ATI FY 3Q 2006 Results and CC

[Geo]

http://ir.ati.com/phoenix.zhtml?c=105421&p=irol-newsArticle&ID=877463&highlight=

Usual caveats --What Geo found interesting, as best I heard it. No warranties offered or implied.

Very chipset-y conference, and more business-y than techie. Tho bullish on a 80nm northbridge coming, and on 80nm yields in general. Also confirmed two or three times that what are presumably RV560/RV570 are somewhat late (without really pinning down just how late). No real visibility on R600, tho some vague hints towards their FY 1Q 2007 for that. Wouldn't take 'em to the bank tho.

Dave Orton, CEO

40 patents issued in Q3.

Highlights consumer. Exceeded 100% growth year over year. 50% margins. Nokia love. MobileTV leadership. DTV record revenues 6th quarter in a row.

Some "Chuck patch" love.

Intel & ATI for Conroe love.

X1900XTX king of the hill. "Others forced to double up gpus per card" to compete. <Zing!>


Patrick Crowley, CFO

ICP/consumer growth offset seasonal lull on the graphics side. PC 77%. Desktop 35%. Chipsets 25%. Notebook 15%.

Chipsets record sales and positive margin growth. Margins now "nearly" 21%.
Record quarter for handheld.

DTV "fantastic quarter". Up 50% sequentially. 9 of top 11 LCD OEMs.

Royalties MS/Nintendo $10M. NRE $5m.

Desktop discrete margins up [but "not within corporate model"? Does that mean less than 34%? I forget what the model is].

Inventory is down.

Cash $518M. Repurched 1.5M shares. $37.2M for Xgi and BitBoys combined.

Next quarter 31-31.5% margins.

Orton

"Margin expansion is paramount". More consumer love. Europe DTV hot [thank you, WorldCup]. New Xilleon coming.

Qualcomm volume shipping begins in CY Q4 [finally!]

New Radeon products in 2nd half of CY2006. Reinforce commitment to single card performance leadership.

More Conroe love. [They seem to think Conroe helps them]

Q&A

Q: Visibility beyond August in PC? A: Lot of excitement about back to school with customers. July-September.

Q: Desktop margins? A: Year over year down slightly. Up quarter over quarter.

Q: What's driving margins? A: Notebooks a bit soft. Desktop flatish. Overall product margins will improve in Q4 in all areas, except flat for consumer.

Q: Order trends? Inventory composition? A: Lowered inventory by plan. Northbridge bulge (had a SB problem that left them with too many NB last quarter) now addressed. AGP inventory is down.

Q: Guidance assumptions? ASP? A: Business primarily OEM driven, which makes for stable ASPs. ASPs were down a touch from desktop, will be up a bit in Q4.

Q: PC vs Consumer? A: Consumer slight improvement in Q4. PC chipset down a bit in Q4 as they are past making up for the SB problem. Conroe hurts existing sales a bit.

Q: Consumer customer changes? A: Handheld something or other. [Didn't really follow it]

Q: CrossFire, how do you break the stranglehold? A: Computex was a revelation on acceptance of CrossFire by OEM. Expect Conroe to really help CrossFire, and are marketing that way. Expect physics to help drive some volumes 9-12 months out.

Q: Discrete 32-35%?. Desktop not there yet and won't in Q4. Some inventory <wonder which?> hangover. "Been late" with some products [Sounded like an admission RV560/RV570 is a little late]. Initial 80nm yields very good.

Q: [Missed one on consumer]

Q: Margins way below competitor? A: OEM is lower margins. Moved the AGP inventory, but the margins are lower. Channel is slow right now, as channel picks up margins will improve.

Q: Q4 margins for chipset? 965/975 will drop some chipset margins on existing. SB600 ramping quite well. Next gen NB is 80nm. Doing phenomenally well in the lab.

Q: ASPs with Intel and Conroe rollout? A: "Teaming up with Intel" re Conroe/CrossFire is a key element. Conroe moves existing chipsets into value territory. New design wins in fall expect significant volume.

Q: Handsets? Motorola doing well. Others as well. "Pretty much as expected". Handset ASPs start rising over the next 6 months.

Q: Price competition on CPU affecting graphics? A: Not really. "Amazed" at AGP volumes hanging on.

Q: 80nm NB chipset end of the year? A: Bring up was hours not days re "Excitement" in the lab. Excited by yields too.

Q: More chipsets and Intel/AMD price cuts A: OEM business helps with visibility. Price drops on CPUs may increase demand. "Primarily making our investments today" in the consumer business.

Q: Even more chipsets [yawn!] A: 90% of their chipset business is OEM based. Expecting channel business to increase in Q4 in part thanks to SB600.

Q: Graphics margins? Upcoming products? A: North America was good, Euro slow. Emerging markets lower mix of products. Refresh of product line late summer/fall FY Q4/Q1.

