EU fines Intel $400M for blocking AMD's market access
neowin.net
neowin.net
It’s shocking that Acer, HP, and Lenovo all accepted incentives (money) to monopolise the market, and are just as guilty in my opinion.
This seems to be black and white, obviously inappropriate; yet has taken decades and still unresolved.
Really…
Personally, I think that all the stuff like this should also be illegal, even without a monopoly. Exclusivity contracts and stuff with that goal should be illegal.
However, we'll have to pass new laws to get there.
If Intel comes to my computer company and offers me a sweet rebate, what do I know about what they're doing with my competitors?
So if it were illegal to receive money to help a company monopolize their market, how would the company receiving the funds be able to determine that's the case?
If I make wheat and you run a bakery I might offer you a 30% discount for the next 5 years, but only if I'm your only wheat supplier.
This sort of arrangement allows us to spread the risk of market uncertainties over that period.
I can then build extra capacity, and you get a cheaper product, or one with less price uncertainty.
Maybe we should do away with this, and say that bakeries must buy wheat like airlines buy fuel, i.e. through more complex financial instruments.
Now, I may be doing this because I'm one of only two companies making wheat, and I'm looking to drive the other one out of business. That's basically what Intel was doing here.
But it's not clear to me that we should conclude that exclusivity deals in general are bad.
Say I'm a local wheat producer. I sign a deal with the local bakery. That stops a multi-national from undercutting me.
The contract cements the benefit that both I and the bakery have from mutually working together, and removes the risk to both parties of some outside party damaging both of us.
In other words exclusivity works for small companies in spaces dominated by behemoths.
Think craft beer and local pub.
Even if you were to get there first, when the contract comes up for renewal the multi-national can offer terms that would be unsustainable for a small producer.
I would be bummed if my local tap roomed signed an exclusivity contract with a local brewery. These days there often many local breweries and new ones open regularly. It would be a shame for the local taproom to miss out on offering a wider range of beers because they got locked into an exclusivity contract.
If you are providing a discount or kickback or whatever for exclusivity, that is an anticompetitive behavioral, regardless of size.
Generally the person who is bribing has higher penalty, because she is an instigator, but both are culpable.
It makes sense, by accepting the bribe, you do damage to some third party. "I was stealing, but only because he offered me money"... nope.
At least in my country and I assume in the rest of EU countries.
Not sure about that.
"I will only stock your product if you can make the deal sweeter for me than your competitor is offering to do."
Bribes are paid by (or directed to be paid to someone else by) the person who is buying (or a sales representative selling) and their employer never sees the money. That's why direct incentives to salespeople (spiffs) should be authorized in writing, by the employer. An example of a bribe would be a purchasing agent requiring me to donate 5% of the deal value to their church to get a deal.
In fact in 2007 Dell would have made a loss if it weren't for the Intel kickbacks.
It goes down to "are exclusive deals illegal ?" and I'd assume it could go as far as small companies accepting to get bought by the market behemoth, or employees getting paid under no-compete contracts to not work at rival companies.
If Intel gonna say "we literally will pay you more to use our parts", that's pure profit to them, customers still gonna buy it because if they cared more about the CPU than the rest of the value propositon of Dell/Lenovo they wouldn't buy from them in the first place.
Try this: find me an in-stock 2-in-1 Ryzen 7 or 9 metal case laptop with high nits 2k+ touchscreen, no discrete GPU, great backlit keyboard, a large battery, and a spare M.2 slot.
Or try to find anything bizarre or exotic (like those laptop with 2 screens instead of 1 + keyboard). Nothing with AMD, ever.
Almost all are / were easy to find in Intel flavor. Have fun trying to put your hands on any worthwile AMD product.
This could have saved me having to carry a bulky charger. Next time!
Here's something exotic soon to be released with AMD:
Really making me question my choices here. Should have just bought her a Lenovo.
You could always cancel your pre-order, get your deposit refunded, and just buy the Lenovo. You could also get a 13th gen Intel Framework 13 right now.
I've pre-ordered a Framework 13 AMD. I'm not as worried about it because I only placed the order recently and I'm across all of the issues.
Not really. They've shipped three iterations of their Intel laptop (11th, 12th, and 13th gen Intels). The Framework 13 AMD and Framework 16 AMD laptops are their first AMD products.
