The Rise and Fall of AMD.

[mrmt]

Couldn't Intel cook their books to inflate the cost of R&D (and make slowing down node shrinking possible)? They're a vertically integrated company with international subsidiaries which would be easier to shuffle costs around wouldn't it?

What's the point of inflate R&D costs? I mean, they can slow down node shrinking whenever they want, nobody is obliging them to do that, why would they have to cook their books to slow down?

 

[bononos]

What's the point of inflate R&D costs? I mean, they can slow down node shrinking whenever they want, nobody is obliging them to do that, why would they have to cook their books to slow down?

IDC mentioned that running at (too) high gross margins might obligate the FTC to step in especially since Intel is looking like it might be the only player in town within the next very few years. So its one of the reasons Intel has to pump extra money to essentially compete against itself and keep cranking out ever better faster chips at a goodly pace.

So I was just thinking if Intel could actually slow the pace of innovation without perturbing the FTC.

Edit: Why would they want to do that? Meeting financial targets would be less risky, Intel could behave like a benevolent company and dole out more leave, everyone wins.

 

[ShintaiDK]

So you are saying they would commit accounting fraud?

 

[Haserath]

Intel needs the node shrinks for an edge. Why would they slow down?

 

[mrmt]

IDC mentioned that running at (too) high gross margins might obligate the FTC to step in especially since Intel is looking like it might be the only player in town within the next very few years. So its one of the reasons Intel has to pump extra money to essentially compete against itself and keep cranking out ever better faster chips at a goodly pace.

So I was just thinking if Intel could actually slow the pace of innovation without perturbing the FTC.

I see your point now. The short answer is no, never. The long answer goes below.

For them to do that they would have to incur in expenses that didn't happen, so they would have to forge an expense, the receipt and move the money elsewhere. The problem is that they would have to put the money completely out of their subsidiaries, as they have to consolidate those in the financial statements in the SEC fillings. I don't think any sane management would do that, as they would not be defrauding the IRS or the SEC, but also the shareholders. It's the kind of gargantuan screw-up that they don't want to bring over Intel, something at the levels of Enron or Worldcom.

There is another alternative, bring some items that would be deemed investments as costs, which would erode their current results but not he future ones, but there is a limit on how far you can go with this. The further you go, the further you are likely to breach some cardinal FASB rule.

I see that you worry a lot with Intel gross margins, but the FTC doesn't. The FTC won't bother with Intel margins, they are likely to bother with consumer prices. If Intel platform costs keep rising from generation to generation then the FTC is likely to step in to rein Intel, but if prices are more or less stable, or slowly decreasing, even if margins were going up a bit, then no.

Intel margins aren't likely to go up too much if AMD retires from the market. The only area where Intel faces competition is in the bottom market and there it is Intel who dictates prices, not AMD, so they are already very comfortable with prices and margins. To move prices up would mean impact volumes, something they really don't want to do in the medium term.

Intel margins are likely to influence the resolve of any would-be competitor to pick a fight with Intel. One thing is to buy an AMD with 35% gross margins and Intel with 60%. But let's say that Intel reaches 70% gross margins, it is highly unlikely that a company a bit less incompetent than AMD could not field products that generated around 45% gross margins and make money. So while Intel must generate enough cash to make shareholders happy and keep them ahead of the other players, they must calibrate their prices in order to not attract undesirable competition.

Keep in mind that manufacturing is a fundamental stone in Intel strategy, much like having the best x86 design on the market. It is the best design making use of the best fabs available that makes Intel chips what they are. I don't see Intel slowing down even if they were alone, much less with ARM around the corner.

 

[frozentundra123456]

Intel needs the node shrinks for an edge. Why would they slow down?

I think they need the node shrinks more to compete with ARM than in their x86 business. They are already far ahead of AMD in performance per watt (and will be even further ahead when Haswell comes out), but the manufacturing advantage is really all they have to try to compete with ARM in smartphones and tablets.

The only reason they might "want" to slow down is that each shrink will be more expensive, technically difficult, and may start to yield diminishing returns. Yes they can get more chips per wafer, but there may be a point where the increasing number of chips per wafer is more than offset by lower yields and higher manufacturing and R and D costs.

 

[Idontcare]

...but there may be a point where the increasing number of chips per wafer is more than offset by lower yields and higher manufacturing and R and D costs.

For what its worth, every new node starts out with that being the case, that is why it isn't in production any sooner than the date it comes to be put into production.

A node takes about 4yrs to develop before it goes into production, and pretty much the entire last two years of that 4yrs are dedicated to getting the yields and reliability up high enough such that it makes financial sense for the company to start producing chips on it versus the node it will be displacing.

Just wanted to point out that what you wrote in your post has been the reality of node development and node cadence/introduction timeline since the 60's. It is the exact reason why Intel is producing 22nm chips now and not 14nm chips right now, because in its present state of yields and manufacturing cost per wafer the 14nm process is so much more expensive than 22nm that is makes no sense to put 14nm into production today (in its current state of yield readiness).