Q: Handsets volumes? A: Quite bullish on demand. Some supply chain adjustment. New platforms coming.

Q: Desktop discrete? Share gain? A: 5 segments with "5" highest. Strengths today are 1, 2, and 5. 4 and 3 are weaker; the classic $150-250 area. Talked about delays earlier [another confirmation RV560/RV570 did slip some]. Expect very strong in Q4 "in front of R600". "Quite bullish". [Damnit, 5 segments now? I just got used to 4!]

Q: [Missed it]

Orton

Thanks the troops. . .setup well for second 1/2 of 2006. Ta ta.

 

[Jawed]

Very chipset-y conference, and more business-y than techie. Tho bullish on a 80nm northbridge coming,

Ooh that's a nice surprise, I thought they were holding onto 110nm for quite a while.

and on 80nm yields in general. Also confirmed two or three times that what are presumably RV560/RV570 are somewhat late (without really pinning down just how late).

Oh dear, dodging the issue

No real visibility on R600, tho some vague hints towards their FY 1Q 2007 for that. Wouldn't take 'em to the bank tho.

Oh well, fingers-crossed that it really is Sep-Nov.

Jawed

 

[Geo]

Come to think of it. . .I think that's the first time I've heard Orton say "R600". . .

 

[tEd]

thanks alot geo

 

[Geo]

Oh dear, dodging the issue

Looking back at the FY 2Q notes, I would guess we'd be talking about them hoping to be in production (whether on sale or not) by the end of May, and they didn't make it. The statement on 80nm yields seemed confident enuf to lead me to believe it likely they may be in production now.

 

[Nelsieus]

Thanks, Geo.

I always love how Orton defends his highend parts...lol

X1900XTX king of the hill. "Others forced to double up gpus per card" to compete.

Anywho, seems like a pretty good quarter for the most part, mostly in the consumer sector.

 

[Mark]

thanks geo,

New Radeon products in Q4. Reinforce commitment to single card performance leadership.

Fiscal Q4 or calendar Q4? One would assume he's talking fiscal considering the audience, but he should really qualify those things.

 

[Geo]

Fiscal Q4 or calendar Q4? One would assume he's talking fiscal considering the audience, but he should really qualify those things.

I just went back and listed again (I hate early morning CCs because I always do them real-time, and I'd prefer to do them from a recording. Real-time makes for more shorthand and mishears. Oh well). He mixed a couple of things in that section. There was a Q4 reference that was FY to Conroe and Crossfire. The leadership part, however, seemed to be tied to "2nd half" of CY 2006 rather than "4th quarter" of FY 2006. So obviously that leaves it open whether that might have been a reference to something like an R580+. . . or R600.

Certainly later on he seemed to make a point of tieing 4Q FY2006 to being "in front of R600" (i.e. R600 would be later than that), which would mean no R600 before September. Dunno about y'all, but I'm not expecting it that soon anyway. At this point I'd be happy to get it in 1Q FY2007 (which ends in November), but am not counting on that either.

 

[NocturnDragon]

Stock is down 11% today, probably due to ATi missing Wall Street's estimates by $14 millions ($650 instead of $640).

Here there is a link to the full trascript:
http://seekingalpha.com/article/12758

 

[Geo]

Stock is down 11% today, probably due to ATi missing Wall Street's estimates by $14 millions ($650 instead of $640).

Here there is a link to the full trascript:
http://seekingalpha.com/article/12758

Whee, that's quick. I can give this up and just start analysing it instead if they do that kind of turn around consistently.

 

[NocturnDragon]

Certainly later on he seemed to make a point of tieing 4Q FY2006 to being "in front of R600" (i.e. R600 would be later than that), which would mean no R600 before September. Dunno about y'all, but I'm not expecting it that soon anyway. At this point I'd be happy to get it in 1Q FY2007 (which ends in November), but am not counting on that either.

Here there is the quote from SeekingAlpha

So we actually expect to have a very strong fall quarter in our GPU business all in front of the R600. We have a new class of products coming to market in the fall in front of the R600 family of products. We are quite bullish about that.

By fall quarter it really looks like he's meaning calendar fall quarter.
So it seems like R600 is after that, in the calendar winter quarter, so it will be 2nd Quarter Fiscal Year. Makes any sense?

 

[Geo]

Bugger. Somebody pointed out to me the following pieces add up to R600 in 2Q FY (i.e. Dec-Feb).

We expect margin percentage to improve in Q4, and again over the fall quarter.

So we actually expect to have a very strong fall quarter in our GPU business all in front of the R600.

Both from Orton. The first defines "Fall quarter" as FY 1Q. The second puts FY 1Q all "in front of R600" (i.e. before).

FY 1Q for ATI is Sep-Nov. So this would seem to suggest no R600 before December (2Q FY).

