrust suit against Hynix, Infineon, Micron, and Siemens which brought the price cartel to light. (Siemens was since fined 418.7 million euros last January by the European Commission for its role, though it is appealing that decision.)

Rambus' public case was made for it by no less than former US Senator John Danforth, the company's senior vice president and general counsel, who said at the time, "From substantial written evidence already in the public record, we believe that these memory manufacturers colluded illegally, thereby limiting consumer choice and depriving our RDRAM products of the opportunity to compete fairly in the marketplace." [CORRECTION: That picture of Rambus' SVP on its executives' bio page looks nothing at all like Sen. Danforth. My thanks to a reader for pointing that out.]

Veterans of the RDRAM era remember Rambus' famous - or infamous - proprietary DRAM architecture. Launched in 1996, it was intended to be vastly superior in performance to standard DRAM. Intel invested heavily in the company, and had plans to devote Pentium 4-based systems to using Rambus' RIMM modules exclusively.

But the expense of those RIMMs even after Intel's investment, in addition to what turned out to be historically inadequate or even poor performance - in the tests that put Tom's Hardware Guide on the map - plus an unfortunate bug in the Intel i820 chipset designed to support RDRAM, are what most believe to have been the factors that truly conspired to defeat Rambus.

Still, the company pressed on. In order to maintain an active assault, Rambus buffered its legal claims with patent infringement suits. Hynix recently sought to have the case resolved by binding arbitration, probably in order to avoid a public recount of its officials' exploits; the California court denied that motion. Hynix appealed, and the State Court of Appeals rejected that too. Its last chance to move the case out of California was the US Supreme Court, but today it said no, thank you.

So even after Hynix and its cohorts' settlement, it may still find itself walking the public gauntlet.