AMD CFO Thomas Seifert Resigns

[CHADBOGA]

AMD's logical future was always a merger with Nvidia, once that was knocked on the head, both companies were doomed and will be lucky to even be at VIA levels by the end of this decade.

 

[ShintaiDK]

nVidia is hard to impossible to merge with anyone due to their company culture. Same reason why Intel would never buy nVidia. Would be like mixing oil and water.

 

[Idontcare]

Yeah I have to believe AMD would have been a far more fierce competitor to Intel had JHH taken over and replaced Hector.

The bottom line is when you look back at AMD pretty much everything bad for the company was set in motion by Hector. Sanders chose poorly and doomed his own company.

 

[NTMBK]

nVidia is hard to impossible to merge with anyone due to their company culture. Same reason why Intel would never buy nVidia. Would be like mixing oil and water.

Trying to make JHH fit into Intel's culture would be like trying to fit an ocelot into a biscuit tin.

 

[Idontcare]

Trying to make JHH fit into Intel's culture would be like trying to fit an ocelot into a biscuit tin.

Honestly though that is exactly what people said/thought of Steve Jobs returning to Apple in 1997.

Companies can live and die by the emergence of disruptive innovation just as well as they live and die by the emergence of disruptive innovators.

Gates (Microsoft), Ellison (Oracle), Jobs (Apple), Jensen (Nvidia), Sanders (AMD), Moore (Intel), Grove (Intel), Chang (TSMC), etc - all were disruptive innovators who had an eye for business and drove disruptive innovation in their respective businesses.

Sometimes that biscuit tin needs an ocelot. And more to the point, the ocelot will find itself a biscuit tin to occupy or it will create its own, much to the ruin of all those other biscuit tins that thought themselves too good to take up company with an ocelot.

Really in hindsight when we see what games Hector was up to in his mismanagement of AMD it is no surprise that he kiboshed the Nvidia merger. The guy wanted no competition that would put a spotlight on just how poorly he had done his job and JHH would have done just that.

 

[Genx87]

The conspiracy theorist in me says that Intel/Nvidia is paying these guys to leave so they can replace them with people who will make sure AMD stays uncompetitive. The logical person in me says AMD is a bad company to work for.

If there was a conspiracy like that Nvidia and Intel would pay the current management to stay at AMD. They havent been competitive in over half a decade on the CPU side. And the GPU side is a little better. Nvidia has done fine since the 5000 series vs 480 over 2 years ago.

 

[Ferzerp]

And the GPU side is a little better

GPU side is a lot better, not just a little. I don't keep up with the financials. I am merely speaking towards the quality of product. On the GPU side that have skus that are competitive from top to bottom.

 

[NTMBK]

Sometimes that biscuit tin needs an ocelot.

Okay, that's going in my sig.

 

[Dribble]

GPU side is a lot better, not just a little. I don't keep up with the financials. I am merely speaking towards the quality of product. On the GPU side that have skus that are competitive from top to bottom.

GPU side isn't that good - it's not competitive where it matters. Yes consumer cards are good, but there is little money in them. The big markups are in workstation and gpu compute, nvidia has had those pretty well locked down for years. AMD could compete there, but that would take a big investment in software and drivers, something AMD doesn't seem to want to do.

While the cpu side isn't very competitive they are most competitive as chips for servers and super computers where the markup is much bigger.

What's most worrying is I don't think AMD has a clear vision of where they are going. All I see with AMD is fusion, which is basically boils down to competing with Intel for the budget laptop market - limited profit in that. Beyond that there is no unified vision - they seem dip their feet in various markets, produce a set of marketing slides but don't really commit so don't really get anywhere.

 

[Ajay]

Does AMD ever have good news?

Yes, it's just that the ratio of bad new to good new is just too high. Intel has 'bad' news, but it's not a big deal when the bad news (bug, delays) occurs 1:10 compared to good news. And Intel marketing seems to do a really good job at pushing that goods news to make sure it get's noticed.

 

[Borkil]

GPU side isn't that good - it's not competitive where it matters. Yes consumer cards are good, but there is little money in them. The big markups are in workstation and gpu compute, nvidia has had those pretty well locked down for years. AMD could compete there, but that would take a big investment in software and drivers, something AMD doesn't seem to want to do.

While the cpu side isn't very competitive they are most competitive as chips for servers and super computers where the markup is much bigger.