The Framework 13 AMD started production a couple of weeks ago:
https://frame.work/blog/now-available-in-taiwan
They'll start shipping them soon.
> Their shipping countries list is small.
Framework is still a small company. They'll add more countries as they grow.
The brief was "exotic". I delivered exotic.
This is about the EU, it's no surprise that Intel continue to be monopolistic and abuse the market in a market that makes no attempts to control abusive companies.
I ended up just replacing the battery on my old laptop in the meantime. TBH the only reason to upgrade would be the screen
[1] during the covid era; here's one example from Feb '21: https://www.youtube.com/watch?v=KuoHW5EiJdQ
Yes, some vendors tried to put a good CPU in a budget chassi, worst screen you could find, no battery life, terrible build quality etc. No shit it didn't sell.
And not sure what "bad experiences" here amount to, has nothing to do with what CPU was inside but most probably the brand of the device. And that might partly because the vendor had no experience in that ecosystem and had to do much from scratch or whatever.
But most likely they just didn't put in the effort (if the poor build quality and components didn't already tell you that), which kind of makes sense given the weird situation we are in.
How much of that can be attributed to intel, and how much can be attributed to vendors having absolutely no vision and no desire to make anything but literal safe crap, I have no idea.
But, even if intel didn't do anything wrong I wouldn't be surprised with the outcome. It is not like the ecosystem for intel hardware is great either. It took decades before microsoft gave up and launched their own Surface-line of computers simply because all the combined efforts of all manufacturers could barely produce anything of value.
Just as google had to launch their own Pixel line because all other vendors would just produce crap that people then associated with android.
As for problems it’s usually reliability issues with power management, network and occasional crashes. And that’s from big brand integrators like Lenovo. You just don’t get the same poor experience with their Intel parts.
I mean when we bought a huge chunk of T14s recently, an actual factor in the decision analysis was “not the AMD variety”.
It might be hardware, it might be software but no one has the time or energy to dig through that.
Anecdotally I built a Ryzen 3700X workstation from scratch a couple of years ago. Nothing but trouble. Some days it wouldn’t even boot until the third attempt. And that gets to decision makers and they make decisions.
And when you get a bad batch "from Intel" you don't exactly ditch Intel do you? No. Ditch Lenovo if they can't make computers.
You are completely missing the point.
My biggest complaint is that it's even harder (or possibly more confusing) to work out how to get an AMD running ECC RAM than it is for Intel, and that's saying something.
AMD or whatever crap Dell put in to keep AMD the cheapest crappiest option, perhaps contingent on “marketing funds” from Intel?
If you build it yourself you can get a fast stable AMD desktop. Prebuilts tend to be shitty and I wonder why…
It was eventually OK for Windows users but for Linux users you had to settle for a cheap wifi chipset in any AMD laptop. I finally managed to at least find a mediatek chipset instead of qualcomm, but I really would like an intel wifi chipset with a Ryzen CPU.
And don't even get me started on high resolution, or touchscreen monitors.
Unless you want to run an access point of course, I have no idea abou that.
Most people have no idea what AMD or Intel — let alone a CPU — is. It seems plausible that AMD isn’t popular because the average consumer isn’t demanding it.
But, more people know the name Intel from those Intel stickers on those old laptops people used back in the day.
So because those worked fine for them, they just choose to ask for the same name, making Intel over AMD.
I'm sure there are more factors than just these that make Intel the most popular CPU brand, but this is the best one I could think.
Manufacturers were not able to design equivalent servers due to pressure from Intel, or need to delay equivalent or more powerful AMD platforms.
AMD is still has more PCIe lanes and better for high bandwidth requiring tasks like dense GPU computing.
The examples cited in the article:
>Specifically, the Commission cited examples where Intel paid HP not to sell AMD-powered business PCs to small and medium businesses through direct channels from 2002-2005. It also paid Acer to delay the launch of an AMD-based notebook from late 2003 to early 2004. Intel also paid Lenovo to push back the launch of AMD notebooks by six months.
For decades, AMD/ATI have been sold to me as "like Intel/nVidia, but cheaper". I've fallen for that trick several times and I have always regretted my purchase.
AMD will never get any money from me ever again. Perhaps I'm not the only one.
GPUs: terrible, awful drivers and compatibility.
I say this as an AMD fan who is very happy with my 5900X/6900XT system.