Not that we've ever over-analyzed these things. . .

 

[sonyps35]

I dont know how they spin positive in Desktop GPU's. It seems they've really struggled this product cycle. Everywhere on the ground their sales seem terrible. So much for "wait for the shader heavy games". It's been a complete turnaround for my expectations for R580 vs G71 (I expected R580 to dominate when final specs for both were originally revealed).

I continue to wonder if R600 can even be competitive with what Nividia puts out, not too mention it looks extremely tardy. Wasn't G80 slated for like, September? The rumors a while back of R600 being 16 TMU's and G80 being 96 pipes sure dont bode well, granted just rumors.

I have to be careful but I continue to just have no idea what they are doing. Nvidia is outmaneuvering them at every turn. What is their response to 7950GX2 for example? It appears none once again. Where is R580+ (which wont be enough anyway being a single card, but might help)? This is where the fact their chips are big and hot (but not any faster) really comes back to bite them. I dont of course think it would be possible to put two R580's on one card to counter 7950GX2 even if they wanted. Too hot, too much power. What is their response to qaud SLI?

I'll probably get banned for this, but I really wonder how much longer ATI will be in business in any serious way in desktops. Perhaps it's way too early to assume R600 will get spanked by G80, but I look at the fact a hypothetical 32 pipe G71 would have spanked them this cycle and they barely avoided that. They are simply not putting resources into making faster chips, but rather, a lot of things that are nice in theory but dont matter to on the ground performance. That's simply not good, and it's not smart.

You can defend there decisions, IQ over performance perhaps, among others, all day long, but when the sales show consumers aren't buying what they're selling, then they need to change course, regardles whether some posters on B3d like what they're doing.

I'm a big ATI fan. That's why they continue to frustrate me of late.

Doesn't anybody else miss the days of even X850PE (let alone 9700pro)? That card was fast, lean, and stands up well to this day. Most importantly it was fast. I wouldn't mind having one today. Still spanks 6800 ULTRA.

It is like you can draw a line between that product and X1k series.

 

[trinibwoy]

I'm really curious to see how big R600 really is. They have to be anticipating a possible (emergency?) dual-G80 flagship from Nvidia and it will probably take a beast of a chip to compete with that.

 

[Nelsieus]

I dont know how they spin positive in Desktop GPU's. It seems they've really struggled this product cycle. Everywhere on the ground their sales seem terrible. So much for "wait for the shader heavy games". It's been a complete turnaround for my expectations for R580 vs G71 (I expected R580 to dominate when final specs for both were originally revealed).

I continue to wonder if R600 can even be competitive with what Nividia puts out, not too mention it looks extremely tardy. Wasn't G80 slated for like, September? The rumors a while back of R600 being 16 TMU's and G80 being 96 pipes sure dont bode well, granted just rumors.

Well first off, I'm not sure where you're getting those rumors. I suppose by the speculation from Kristopher during Computex that G80 was a dual-core with 48 pixel shader processors (not pipelines) per core, but that seems rather unlikely, imho.

And as far as R600, there's more to look at than just the TMU count. Until I see more of the USA, I'm with-holding judgment on the speculative rumors being tossed around.

I have to be careful but I continue to just have no idea what they are doing. Nvidia is outmaneuvering them at every turn. What is their response to 7950GX2 for example? It appears none once again. Where is R580+ (which wont be enough anyway being a single card, but might help)? This is where the fact their chips are big and hot (but not any faster) really comes back to bite them. I dont of course think it would be possible to put two R580's on one card to counter 7950GX2 even if they wanted. Too hot, too much power. What is their response to qaud SLI?

Well, I think what's hurting them more isn't what's to compete with Geforce 7950GX2, but instead, what's competing with Geforce 7600. As ATI even stated, that's where their lacking, and where R560/R570 should make up. Both of those GPUs were supposed to be out around this time, but obviously have been delayed. Nonetheless, I suspect when they do arrive, we'll begin to see fierce competition once again in this particular segment.

R580+ has me totally uninterested. It will be out around the same time as nVidia is expected to launch G80. Obviously, if that is the case, I don't expect the acceptance of R580+ to be very loving (obviously as G80 would/should be much faster and be DX10 compliant.) So on this issue, yes, I think ATI will be too late.

I'll probably get banned for this, but I really wonder how much longer ATI will be in business in any serious way in desktops.

That's what people said this same time last year with the R520 delays, and obviously, ATI is still around, going quite strong. In the discrete desktop/notebook markets, they are lacking, but as this quarter proved, they have other strong sectors supplying growth.

You can defend there decisions, IQ over performance perhaps, among others, all day long, but when the sales show consumers aren't buying what they're selling, then they need to change course, regardles whether some posters on B3d like what they're doing.

