I haven't had much time to listen to the Intel announcement. However, the gross margin was disappointing and as I said below, if Intel misses then the stock is heading down about 5 points which it did after hours. However, I think this will stabilize and possibly go back up 1-2 points as folks look at the results with a cooler head. What was really disappointing was Intel's Q3 revenue forecast and even though they did pop the margin number back up, what it says to me is they are unable to capitalize on the technology lead they have and proportionately converting it into revenue. Unless AMD is doing what I said they might earlier and almost give away their low end at cost. In which case Intel would have had to decide to stem the bleeding at some point and not fight AMD all the way down to the mud.
The impact to AMD is that their GM's are going to be awful. Further down from the 29% they had in Q1. It feels like Intel took more share in server and desktop and potentially lost some more in mobile. The impact of a further hit on servers along with unit pressure on desktop is not going to be completely offset by mobile upside considering the small base of mobile AMD is coming off. Otellini did say they felt pricing pressure on the low end of desktop and mobile. Which was AMD trying to salvage unit share and hence hit their revenue number. So expect the revenue to be at risk. More importantly, with the margin hit their EPS numbers will be worse than forecast so I think they will lose more money than the 85 cents per share Wall Street is predicting. My earlier call was around half of their Q1 loss and I think that's now back in play. Expect somewhere in the region of a 200-250 million $ loss. The other thing to consider is the start up costs of Barcelona. If the product is pushed to Q4 effectively then some of those ramp costs will hit them in Q3 thereby suppressing margins further.
Right now - I have nothing positive to say about AMD. But Intel's execution while good, is not as good as it could have been. I think they have made the critical mistake of allowing AMD to get established - both with the OEM customer base and to some extent their brand. Which is allowing the OEMs to move their parts even if they're inferior which allows the OEMs in turn to take a tougher stance at the negotiating table with Intel.
On the market share, we won't see the results this week but I think we're on track to see some minor upside for Intel. I'm still comfortable with a point as I called it below.
Subscribe to:
Post Comments (Atom)
15 comments:
Paul Otellini
Well, ours has certainly been cleaned up and is in very good shape as we talked about. I think there remains to be seen what others will say. The rumor out there is that the inventory in other parts of the channel is a bit long right now.
http://www.arnnet.com.au/index.php/id;2085170421;fp;4;fpid;399285820
See what happens when you switch to AMD? Poor Dell. Recently Gateway and eMachines have dumped AMD with resounding success.
"The other thing to consider is the start up costs of Barcelona."
Could you expand on this - K10 is produced on 65nm so any capacity that was targeted for K10 production, could be shifted to K8 production.
Unless AMD is doing a ton of risk wafer starts for K10, I'm not sure what the "start-up" costs would be. The ramp costs would just be used to support K8 production no? (it's the same 65nm tooling for both chips)
My mentor at work has a friend on the inside at ATI (we do graphics and visualization work), and AMD's been laying off both their own and ATI's people at a rapid place. After that started happening, the talented software and hardware engineers didn't see too much of a point in staying and just booked it. They left for greener pastures (not an nVidia pun).
Right now, I'm of the mindset that even if Barcelona kicks the proverbial crap out of Penryn, I won't buy AMD because I don't want to support a company that pulls this kind of crap.
$600M gone. How much of that $2B or how much ever they got lent is left?
"To clear out inventory, AMD said it wrote-off $30 million in older microprocessor parts" (Q2 report)
Funny was Sharikou hammering on Intel that they were going to have to write off all that old Netburst inventory with Core 2 ramping up?
But hey, I'm sure this is simply an anomaly and AMD will be able to clear out all of those 3600, 3800 etc... especially after the recent cuts when their top of the line desktop (I don't count FX) is sub $200...
Startup cost: Reticles which are pretty expensive at a few million a pop. Then figure AMD must be on the 3 or 4th all layer stepping that is a few million from their rapidly dwindling cash reserves.
Then they got to run all those engineering silicon at really fast priority that will displace a ton of revenue wafers. I figure each Barcelona wafer start probably costs AMD 3 to 4 production wafers displaced. They got one small fab and they need every die right now.
AMD really blew it.
I'm really curious of that inventory write down in their quarterly annoucment. I'm sure there is some risk production Barcelona hidden their that was scrapped crap.
What is AMD to do.. with their next stepping? THey are in a cash crunch. DO they risk committing valuable silicon to another stepping that might be another bust and have hidden bugs.
AMD is finished.
180
"The impact to AMD is that their GM's are going to be awful. Further down from the 39% they had in Q1."
I don't know what you are referring to. What was 39%?
AMD's Gross Margin Percent was:
Q4 06 - 36.1%
Q1 07 - 28.1%
Q2 07 - 33.5%
That's a good increase from Q1.
Microprocessor Revenue Share:
Q4 06 - 17.1%
Q1 07 - 13.3%
Q2 07 - 18.4%
Again, a good increase from Q1.
I was surprised that ATI revenue didn't drop. with R600 arriving so late it couldn't have had much effect on Q2. This suggests that ATI's sales should be up a bit in Q3 (maybe they'll break even).
"On the market share, we won't see the results this week but I think we're on track to see some minor upside for Intel."
What would that be? I wouldn't be surprised if Intel lost a bit share in each market: server, desktop, and mobile.
Are you overlooking the numbers? AMD claims that they increased unit shipments by 38% while their gross revenues only went up 20%. So, obviously, they had some drop in ASP. However, in spite of the decrease in ASP AMD actually made more money per unit than they did in Q1. This means that AMD's production costs dropped faster than their prices.