The exact reason why I stuck with Nvidia since buying a 1080. I got it for the perf, but I stayed with Nvidia for the drivers.
I don't really see the issue, there are Ryzen think pads I could order right now.
> Try this: find me an in-stock 2-in-1 Ryzen 7 or 9 metal case laptop with high nits 2k+ touchscreen, no discrete GPU, great backlit keyboard, a large battery, and a spare M.2 slot.
Maybe it just that your tastes are not that common.
cough Acer Swift X
The fact that such moves might make financial sense says something about our underlying financial system.
One with patents.
I presume that if we didn't have patents for processor architectures, competition would have made these practices completely unviable, and we wouldn't have had a decades-long duopoly in the CPU market.
Reverse engineering the x86 instruction set would be rather trivial for even a small company. Releasing a x86 product, however, is not possible because of patents and Intel lawyers.
I think we should swap the burden of proof: patents should be abolished if it cannot be shown that they’re a net positive for consumers. We shouldn’t be supporting a practice whose effect is unknown and possibly detrimental to consumers.
In the 90s you had other companies that could have built x86 chips, if they would have been allowed two. Digital Alpha was the bases for the AMD chips that kicked Intel so hard that they resorted paying their supplier to not adopt it. Digital could have done a x86 but they would have been in a lawsuit with intel.
Intel also didn't really have an advantage on the node for most of that period.
https://ec.europa.eu/commission/presscorner/detail/en/qanda_...
"Antitrust: Commission re-imposes €376.36 million fine on Intel for anticompetitive practices in the market for computer chips"
https://ec.europa.eu/commission/presscorner/detail/en/IP_23_...
Like fish in water we dont even know what a truly competitive information technology sector would look like.
My rough guess is that we would be two decades ahead in the quality and variety of hardware and software (a 2 x acceleration vs the stagnation of locked-in artificial equilibria).
Information technology is still a huge phase space of possibilities that lie unexplored so that a tiny set of vested interests can have a predictable monetary return.
Even into the early '90s, I don't think that was really the case.
> My rough guess is that we would be two decades ahead in the quality and variety of hardware and software
Literally based on what? You think Moores law would be 2x faster? Why? Intel never had a monopoly outside of maybe in 2000s for a few years.
I am not conducting a legalistic argument. Name a period where digital tech was not controlled by a cozy arrangement of one or two entities, offering all the menu options of Cola vs Pepsi.
> Literally based on what?
It is hard to make it an "exact science" when we don't have actual long periods of competition driven innovation to juxtapose. But there are plenty of hints. Take for example the story of RISC. The architecture is only now being "rediscovered" after being crushed for decades.
Do you know what the landscape was like in the 80s?
> Name a period where digital tech was not controlled by a cozy arrangement of one or two entities
Ok, lets talk about the 80s. I could do this for other decades.
Monopoly in Operating Systems? No clearly not, IBM had its own (more then 1), various Unix, VMS, CSM, various DOS, lots of costume proprietary stuff.
Monopoly in machine architecture? Clearly not, you had IBM mainframe and midrange, you had DEC Vax (and PDP11 still around). DEC had many competitors, most failed in the early 80s. But were replaced by Motorola based workstations and servers. HP was a big player in the 80s. You had Apple doing its own thing. PC based systems were on the rise, including being used as servers. You also still had a huge amount of 8-bit micros being sold as well. I don't see a monopoly.
Lets talk about processors fabs. Lots more companies had fabs back then. IBM, DEC, Intel, AMD, Toshiba, NEC, Motorola, National Semiconductor, HP and even Commodore had their own fabs and you had contract fabs like VLSI. Monopoly?
And there was so much more strange stuff happening in Britain, Germany and so on.
So please, actual tell me what monopoly you are talking about instead vaguely referencing abstract concepts!
> Take for example the story of RISC. The architecture is only now being "rediscovered" after being crushed for decades.
This is just complete nonsense. First of all the RISC/CISC division isn't as relevant today as it was then, and the idea that 'RISC' was lost is complete nonsense.
Sun/Oracle/Toshiba and co build new SPARC to this day, Tosbia mainframes have been running on SPARC this whole time. Lots of IBM stuff runs on POWER to this day as well. Even in the 2010s SGI MIPS based clusers were common. And during this whole time MIPS and RISC basically owned the embedded markets between them. So its really only on the PC and server PC where x86 was dominate. Turns out that's not all of computing.