What's most worrying is I don't think AMD has a clear vision of where they are going. All I see with AMD is fusion, which is basically boils down to competing with Intel for the budget laptop market - limited profit in that. Beyond that there is no unified vision - they seem dip their feet in various markets, produce a set of marketing slides but don't really commit so don't really get anywhere.

Isn't that where the market is heading? the average person doesn't drop a grand on a high end desktop. I think a smart move would be to focus more on the low power and mobile space

 

[Dribble]

Isn't that where the market is heading? the average person doesn't drop a grand on a high end desktop. I think a smart move would be to focus more on the low power and mobile space

Well fusion is very much for budget laptops only right now (a netbook being a very budget laptop) - AMD hasn't got the power usage down enough to be considered seriously elsewhere. Being as Intel are finding that really hard and they have infinite funds and a much more power efficient manufacturing node I wonder how well AMD can do in true mobile devices.

However assuming they do manage it then if you look at the breakdown of a PC, say 1/3 of cost is cpu + graphics. If you look at break down of a tablet/phone, then the soc might cost $15-25 which includes sound, graphics, cpu, etc. That's maybe 1/10-1/20 of the cost of the finished device. Both cost about the same e.g. iphone/ipad costs about as much as a cheaper laptop/pc.

So if that's the way the future is headed, where the cpu/gpu are a smaller % of finished devices costs then you've got to be careful. Sure there is money there if you sell enough chips, but I would have hoped AMD would have a stronger strategy in markets where the chips cost more and the mark-up is much bigger.

 

[frozentundra123456]

GPU side isn't that good - it's not competitive where it matters. Yes consumer cards are good, but there is little money in them. The big markups are in workstation and gpu compute, nvidia has had those pretty well locked down for years. AMD could compete there, but that would take a big investment in software and drivers, something AMD doesn't seem to want to do.

While the cpu side isn't very competitive they are most competitive as chips for servers and super computers where the markup is much bigger.

What's most worrying is I don't think AMD has a clear vision of where they are going. All I see with AMD is fusion, which is basically boils down to competing with Intel for the budget laptop market - limited profit in that. Beyond that there is no unified vision - they seem dip their feet in various markets, produce a set of marketing slides but don't really commit so don't really get anywhere.

Are they really competitive in the server segment? Doesnt Intel have like 90 or 95 percent of the server market?

 

[Ferzerp]

In the x86 server space, AMD's marketshare is in the 5-7% range, correct. Intel sells 15-20 Xeons for each Opteron sold.

 

[exar333]

Are they really competitive in the server segment? Doesnt Intel have like 90 or 95 percent of the server market?

Yeah. They had something like around 20% just 4-5 years ago and is now less than 5%. That is a HUGE loss in market share.

Once Intel's Core tech made it's way into the server space around 2007, it started really eating into the ground that Opterons had made from 2003-2007.

 

[Ferzerp]

Yes, it is AMD's weakest space by far.

edit: In my view, the worst thing for AMD is their image. In the GPU arena, even though they tend to trade blows with nvidia, nvidia typically outsells them by quite a bit. And this is in their strongest market. They have a real issue with marketing, and it has always seemed that their marketing department feels that pulling stunts, using hyperbole, and being misleading is the way to improve their image despite it not really helping up to this point.

 

[thilanliyan]

using hyperbole, and being misleading is the way to improve their image

Works for Apple...

I agree though...the average person doesn't know what an AMD CPU is, nor what a Radeon video card is, whereas they definitely know what an Intel CPU and Geforce video card is.

 

[Ajay]

Yeah. They had something like around 20% just 4-5 years ago and is now less than 5%. That is a HUGE loss in market share.

Once Intel's Core tech made it's way into the server space around 2007, it started really eating into the ground that Opterons had made from 2003-2007.

Better margins though, even with less market share. Cloud computing on the consumer side is driving up the need for high density data centers. I think Rory Read said something about about still having a focus on servers, so that's probably what Steamroller is optimized around. For AMD's sake, it better be good, because it looks like PD missed the mark.

With the kind of compaction allowed by synthetic tools, AMD could hit a sweet spot in lower clocked, low power and lots of cores. That would be a good setup where compute power can be brought up on demand using vSphere. But HW-EP will be right behind it. I imagine that Excavator is targeting 20nm SHP, it will take a miracle for AMD to put out something that that can compete with Broadwell in servers though.

