29 August 2007

Ambushed by Patents

I've not written about the Rambus case before, because it seemed frankly rather dull. But I was wrong: there is an important principle at its heart:

European Union regulators have charged Rambus Inc. with antitrust abuse, alleging the memory chip designer demanded ''unreasonable'' royalties for its patents that were fraudulently set as industry standards.

The EU's preliminary charges, announced Thursday, come weeks after the U.S. Federal Trade Commission ruled the company deceived a standards-setting committee by failing to disclose that its patented technology would be needed to comply with the standard.

As a result, every manufacturer that wanted to make synchronous dynamic access memory chips had to negotiate a license with Rambus.

Both EU and U.S. antitrust officials allege that this allowed Rambus gain an illegal monopoly in the 1990s for DRAM chips used in personal computers, servers, printers, personal digital assistants and other electronics.

Clearly these kinds of patent ambushes are potentially a general problem, and indicate why real standards must only allow completely patent-free technologies. If a company wants its patented technology to become a standard, it must give its patents.

2 comments:

Anonymous said...

Funny thing is that Rambus didn’t have any “relevant patents” to disclose when SDRAM/DDR standards were discussed. They have invented fast memory in 1990 and they’ve been teaching memory manufacturers how to make it, under NDAs. They then have been invited to join JEDEC because of their expertise, seemingly. They have left a few years later after realizing that their inventions were being hijacked and incorporated into the standards, by the very memory producers they’ve been teaching. Everybody knew what Rambus invented but some didn’t want to pay royalties. So they pretended not to remember where the inventions came from and accused Rambus of deceiving a standard-setting organization. And called in the government. Isn’t that a wonderful way to strip somebody of patent rights? It’s cheap too. That may have been the real reason behind JEDEC invitation.

Also funny is the fact that JEDEC continues to openly incorporate additional Rambus inventions into its new standards, DDR2, DDR3, GDDRx, etc. A decade after Rambus left. Nobody was able to come up with anything better, so they just take more of what works. US FTC only limited royalties on the old SDRAM/DDR standards. The mainstream has long moved to DDR2. Of course, big memory companies wanted free unlimited use of all Rambus inventions. They didn’t get it. The little they did get is under appeal. What will the EU Commissioners be left pointing to if FTC is overturned by a real court?

Indeed, there was a patent ambush. Only it was the other way around. Certain memory manufacturers ambushed Rambus in attempt to strangle it, so they could use its IP for free. Although unfortunately the “savings” didn’t reach the consumer, as these companies were convicted by DOJ of memory price fixing. So for a long time consumers paid inflated prices for stolen, stripped down designs, without even knowing it. The memory in your computer could have been 10 times faster, just look at Rambus’s XDR memory design. If there were no Rambus, it would quite possibly be 10 times slower. Is 1-4% royalty for inventions facilitating 100x faster memory - “unreasonable”?

http://www.rambus.org/story/

Glyn Moody said...

Thanks for the comment and the link. Another reason I haven't written about this case is that I can't claim to know the full background or who exactly did what to whom. So I will reserve judgement on Rambus's case here until a later date when the facts are settled - if they ever are.

However, irrespective of this particular case, the threat of patent ambushes is still there, and if this isn't a good example, I'm sure that we'll get one all-too soon.