Please do this for the 60's and 70's and IBM's 70% market share that actually eventually led formally to an anti-trust case against the company [1].
You seem to conflating the availability of occasional and niche alternatives with an actual competitive market concerning products with major mainstream adoption. In a real competitive market you would have like five to ten providers for substantially the same product or close substitutes. This kind of market is when buyers benefit and this is when producers are forced to compete on price or features. Of course producers hate competition. Its less profit for the same work and a constant threat of extinction if you don't deliver.
Various arguments have been advanced (by vested interests, most notably Microsoft trying turn black into white during its own anti-trust case in the 90's) to justify ex-post the pervasive tech exceptionalism in matters of market concentration and associated regulation: "network effects" and innate "winner-takes-all" in tech being the most prominent.
> its really only on the PC and server PC where x86 was dominate. Turns out that's not all of computing
Its a very strange view: If an entity is not dominating 100% all related markets its not a monopoly. Monopolists would love to have you as jury member in an anti-trust trial.
And yes I know about IBM, I have heard of them before.
> Its a very strange view: If an entity is not dominating 100% all related markets its not a monopoly.
Where to draw lines in what is considered a market and a monopoly is always questionable. But its besides the points, because your statement was not, monopolies sometimes exist in the broader computer industry. Your statement was something inherent in 'Digital tech' was somehow extremely prone to monopolies. And I don't think that view holds water. Different market segments had different periods of consolidation and centralization.
Yes, IBM and Microsoft were sued in anti-trust cases, but in both cases they didn't lost their dominate positions because of those anti-trust cases.
Now we're "fining" them a few million while pumping billions because we "need fabs in Europe". We do, but not theirs
It's entirely wrong to give money to intel, whoever you are. They are the devil and have treated us horrible. These guys deserve nothing but contempt
But the important part is that Intel stopped doing those things after the investigation started so that AMD could compete. If they didn't it would be very easy to prove that they are anti competitive. So even though the fines are late the effects have already been here for a long time with AMD doing well today.
A lot of online apps tend to operate at a loss to gain a large audience and then once they've gained market share they often increase fees or more generally make it more expensive for users. Their business model forces smaller competitors out because a larger funded company can in theory hold out for a longer period of time at a loss.
It's surprising this is ok but selling a piece of hardware at a loss is not ok. What about Costco selling rotisserie chickens at a loss to get people in the store or the idea of a "loss leader"?
> What about Costco selling rotisserie chickens at a loss to get people in the store or the idea of a "loss leader"?
Isn't the usual argument that they got surplus stock in some way? Then it isn't below cost, it is loss minimization from a bad investment, and it doesn't hurt competition for chicken since it is a limited amount of time.
Also, it isn't limited amount of time. This has been Costco's MO for decades.
Have they sold chicken at below cost for decades with no stop? Really? I think you mean they have sold small portions of their wares at below cost, not that they have sold a specific thing at below cost for extended amounts of time.
> Whether you have 100 chickens or 100 million, if they cost you $6 and you sell them for $5, that's below cost.
Yeah, but the store as a whole isn't selling at below cost, and on average since those chickens aren't sold at below cost all the time they still make money from chickens. Those things means that on average the store is net positive, and them selling chickens is a net positive. Hence overall they don't sell anything at a net loss, its just for short periods of time they do it and those can be hand waved away. There are a lot of laws surrounding limited time offers that they have to follow.
Intel stock is a lot of things, but 'up 80%' hasn't been something true since the early 2000s.
ETA: That was 16 years ago I was thinking of, I forgot about the nasty dip in 2008. I misremembered, it was ~26 when I was hired, dipped to 18 the next year, and slowly climbed a while and has been around that $30/share number for ages.
$1 billion fine seems like a lot, but let's assume $100/CPU, 50% profit margin and 5 years (2002-2007). 1000_000_0000 / 100 / 2 / 5 = 1 000 000. So you would need to sell 1 million CPUs to make a $1 billion profit over 5 years. Eminently doable on EU laptop market, even back then.
Definitely doesn't seem disproportionate.
All CPUs today are RISC descendants: yes, even Intel and AMD's x86.