 

[Dribble]

Are they really competitive in the server segment? Doesnt Intel have like 90 or 95 percent of the server market?

I suppose not now, but still more competitive then in most markets, the new BD architecture was built for servers first. While they have lost a lot they still have some % of the market, and you still see some AMD powered super computers as their interconnect is so good.

 

[Ferzerp]

I suppose not now, but still more competitive then in most markets, the new BD architecture was built for servers first. While they have lost a lot they still have some % of the market, and you still see some AMD powered super computers as their interconnect is so good.

No. As we've referenced, the server market is AMD's absolute weakest market. 5-7% or so. That is not "more competitive than in most markets".

 

[frozentundra123456]

I suppose not now, but still more competitive then in most markets, the new BD architecture was built for servers first. While they have lost a lot they still have some % of the market, and you still see some AMD powered super computers as their interconnect is so good.

That to me is the biggest failure of Bulldozer. I could excuse its weakness as a desktop chip with the "it was designed as a server" rationale, but it hasnt taken off in the server market either, probably because of its excessive power consumption.

 

[Blitzvogel]

How did AMD lose to Nvidia for the current round of Macbook Pros and iMacs with dedicated graphics? That's a big loss in terms of cash (5 million+ GPUs across all models per quarter term?). Did Nvidia's products beat the rival AMD GPUs in OpenCL performance at set price points, because I figured AMD had OpenCL locked down pretty hard or was it just Tahiti that truly outdid Nvidia?

Oh, AMD........

I hope the return of Jim Keller turns things around, but anything he does will take two years minimum before the changes come into effect or see the light of day in production processors.

 

[meloz]

How did AMD lose to Nvidia for the current round of Macbook Pros and iMacs with dedicated graphics?

They might not have the ability to supply the amount and type of SKUs Apple wanted due to production issues at TSMC. (TSMC always have wafer supply issue, not yield issues. )

Or they might not have wanted to supply to Apple when they realized what Apple was willing to pay.

That's a big loss in terms of cash (5 million+ GPUs across all models per quarter term?).

It might be a loss in potential cashflow, but not necessarily a loss for profitability. Apple is famous for sucking every last drop of blood out of their enemies, ur, I mean "partners". Selling 5 or 50 million GPU / APU for a $0.1 profit on each unit won't make AMD much money.

The whole integrated / iGPU market is steadily being taken over by Intel, anyway. Not just because of the hardware improvements (Haswell....), Intel are the only one of the big three to have have open-source graphics drivers (which is also steadily improving).

In another generation or three we might see the end of dedicated graphics as far as notebooks are concerned. Eventually the desktop and workstation markets will feel the heat, too.

I hope the return of Jim Keller turns things around, but anything he does will take two years minimum before the changes come into effect or see the light of day in production processors.

Best wishes to Jim, we really need a strong AMD, if only to spur Intel.

For sure, there is this 'latency' between a decision taken at executive level, and how it affects the company's products and performance. Most of the bad things we have observed about AMD are really the doing of imbecile Ruiz. The executives after him were left there to take the blame while he went on to enjoy a wealthy and thorougly undeserved retirement.

 

[MarkLuvsCS]

How did AMD lose to Nvidia for the current round of Macbook Pros and iMacs with dedicated graphics? That's a big loss in terms of cash (5 million+ GPUs across all models per quarter term?). Did Nvidia's products beat the rival AMD GPUs in OpenCL performance at set price points, because I figured AMD had OpenCL locked down pretty hard or was it just Tahiti that truly outdid Nvidia?

Could it be related to how well Optimus works compared to AMDs switchable graphics? I know they have had a really rough time for over a year trying to get their switchable graphics working properly. I'd imagine that alone could have been a deal breaker, because without functional switchable graphics for GPUs, they consume too much power and significantly effect battery life.

 

[exar333]

Could it be related to how well Optimus works compared to AMDs switchable graphics? I know they have had a really rough time for over a year trying to get their switchable graphics working properly. I'd imagine that alone could have been a deal breaker, because without functional switchable graphics for GPUs, they consume too much power and significantly effect battery life.

That's definitely a consideration. Because of the adoption rate of Optimus, and the fact AMD has no mid to high-end CPUs on mobile, the Intel + NV option is by far the most used for middle to high-end gaming laptops right now.

The sour taste that the 7970M has put in a lot of high-end gamer's mouths probably will not help them in the near-future.