And that is why node-cadence has slowed for most companies other than the top-10. If you aren't pumping in the R&D dollars then it takes you 3 yrs to get that last phase of yield entitlement completed instead of the usual two years, so your node then takes 5yrs to put into production instead of the usual 4yrs, and then you find yourself a year behind Intel.

Or you start trying to take shortcuts, going with less complex projects like gate-first integration instead of pursuing gate-last integration and so forth, so you hit your timeline in a competitive fashion but the underlying process node itself is not competitive in terms of power consumption, clockspeeds, and xtor density.

The thing with Intel is that since they are already the world leader in the field of process technology, we wouldn't know if they slacked off in their node cadence because they truly ran into R&D diminishing returns or if they decided it was time pull back on the R&D investments so as to intentionally slow down the node cadence (like AMD did with their 65nm node). Only the folks on the inside would know if that was going down.

 

[Idontcare]

Couldn't Intel cook their books to inflate the cost of R&D (and make slowing down node shrinking possible)? They're a vertically integrated company with international subsidiaries which would be easier to shuffle costs around wouldn't it?

They could, it wouldn't be beyond the capacity of humans to fall victim to the temptation of manipulating their accounting as we have seen in countless examples from the semiconductor industry to the banking and investment industry.

But my contention is that the outcome there is the same, eventually they will get caught out because eventually some whistleblower will find out OR someone at Intel who was in the know would get spurned (passed over for a promotion, or an office romance sours, etc) and they would turn evidence just for the human schadenfreude aspects of watching their ex-coworkers get burned over it.

And that'll be the day the party at Intel HQ would end, the same as if they decided to jack prices to $2000 for a celeron, the same as if they decided they would only release new nodes once every 10yrs, etc.

All these moves that Intel could do to screw over the consumer, and to be sure they most certainly could do them, are all short-term plays that result in the demise of Intel in the span of a few short years.

So we either have an Intel that is well managed and continues on the path it has been on, and the consumers as well, or we have an Intel that elects to commit corporate suicide by any number of means and in that case the consumer suffers a brief short-term (1-3 yrs) period of dealing with Intel jerking them around before *poof* Intel no longer exists beyond name only because the FTC or DoJ busts them up.

There is of course a third option - collusion between Intel, the FTC and the DoJ - in which case Intel reigns unholy terror down upon the hapless consumer whilst the FTC and DoJ, on the take no doubt, turn a blind eye to it all.

But if that happens the consumer has bigger issues than an evil monopolistic Intel, because at point the consumer doesn't have a functioning FTC or DoJ and that reality (if it came to be reality) should warrant a bit more activism above and beyond whatever we would have in store for Intel at that juncture.

 

[Matt1970]

AMD would do well to stop putting those E-300 processors in Laptops and PC's. Those are pathetically slow processors and are probably leaving first time AMD users with a bad taste for AMD.

 

[mrmt]

AMD would do well to stop putting those E-300 processors in Laptops and PC's. Those are pathetically slow processors and are probably leaving first time AMD users with a bad taste for AMD.

It's better you get used to it because E-300 series is *exactly* what AMD wants to promote.

 

[Ancalagon44]

AMD would do well to stop putting those E-300 processors in Laptops and PC's. Those are pathetically slow processors and are probably leaving first time AMD users with a bad taste for AMD.

The irony is that the AMD Brazos chips are the last CPU products that outperformed their Intel counterparts upon release.

In other words, they were good for their time.

Now, they are outdated and slow, with Ivy Bridge low voltage CPUs draining a similar amount of powe while being an order of magnitude faster.

 

[NTMBK]

AMD would do well to stop putting those E-300 processors in Laptops and PC's. Those are pathetically slow processors and are probably leaving first time AMD users with a bad taste for AMD.

Blame the laptop makers, not AMD. I've got a C-60 netbook, and it's damn fast for a tiny little thing like that. But a souped up version in a 15" chassis? Shoot me now.

 

[makken]

That's cool, especially since I have open office downloaded on all 4 of my computers for free. Microsoft office is good except for I believe it should cost half as much (and the price is about $90 I think for it). Operating systems are overpriced too. When a copy of windows cost 1/4th the cost of the last pc I built, I realized something was wrong. Not saying that I'm paying for a bad product, but considering that it costs that much just for a few simple programs is a little too much for me to be happy with. They're not outrageous prices, but just enough for me to convince myself they're worth buying again rather than switching, which is exactly where they want to be. Especially with me being in college and everyone using word, open office just doesn't work sometimes, and that's something I'm not happy with.

All in all, that's where Intel wants to be now and will want to be in the future too. High prices, but cheap enough to convince us we want to buy. Marketing sucks right?

I brought windows 8 for $15
I brought office 2010 for $10
I brought window 7 for $30

Microsoft is just one of those companies where you should learn to take advantage of their discounts.

 

[serpretetsky]

Blame the laptop makers, not AMD. I've got a C-60 netbook, and it's damn fast for a tiny little thing like that. But a souped up version in a 15" chassis? Shoot me now.

agreed. I had a lenovo x120e with an e-350. I loved that machine. So much faster than atom.

However, all of these full size laptops and desktops with e-350? I don't get it.:|

 

[mrmt]

agreed. I had a lenovo x120e with an e-350. I loved that machine. So much faster than atom.