However, we don't see this with Intel. Not only are Intel's revenues down 19% but Intel's Gross Margin Percent dropped as well. So, Intel made less money per unit than they did in Q1. This means that Intel's prices fell faster than their production costs.
Intel's Gross Margin Percent:
Q4 06 - 49.6%
Q1 07 - 50.1%
Q2 07 - 46.9%
So, the opposite of AMD. Intel has a drop from Q1.
Microprocessor Revenue Share:
Q4 06 - 82.9%
Q1 07 - 86.7%
Q2 07 - 81.6%
Also down from Q1.
Basically, AMD is moving in the right direction but isn't there yet. Intel is currently moving in the wrong direction but is still way ahead of AMD.
Okay, I found a typo on my spreadsheet. Let's see, I need to correct the Q2 processor revenue share numbers:
Intel - 84.2%
AMD - 15.8%
So, still an increase for AMD and a drop for Intel from Q1. Intel's Gross microprocessor revenue only dropped 2.3% while AMD still gained 19.6%. That should be right now.
Typo - should have been 29%. Have corrected it in the post.
Oh Scientia. This is great for AMD! Burn baby burn!
Intel's Gross microprocessor revenue only dropped 2.3% while AMD still gained 19.6%. That should be right now.
In other news, outside of the fantasy land where fools burn money chasing share and revenue, Intel's net income increased 44% year over year, while AMD's decreased an asstouding, 774%.
"Are you overlooking the numbers? AMD claims that they increased unit shipments by 38% while their gross revenues only went up 20%. So, obviously, they had some drop in ASP. However, in spite of the decrease in ASP AMD actually made more money per unit than they did in Q1."
Scientia has now said this several times (I think it his understanding the more he says it the more it might be true). Revenues were also not up 20%, Scientia is just rounding it to make it seem bigger (yet didn't round the unit shipment #)
If AMD earned more money per unit and their unit volume went up by 38%, what should have happened to their net income? (Flat, down, or up by more than 38%) Again he is confusing gross margin with profit margin....
So let's be happy about AMD's GPM's no? However as Scientia likes to point out, AMD was earning 38% in Q4 on mostly 200mm,90nm process (with some 300mm, 90nm).
AMD is now vastly higher in terms of 300mm % and also 65nm.... yet margin has done what? K10 will help this no with higher ASP's? Well only if they are getting more than DOUBLE the margin on these as the die are 2X bigger (and likely costs AMD more than 2X due to lower yield than dual core)
As for margin analysis, perhaps Scientia should factor in flash, mobo and chipset biz as Intel's margin is not CPU margin. As Intel indicated, chipset has gone up (which in turn will drag a bit on gorss margin as these tend to be lower margin than CPU). Similarly this should have HELPED AMD's GPM as they sold fewer lower margin chipsets.
As always Scientia is taking individual data points and making conclusions without a ocmplete set of data (much like he did back in Q3 when his assumption that Intel's market share gains was due to Via's loss of market share!)
Similarly one needs to factor in graphics biz. The Q4 # is a bit high as it did represent only 2 months worth of graphics biz (subsequent quarters obviously had 3). This obviously helps AMD's margin recovery as the 2007 #'s will be slighlty deflated over 2007 due to inclusion of graphics biz.
It's funny Scientia has neither a business background or a process tech background yet blogs and speaks authoritatively on both.
In one of his comments he picks on a reader who claims AMD 65nm process is limited to 2.6GHz. Scientia claims he has no proof (the original commenter was basing this on the lack of 65nm products above 2.6GHz)....
The irony of Scientia's comment? It was in the 3.0GHz barrier bog - which he uses the lack of Intel releasing 3.0+ GHz parts with the conclusion that their is a 65nm process barrier! Pot... kettle... black...
Sure he wraps it up in terms of some BS on TDP's and Tc's to make it look like he's proving something but he has absolutely ZERO process data to say that Intel can't make 3+ GHz parts... what is IDsat, Vt, Vcc limitations? Yield impacts? Is poly CD limited? gate leakage? Isub?
Oh right it's a temperature thing...
"Basically, AMD is moving in the right direction but isn't there yet. Intel is currently moving in the wrong direction but is still way ahead of AMD."
Funny Scientia dismisses the Q1 market share data point saying that the # could just be temporary and you can't look at 1 quarter for market share trends....
...now he's saying AMD is heading in the right direction BASED on 1 quarter of market share info! Could Q2 not just be a temporary data point from AMD, as they caught up on shipments after dumping on the channel in Q4/Q1? I'm not saying it is, but if Intel's Q1 # was temporary, why is AMD's Q2 # not also potentially temporary?
Pot... kettle... black....
Scientia can make up whatever market share calculation he wants... bottom line, last 5 quarters (Q2'06, Q3, Q4, Q1'07, Q2.07-prelim)
Intel 73.4%, 74.0%, 75.8%, 80.8%, 80.3%
AMD 16.4%, 16.8%, 15.7%, 10.9%, 11.4%
Other 10.2%, 9.2%, 8.4%, 8.3%, 8.3%
Source: iSuppli (no offense but this is just a bit more well known for this sort of thing than Scientia)
http://www.digitimes.com/mobos/a20070720PR201.html
0.5% RECOVERY! At this rate AMD will be back to 2006 levels in ~2 years! Don't get me wrong AMD did well to stabilize market share, but 'recovery' it is not...
Good quarter for AMD? Well it wasn't a bad one for AMD in terms of units... but AMD's largest concern right now is not unit volume it's how quickly they can return to the black so they don't have to do things like push out capex which slows their ability to grow organically.
Post a Comment