The first "not actually a CISC x86 at all internally, but 100% a RISC" was the Pentium Pro (1995, and recycled a lot of work put into the i960, thanks to Fred Pollack), and AMD's first real[1] CPUs weren't x86 but a RISC descendant called the AM29k, which when they slapped an x86 decoder on it and it became the original K5 (1996), which sometimes won benchmarks against the Pentium, especially on code that wouldn't take advantage of the Pentium dual ALU design; the AM29k/K5's OoOE was superior due to it's RISC nature.
So, although, yes, technically there is a x86 duopoly, but x86 kind of ceased to exist meaningfully in the 90s. It is a RISC descendant monopoly, and whoever is in the lead at any given moment is who has the best fab tech; currently it's a fist fight between two Jim Keller architectures (he worked on both AMD's Zen and PASemi's ARM family) both using top tier TSMC tech. Jim Keller himself is a very accomplished RISC jockey, and I absolutely respect him and any team he's been part of.
[1] Depending on how you define real; the team that left Intel after management screwed them and started AMD, reverse engineered and reimplemented the CPU that put Intel on the map, the 8080, called the Am9080, after making a lot of "bag of chips" sort of ICs that weren't CPUs. All the other CPUs between that and the AM29k were second source chips under license.
At the very least, if AMD had been able to maintain competition for longer, would we have been stuck at 2/4 cores for so long? Still struggling to make ECC RAM mainstream?
https://en.wikipedia.org/wiki/Advanced_Micro_Devices,_Inc._v....
We have to remember that Intel went all in on Itanium and it was AMD who made 64bit happen. They absolutely kicked Intels ass on performance and price for quite a while but were not able to leverage that.
So yes, the future would have been quite different.
The only reason why they're floundering now is that Lisa Su resurrected AMD and built a star team of CPU designers that are kicking Intel's ass. However, that is not a reason to ignore antitrust as "outdated regulation". If anything, AMD is doing a some of the same shit Intel did when they were dominant, because that's just how the CPU market works now, itself because we let Intel set the standard of behavior a decade ago by failing to punish them for cheating.
Anyway, that's why we need more overly punitive fines detached from reality.
They’re operating at between a 2-4 process node handicap (depending how one counts) and turning out workable stuff (e.g. Raptor Lake) strictly by competing their butts off on micro architecture and other hard, legitimate stuff. My hat is fucking off to the people turning out product that keeps them in the race with thermal/transistor budgets from like, before COVID.
There’s a dude in my building who wears an ASML badge and I always ask him: Gelsinger going to get EUV going? He obviously can’t answer but he gets this little smirk.
I would love it if it all landed at once: HDET SKUs with 128 physical cores, ARC cards competitive with 4xxx and drivers, and a domestic pure-play foundry business that the PRC can’t even remotely disrupt in their wildest dreams.
AMD wants to generate some cap? Pay the big bucks to get ROCm up to par and trim some of that trillion off Jensen’s “how much you got on you?” pricing strategy.
It’s more like a competition fanboi alert, and watching NVIDIA sit at 1.2T in cap while AMD sits at 175B with competitive (if not strictly parity) GPUs because apparently writing a driver stack that works is out of reach for a company that could trivially finance a Sheryl-style “triple the comp of all the CUDA ballers” just…decides not to?
Yeah, I’d like to see some pressure applied by people who know how to write drivers, document their gear, and obviously haven’t cut a deal about supercompute vs. AI-per-se.
Zen4 is awesome, add in the TSMC edge? It’s what you’ve got on multi-core x86_64 at the moment.
Would I like to see alternatives to Epyc and 4090s that have the same 24Gb of GDDR6 that my 2090s did? Yes I fucking would.
And would I like to see Taiwan treated like a nation and not a geopolitical football? That sounds pretty much ethically right too.
I cordially invite your input on where I’ve gotten this wrong.
When they can stop their compiler bullshittery, I might be convinced to think they're actually competing.
And I’m well aware they’ve historically been the baddies: I was doing this in the 90s.
But no competition for cutting-edge process, no alternative on GPUs, no pure-play foundry capacity without IRBMs trained on it are like, no fun man.
The enemy of my enemy is still usually a bad guy, but balance of power is probably as good as we’re going to get on stuff that’s Capital Intensive from Outer Space.
Is this gang tackle really indicative that an Apple monopsony of current generation fab, an AMD monopoly on serious multi-core, an NVIDIA monopoly on GPUs you can just use are all preferable to a once evil company down on its luck getting back in the game?
It’s not like any of these other actors are exactly leaving margin on the table.