However, all of these full size laptops and desktops with e-350? I don't get it.:|

Intel received a lot of flak because of their decision to confine Atom to 10.1 inch screens. Now AMD is receiving flak for doing the opposite. On this one I have to defend AMD. If OEMs want to do those builds and there are people buying, why not?

Plus AMD haven't a choice on this one. The Bulldozer ghost haunts them here, as the big size of the die makes the chips unsuitable for the bottom market. You can bet that Intel is leveraging heavily on their smaller chips when selling to OEMs, probably selling chips charging less than it would be viable for AMD. Only Brazos can compete on price on Intel, so the decision isn't "should I allow OEMs to put Brazos on 15 inch notebooks?" but "Do I sell them Brazos or nothing?".

 

[SickBeast]

I think that the Brazos platform has a lot of potential going forward, and if I were AMD, I would be putting most of my resources there rather than for enhancing Bulldozer. That's how Intel recovered from the P4 and I think if AMD can come out with a really solid low power chip for ultrabooks with good integrated graphics they will do quite well.

 

[exar333]

Intel received a lot of flak because of their decision to confine Atom to 10.1 inch screens. Now AMD is receiving flak for doing the opposite. On this one I have to defend AMD. If OEMs want to do those builds and there are people buying, why not?

Plus AMD haven't a choice on this one. The Bulldozer ghost haunts them here, as the big size of the die makes the chips unsuitable for the bottom market. You can bet that Intel is leveraging heavily on their smaller chips when selling to OEMs, probably selling chips charging less than it would be viable for AMD. Only Brazos can compete on price on Intel, so the decision isn't "should I allow OEMs to put Brazos on 15 inch notebooks?" but "Do I sell them Brazos or nothing?".

Sales (revenue) is important, but not to the extent that it hurts your brand. This is one area that Intel has managed very well over time, and they can sell their CPUs for more, because of it. If consumers buy 'AMD' branded hardware and it sucks because of how the 3rd party implemented it, AMD suffers as a brand. Most consumers don't really buy a phone because it has 'Qualcomm inside' or anything like that, they are attracted to the handset company (LG, Samsung, Apple, etc). Desktops brands, though, definitely do sway the masses, so to speak, when looking to purchase.

 

[exar333]

I think that the Brazos platform has a lot of potential going forward, and if I were AMD, I would be putting most of my resources there rather than for enhancing Bulldozer. That's how Intel recovered from the P4 and I think if AMD can come out with a really solid low power chip for ultrabooks with good integrated graphics they will do quite well.

Totally agree. AMD should focus on Brazos and make it better. It's a great product.

 

[Lepton87]

The irony is that the AMD Brazos chips are the last CPU products that outperformed their Intel counterparts upon release.

In other words, they were good for their time.

Now, they are outdated and slow, with Ivy Bridge low voltage CPUs draining a similar amount of powe while being an order of magnitude faster.

An order of magnitude? Really? Maybe desktop chips certainly not ULV IVY.
Even the fastest IVY might not be up to the task
http://www.anandtech.com/bench/Product/328?vs=551

I won't bother to multiply E-350 numbers by ten, but if I did it would certainly be competitive to the fastest desktop IVY on the market.

 

[mrmt]

Sales (revenue) is important, but not to the extent that it hurts your brand.

And what is AMD supposed to do? Not sell to OEMs? Give the market entirely for Intel?

 

[Puppies04]

Couldn't Intel cook their books to inflate the cost of R&D (and make slowing down node shrinking possible)? They're a vertically integrated company with international subsidiaries which would be easier to shuffle costs around wouldn't it?

Sounds risky.

A safer way to slow down innovation would be to secretly force AMD out of large parts of the market where they wanted things to progress less quickly. Off the top of my head I would assume they could achieve this by paying large pc manufacturers a load of money to only use intel products thus placing a strangle hold on AMDs revenue........only kidding

 

[Lepton87]

Sounds risky.

A safer way to slow down innovation would be to secretly force AMD out of large parts of the market where they wanted things to progress less quickly. Off the top of my head I would assume they could achieve this by paying large pc manufacturers a load of money to only use intel products thus placing a strangle hold on AMDs revenue........only kidding

They already did that, use some imagination

 

[frozentundra123456]

They already did that, use some imagination

Intel doesnt have to force amd out of the large segments of the market. AMD has managed to do that to itself with poor performance at the mid range and high end, and pretty much forced itself out of the most profitable market (servers) with poor performance per watt, even though their flagship chip was supposed to be great for servers.

 

[exar333]

And what is AMD supposed to do? Not sell to OEMs? Give the market entirely for Intel?

This is exactly why so many companies fail. They only look 1-2Q ahead and long-term brand-recognition be damned.

Look at the root of the issue. If they had a well-priced, decent-performing more powerful option for a standard 15'' laptop, partners would not be looking to stuff a Brazos in there to begin with.

 

[acx]

AMD now selling assets to lease back. Trading increased operating costs to raise cash.

http://www.reuters.com/article/2012...feedType=RSS&feedName=financialsSector&rpc=43