Nine Chip Makers Fined $400M In EU For Price Fixing

eldavojohn writes "In a disturbing case for average consumers, nine DRAM chip manufacturers have been fined more than $400 million for price fixing. The named companies are Samsung, Hynix, Infineon, NEC, Hitachi, Mitsubishi, Toshiba, Elpida, and Nanya. A tenth company, Micron, avoided fines by reporting the other nine to the authorities. Since all companies cooperated with the probe, they received a 10% reduction in fines, so it could have been worse. The US DoJ has had its own history with chip makers and LCD makers in price fixing scandals."

Not really that disturbing

[DavidRawling]

There are many other cases where pricing appears to be fixed, but it's a deliberate lack of competition (eg in Australia, the weekly fuel price cycles where everyone drops prices at the same time). At least this occurrence will be punished, and yes it will eventually come from the consumer wallet ... but I don't see much else that can be done other than fining (and imprisoning the human culprits if possible).

ban one company at random

[Anonymous Coward]

I say rather than fines, we ban one of those companies from the US market forever. We repeat this process ever time there is price fixing incident. Shareholders of those companies will not tolerate the risk and management will be too scared to pull this shit again.

Re:Disturbing?

[purpledinoz]

Is this the same price fixing that occurred when Intel tried to shove Rambus, a crappy, expensive, proprietary RAM technology down our throats? And they colluded to LOWER their prices to kill Rambus? In my opinion, they did us all a favour. If Rambus became the "standard", we would be paying a lot more money for memory now.

Re:Disturbing?

[purpledinoz]

Last time I checked, all companies except Samsung are losing money on each DRAM chip they sell. These companies are competing themselves to death. Qimonda already went bankrupt last year because of such a competitive environment. In fact, I can't think of many other products where companies compete so hard to make.

Re:So what?

[Anonymous Coward]

yep. look what one of the fuckers had to say: [elpida.com]

"Since the fine is within the range of a reserve already fixed for such issue in FY 2008, the company believes that the fine will not have a material impact on its current year (FY 2010) consolidated financial results."

"for such issue" - gotta love the Japanese.

Not even close to reality are you.

[Chris Burke]

You are correct, because DDR uses the same patented techniques that RDRAM did, it was just as fast.

What? Are you insane? Same techniques? Just as fast?!

The only thing DDR and RDRAM have is that they transfer data on both edges of the clock signal like a thousand other technologies that already existed at the time, on chips and on PCBs. Rambus did not invent double pumping data busses; it was already a standard technique for reducing signal integrity issues on the clock signal. See a variety of FSBs that were out at the time.

Other than that, the two memories couldn't be more different. RDRAM is a serial, in-line, packet-based, wave-pipelined memory technology with differential signaling and resistor termination. DDR is a parallel, multi-drop, bus-mastering, logically pipelined, non-terminated and non-differential technology.

They two memories are nothing alike except in the most superficial way, and unsurprisingly to those with a clue have completely different performance characteristics. RDRAM had higher clock rates and higher sustained memory bandwidth despite having a much narrower interface, but had substantially higher latency especially as you increased capacity. DDR on the other had much better latency, but due to the multi-drop nature of the bus had issues with increasing the clock speed and so was bandwidth limited. Nevertheless, DDR outperformed RDRAM in most applications, even in architectures like the Pentium 4 which were designed around high latency and high bandwidth.

The PS3's use of RDRAM that you mention in another post is an interesting case. RDRAM works here because it gets relatively high performance for a small number of motherboard traces, and fewer traces translates into a cheaper motherboard, which is a good tradeoff for a home console.

Ask yourself why if these companies colluded to fix prices as they have admitted to doing why did they decide to fix prices so low?

That's obvious. They did it to ensure that despite the huge push by Intel, RDRAM would fail to become the dominant memory type. It wasn't enough that DDR performed better, it had to cost customers less otherwise they'd just go along with what Intel said. They did this to avoid having to pay Rambus royalties on RDRAM.

If you've actually been following the case, then you know that the patented technologies the DRAM makers have been accused of violating are not technologies implemented in RDRAM. They are patents that specifically cover DDR (and not RDRAM) filed by Rambus and specifically modified while JEDEC meetings were going on to cover what JEDEC was discussing for their new, supposedly patent-free memory standard. Rambus did not disclose these patents despite the JEDEC rules requiring it. Rambus deliberately created a patent minefield in the DDR spec that had nothing to do with RDRAM.

The courts have repeatedly found that this is what happened, that Rambus violated the JEDEC rules and operated in bad faith. The only question has been whether the JEDEC rules are legally binding, and whether their actions constituted legally actionable "bad faith", which the appeals court ruled they did not.

The DRAM makers price fixing, however, is certainly actionable, and I'm certainly glad they were fined for it. They're giant assholes too. But they did not steal Rambus technology. Rambus stole from JEDEC and manipulated the system so they came out with ownership of the technology that had nothing to do with what they had actually invented.