In a bit of news that’s unfortunately not an April Fool’s joke, a US District Court has ruled that AMD must face claims from investors over potential securities fraud committed by the company.

At the heart of the matter is AMD’s Llano APU. Launched in 2011, in Q3 of 2012 AMD had to take an inventory write-down of $100 million on unsold Llano inventory, as the company had to further reduce prices on the chips in order to sell them in the face of competition from Intel along with the ramp-up of their own Trinity APUs. The writedown in this case did not directly cost the company $100M, but it essentially reduced the value of the company by that much to AMD’s shareholders, whose stock in turn suffered a hit in value.

What makes this writedown lawsuit material are the events that led up to it and how AMD handled it. The participating investors are accusing AMD of committing securities fraud over how they presented the state of Llano production. The suit claims that Llano production was not as strong as AMD was claiming – a consequence of supply issues with GlobalFoundries’ 32nm process – and as a result AMD artificially inflated the value of the company in 2011 and 2012, and in the process produced too many Llano chips once GlobalFoundries was finally able to catch up. This in turn led to AMD’s $100M writedown and overall decline in value of the company and its stock price (with AMD losing about ¾ of its peak value in 2012).

These types of lawsuits are not particularly uncommon, especially as institutional investors seek restitution for money they lost from the drop in stock price. That said, today’s ruling is only over whether the lawsuit can go to trial and not over the validity of the claims themselves, never mind what specifically the investors are asking for. So it is likely that the actual lawsuit will take quite a bit longer to resolve.

Source: Reuters (via SH SOTN)

Comments Locked


View All Comments

  • dragonsqrrl - Thursday, April 2, 2015 - link

    I'm not sure what Intel would've used. Atom was still a year away, and even when it launched it was never intended for the smartphone form factor. It was too expensive, the TDP too high, and the platform too complex. But Atom was way more powerful than any mobile architecture targeting smartphones in 2008. That's probably why Apple didn't go with Intel, Intel didn't have anything compelling to offer at the time.
  • melgross - Thursday, April 2, 2015 - link

    I believe it was an ARM chip Apple asked them to make. They didn't do it because they underestimated the number Apple would need, and thought Apple's costs were wrong. It turned out that Apple needed a lot more than Intel thought, and the costs were much less than Intel thought.

    They've been paying for that mistake ever since. It would have been 300 million chips for Apple last year at an average of $20 per chip.
  • Thermalzeal - Friday, April 3, 2015 - link

    Intel couldn't provide them graphics nor were they flexible on letting Apple engineers build a more custom gpu. Same reason why the game consoles went to AMD instead.
  • Operandi - Thursday, April 2, 2015 - link

    That assessment is pretty wrong.

    Phenom and Phenom II were never meant to be amazing, they were the end of the line as far as the K8 architecture goes and considering that the Phenom II was pretty good.

    Bulldozer was an ambitious start from scratch ground up redesign and yeah that was supposed to amaze and bring AMD on par with Intel though via a completely different approach. Instead of amaze it fell flat on its first outing and never really caught up, it only really ever got to "good" under ideal work loads. AMD was never going to catch up to Intel in pure CPU performance with Dozer tech, but they haven't left the performance market vacant, just stagnant.

    Right now they starting over with Zen and I'm sure they will target as much of the high performance market as they can as that's where the margins are.
  • Coreyscomputerstuff - Sunday, April 5, 2015 - link

    the problem with bulldozer and again with piledriver is they were hedging their bets on the market pushing for multithread. The prob is as it seems with AMD they pushed way too early. The I5 4 years later has only just caught up with the FX-8350 in multithread (comparing price point). Not saying the FX chips are better but they set out to do exactly what AMD intended and built very powerful multithreaded chip using a clustered multithread configuration. The problem is they are <20% of the market and they don't get to push what standards get adopted... Intel do. As a result hardly anyone programmed to match the multithread capabilities of the bulldozer/piledriver series as it was too hard to do so for the amount of people that used AMD chips at the time. Don't get me started on games as we all know why AMD aren't strong in most games. If you don't know its due to only supporting 1-4 threads meaning most games could only utilise less than half the processor. Morale of this story and with Piledriver and bulldozer is AMD read the market wrong. They produced a powerful chip that was ahead of its time that no one wanted to adopt because it was too hard for developers to program for little gain. The issue is they had to ride it out because they spent a lot of money developing a chip no one wanted on a platform that was already around for too long (AM3+). I moved from the 8350 to a 2600 (little bit slower in multithread but way faster in single thread and most games). The AMD is in my home server which it is fantastic in. Its a multithreaded environment and it loves it the Intel is in my gaming PC which is mostly a single threaded environment and it thrives in it. This is why AMD is starting with a new CPU design. They will go from clustered multithread to simultaneous multithread like Intels hyper threading. Anyhow my rant over. AMD's APUs for the record (both desktop and laptop) are brilliant.
  • FlushedBubblyJock - Monday, April 6, 2015 - link