I think the slow sales are a result from one of two things. The first thing, which is mostly minor, is mindset of nVidia's offerings being superior (that began when G70 came out and had absolutely no competition). But the other reason I think it just seems like sales are bad is because ATI doesn't have strong multi-GPU support like nVidia does (SLI). ATI might be selling GPUs right on track with what's expected, but obviously compared to nVidia, who can enjoy now selling two GPUs instead of one per consumer, it's not going to seem like it.

I'm really curious to see how big R600 really is. They have to be anticipating a possible (emergency?) dual-G80 flagship from Nvidia and it will probably take a beast of a chip to compete with that.

Yea, that's definately an advantage nVidia has with shrouding their upcoming products in mystery, and causes ATI to sometimes under-estimate them.

 

[Geo]

I think it's incredibly silly to say that consumers aren't buying what they are selling. They are making a profit; their overall margins are nearly back to historical levels for them. Nearly 80% of the company is PC; if people weren't buying it they'd be out of business in a big hurry, not making a profit of any kind.

 

[Nelsieus]

I think it's incredibly silly to say that consumers aren't buying what they are selling. They are making a profit; their overall margins are nearly back to historical levels for them.

I think he's just comparing them to nVidia.

 

[Moloch]

Nvidia= king of 3d.
They know how to sell cards and make huge profits, something ati fails at.

 

[poopypoo]

I'll probably get banned for this, but I really wonder how much longer ATI will be in business in any serious way in desktops.

lol, you won't get banned, just laughed at. We've heard this so many times and not for nothing, but the arguments are ignorant of the larger business as a whole. They're really doing quite competitively, and it shows.

I'm a big ATI fan. That's why they continue to frustrate me of late.

This I can relate to. The marketing machine of nV is much more powerful, and I hope to see this change. Right now, however, ATi has been playing relatively conservatively, perhaps has had a few too many eggs to juggle with chipset and HDTV concerns, and has yet hung tough. The nature of the beast is against them until R600 delivers -- hardware mags sell new products above all others, and general hype is that ATi is being beaten even though despite their difficulties, they are managing to keep nV from pulling too far ahead. We're all graphics fans and want to see a blockbuster chip soon, but blasting R600 before it arrives is just foolhardy. If you really want a $1200 graphics solution, go buy it. I'll be waiting to see what the next generation brings. I must say, it would be nice to see a refresh to R580, but I think we aren't getting it.

 

[Razor1]

I dont know how they spin positive in Desktop GPU's. It seems they've really struggled this product cycle. Everywhere on the ground their sales seem terrible. So much for "wait for the shader heavy games". It's been a complete turnaround for my expectations for R580 vs G71 (I expected R580 to dominate when final specs for both were originally revealed).

I continue to wonder if R600 can even be competitive with what Nividia puts out, not too mention it looks extremely tardy. Wasn't G80 slated for like, September? The rumors a while back of R600 being 16 TMU's and G80 being 96 pipes sure dont bode well, granted just rumors.

I have to be careful but I continue to just have no idea what they are doing. Nvidia is outmaneuvering them at every turn. What is their response to 7950GX2 for example? It appears none once again. Where is R580+ (which wont be enough anyway being a single card, but might help)? This is where the fact their chips are big and hot (but not any faster) really comes back to bite them. I dont of course think it would be possible to put two R580's on one card to counter 7950GX2 even if they wanted. Too hot, too much power. What is their response to qaud SLI?

I'll probably get banned for this, but I really wonder how much longer ATI will be in business in any serious way in desktops. Perhaps it's way too early to assume R600 will get spanked by G80, but I look at the fact a hypothetical 32 pipe G71 would have spanked them this cycle and they barely avoided that. They are simply not putting resources into making faster chips, but rather, a lot of things that are nice in theory but dont matter to on the ground performance. That's simply not good, and it's not smart.

You can defend there decisions, IQ over performance perhaps, among others, all day long, but when the sales show consumers aren't buying what they're selling, then they need to change course, regardles whether some posters on B3d like what they're doing.

I'm a big ATI fan. That's why they continue to frustrate me of late.

Doesn't anybody else miss the days of even X850PE (let alone 9700pro)? That card was fast, lean, and stands up well to this day. Most importantly it was fast. I wouldn't mind having one today. Still spanks 6800 ULTRA.

It is like you can draw a line between that product and X1k series.

ATi did seem to loose focus in the desktop segment, but they are still in a good position to take the speed crown backm but this is not as important as it so seems, what they have to do is get a better marketing team. nV has to use more transistors in thier g80 to keep up with ATi on the IQ and features front so this will even out the heat and size I think. Hypothetically if there was a g71 with 32 pipes, that would put it with a trany count of higher then the r580, so there would not have been a 7950 or quad sli. So this really wasn't an option.

The G71/r580 product cycle is similiar to the nv40/r420, 480 product cycle just flip floped, and no production problems.