Under the terms of the settlement, Intel cannot enter into agreements or
enforce existing agreements with OEMs that require the OEMs to purchase
chips exclusively from Intel in any geography or market segment, or
requires OEMs to refuse to purchase or delay purchase of products from
Intel competitors. This precludes Intel from offering discounts on
microprocessors or threatening legal action against OEMs for not
purchasing other chips such as graphics processors from Intel.
Intel may continue to offer discounts to customers on parts in excess of a specified unit volume threshold, but the company may not offer a volume discounted price to OEMs on all parts contingent on total sales being above a specified threshold.
The terms of the settlement also specify that Intel may not deny an OEM the ability to market its products as having "Intel Inside," in the event that, for example, the product also includes graphics processors from Nvidia. The settlement also prevents Intel from withholding marketing dollars from OEMs that include competitors' products.
Intel agreed that it will include in each of its mainstream microprocessor platforms an interface to a standard PCI bus. Intel also agreed not to design any required interface for the purpose of intentionally limiting the performance of a graphics processor that would make it non-complaint with the applicable PCIe base specification. The presence of bugs or errata that would make any product non-complaint with the relevant PCIe specification will not except the company from the latter requirement, according to the settlement.
Under the terms of the settlement, Intel must also offer to amend its patent agreements with AMD, Via Technologies Inc. and Nvidia that would allow them to disclose to customers and foundry suppliers its licensed rights to Intel patents, provided that the customer or foundry agrees to keep the information confidential. The settlement requires Intel to offer other specified amendments to patent agreements with Via.
McCarron of Mercury Research (Bee Caves, Ariz.) said that ironically it was probably Via, by far the smallest of the Intel competitors named in the investigation, that got the most out of the settlement deal. "They have x86 license, and that gets expanded [in the settlement] at no cost to them," McCarron said.
The real problem with ‘settlements’ like these is they legitimize unacceptable behaviour and that strikes at the very heart and soul of society. Whatever the grand posturing, this is an unequivocal whitewash with self-congratulations all round for the regulators who, in reality, have failed to protect the market. The real world knows this of course so all these 'slap on the wrist’ judgements do is to undermine regulatory authority. This is yet another bad day for capitalism and a sad day for America. It’s a sad day too when firms have to resort to cheating and dirty tricks to stay on top of the game. Regulators are there to rein back these excesses, not condone and legitimise them.
Intel has started a pattern that questions its ethical behavior in the market. When I was approached to join this great technical company, I was turned off because for all its microprocessor, Intel does not come our 'normal' to me. It fires staff anyone and bullies in the market. That is not the mark of a leader. They better learn how to act in this world
I'd love to hear more comments from folks like BobbyTsai on whether access to PCI interfaces is enough to open the door for third-party graphics--or do they need access to cache coherent interconnects like Intel's QPI.
Given the speed of technology changes I would expect that in 10 years all of the chips/technology would not be relevant anymore. What is relevant is the settling of the the charges and the fair play ruling for Intel and non-Intel devices. I am a big fan of competition and using newly developed technologies to create market differentiations but blocking out competitors while good for self interest does not help consumers or technology advancements.
@CamilleK- agreed about the $2 million. It's hard to imagine that will cover the salaries of multiple technial people over 10 years. Even though the order is supposed to be in effect for 10 years, I get the feeling the FTC doesn't it expect to to be necessarily relevant for that long.
The FTC should also not have agreed to their technical consultants being approved by Intel even if Intel paid for them. The $2 million limit over 10 years for the cost of those consultants seems on the low side. It would have been better to have a team of independent consultants under NDA. Alternatively if the consultants would not be acceptable, a reason should be given with an override by FTC a-la 'advise and possible no consent'.
The agreement needs to cover QPI. It is needed to connect to IO hubs and interconnect CPUs in multi-processor systems. (can also be used for graphics processor and other co-processors). Guys @ FTC missed the boat on this one. Sure DMI and PCIe can be used but the high margin products are not included. AMD and Nvidia product were always 8-12 months ahead of Intel products before QPI. Without QPI, Nvidia / AMD / VIA / SIS can't provide a full market portfolio.
Intel should have been fined a billion or so dollars, but more importantly the FTC should have pursued criminal charges as well. Why should "big oil" receive the brunt of media hatred while criminal monopolies like Intel and Microsoft escape prosecution? I need oil and gasoline, but I'm beginning to see that chips from Intel and bloatware from Microsoft aren't worth much. I say take the executives to court on criminal charges, and put them in shackles and orange jumpsuits to and from court, then put them in prison for long sentences. That should cut down on criminal behavior in the tech sector.
The fact Intel must offer a PCI interface on its processors should open the door for Nvidia to sell graphics and chip sets for all Intel parts without needing to license Intel's new QuickPath Interconnect, an agreement Nvidia had not been able to secure. That's big in itself.
Join our online Radio Show on Friday 11th July starting at 2:00pm Eastern, when EETimes editor of all things fun and interesting, Max Maxfield, and embedded systems expert, Jack Ganssle, will debate as to just what is, and is not, and embedded system.