    That's just total crap. Piling the lipstick on the quad piglets with half siamese twin deformities on all them and calling it a great hog farm.
    AMD blew it, they couldn't even get their turbo working correctly, and right now the G3258, a $70 chip smokes them OC'ed for dirt cheap in all but a few cases.
    So desperate AMD was an is, that they took the 8350, OC'ed it to 220watts 300+ actual, and marketed it to fanboys for $900.00 released.
    At least the monster electric housefire 290x can, doubled up, and sometimes singly, be respectable and cannot be entirely discounted, because it actually performed without unrealized gimmicks and blaming everyone else, some industry, some "lack of multithreading realized by others doing coding" or some other pathetic excuse.
    Note - AMD actually tried to claim a profit from their discrete graphics division a couple quarters.

    So, looks like AMD just cannot produce single thread performance, and you know what ? If they did, they could just add cores... and WIN.

    Instead, that's what Intel did.

    AMD failed, 20, 30, 100 cores won't do it, when 1 core can't hang.
  • KAlmquist - Monday, April 6, 2015 - link

    Misreading the market was one of AMD's problems.

    If you compare the design goals of Intel Sandy Bridge (4 core) vs. AMD Bulldozer (4 module), we have:

    1) AMD designed a 315mm^2 die containing 1200 million transistors, whereas Intel went with a smaller design (216mm^2, 995 million transistors, including an iGPU). So AMD is going for raw processing power whereas Intel is aiming for lower manufacturing costs.

    2) AMD emphasized raw CPU power whereas Intel placed more emphasis on CPU power per watt (125W vs. 95W TDP).

    3) Intel placed more emphasis on floating point performance than AMD.

    4) Intel uses hyper-threading to get to 8 threads, whereas AMD has dedicated integer processing units for each thread. So the AMD is oriented towards work loads with 8 or more threads, whereas the Intel design will do relatively better on workloads with 4 or fewer threads.

    So Bulldozer tends to be faster than Sandy Bridge on heavily threaded integer workloads, but that's because each of the four design goals listed above point in that direction. But if your needs differ in any way from the optimal target for Bulldozer--for example you want to run lots of threads doing floating point computation, or you don't care about floating point perfomance but your software only uses a few threads, or you are looking for an energy-efficient way to run lots of integer threads--Bulldozer's advantage disappears.
  • Frenetic Pony - Thursday, April 2, 2015 - link

    So the suit claims that AMD misrepresented Llano chips as being produced in high quantity than was reality, which then lead AMD producing too many chips, which then lead to the writedown. No maybe I'm missing something, but that seems that what is being claimed is that one thing lead to the exact opposite and opposing thing you would expect it to lead to.
  • psychobriggsy - Thursday, April 2, 2015 - link

    It doesn't seem that consistent an argument really.

    Also, it's shares. Value can go down as well as up. Anyone doing a bit of research into the market would have seen that Intel had a strong proposition at the time, and AMD was at the end of a product line and behind Intel technology-wise.
  • ( mojah ) - Thursday, April 2, 2015 - link

    I'm not experienced in stocks but what I got from the article was that amd made it seem that they were able to produce more than they were capable at the time in order to draw investors. By the time they were able to produce the amount of chips they claimed they could, competitors already had a steady hold on the market and in order to keep their chips competitive amd had to reduce the prices of them, lowering the value of the company and thus the investors' stocks - had they known amd's chip production wasn't as profound as amd stated they likely wouldn't have made the investment in the first place. I could easily be reading this the wrong way but that's just what I got from it

Log in

Don't have an account? Sign up now