they might get away with stuff elsewhere but in new yawk they're messin' with spitzer...

gb

I doubt the NY thing will ever happen. AMD is in no position to take that on. They may have to delay the Fab30 to Fab38 conversion to 65nm to save cash. I don't know how they could justify breaking ground in NY in light of that.

It will be interesting to see how they keep Fab30 busy on 90nm if they delay the conversion.

gb

As far as price/book:

AMD's design teams are now in Canada, Texas, California, Colorado and I think a small one in Washington. I think the hard assets consist of non-production real estate and office and lab equipment. Likely inconsequential dollar wise.

Their only fab facilities are in Dresden and while they might be able to sell equipment from those places it certainly will sell at a discount. Who is going to buy a logic fab in Dresden?

Lastly there is the value of the IP which has a major foundation in patent cross licenses which is suspect.

So while AMD may have a paper book value, I suspect reality is well below that.

What ever hard assets they do own will likely be collateralized for this next round of financing, much like the restrictions on the proceeds from the sale of Spansion shares that occurred with the ATI financing.

gb

I don't think that Barcelona will affect AMD financial performance in a material way this year in any segment. It will be a big, unproducible 65nm die as a quad. I haven't seen any indication from AMD that they will launch a dual core variant of it this year. Did I miss it?

If you haven't seen today's Intel presentation go to their investor page on their website and listen to Skaugen.

I think after AMD's April Q1CC the institutions will throw in the towel as it becomes clear that Hector's oratory "from the gut" doesn't offset the prospect of a steady stream of negative cash flow quarters.

And I don't think the institutions will be happy to see their shares diluted by another round of convertible financing. I think they'll want to get out before that happens which I'm guessing will be in Q2. AMD can't let their cash get too low and risk the loss of their cross license through insolvency. Remember people only loan you money when you don't need it. The more you need it the less likely you'll get it. That's why I think AMD will raise cash in Q2 rather than later in the year when their problems are much more immediate.

As far as reducing cash burn...I'd like to see them explain pushing out capex to the analysts in light of Intel shipping 45nm this year. I don't think they could explain shedding any architecture or design teams either in light of Tick/Tock.

And their acq of ATI merely greenlighted an Intel discrete graphics play while burdening them with debt for a money loser.

Rock and a hard place.

gb

I'm don't think that Ruiz is suicidal. I just think that AMD has few options to continue as they have in the past. Until they can figure out how to dig out of the hole they're in they have to talk big.

Intel's Tick/Tock strategy if executed properly is one that is extremely difficult to compete with.

I'm always surprised at the level of risk some investors will take. It will be interesting who steps forward this time and what risk mitigation they demand in addition to premiums. What AMD has to do to obtain financing may indicate where they are heading. I doubt they could find deep pockets to continue their current strategy unmodified.

I wonder how much longer the institutional investors will hang on? It's one thing for the retail investor to take a clobbering but the institutions must be squirming in their seats given the magnitude of their holdings.

Anyone have the latest institutional holding data?

gb

i don't think they'll make it "well past nehelem" as a head to head competitor with intel. if they are to survive as a separate publicly owned company they are going to have to change their business model.

this summer's cash raising activity will be their last in my opinion with their current business plan. frankly i think they'll be insolvent by the end of '08.

the only way they will survive is if intel's 45nm is a complete disaster.

gb

actually i think that clause was intended to keep the cross licensed patents out of the hands of those who might obtain the bankrupt company's assets on-the-cheap then shop them around.

gb

i agree.

but it will come at a dear price.

gb

except the last time amd went for a major pile of dosh was when they were just launching opteron with a major fanfare.

i don't see anything like that to attract potential investors. folks only want to lend you money when you don't need it. and just because you get some cash doesn't really help if you don't have a way to turn a profit on a regular basis.

as someone else posted amd can't risk insolvency and risk their cross license. they can't even do chapter 11 reorg so they have to continue to go to the well to stay liquid. unfortunately every time they do so they further poison that well.

i think this next round will come with "a-knuckle-a-week juice"...and it may well be the last round.

gb

Someone who knows much more than I do about startup costs made an interesting observation. A big chunk of "startup costs" are wafers used to wring out the fabs that are thrown away.

If suddenly those wafers become saleable that changes the accounting considerably.

So Intel's startup costs overweighted in 1H07 per Bryant may have a lot of potentially scrap wafers that suddenly become valued inventory if things go right.

gb

interesting about the escape clause. i didn't know about that.

and a consortium around building a "something for everyone" process will be slower even if they are successful. and success for the consortium may still not produce a process which amd can use to compete with intel at least not without a lot more work of their own.

up sh** creek without a paddle...

gb

I think this allows Intel to evaluate a lot of technologies that may be interesting for the graphics chips as well as general computing.

Simulation is good as far as it goes but sometimes you need to do an implementation to find things you hadn't thought of.

And it's better to find them out in a research context than when trying to make a product schedule.

BTW, this kind of research is what separates Intel from AMD. There's no way AMD could afford the team necessary to do this kind of work. And I'm sure AMD has had some internal discussions about what impact this disparity holds for their future. I don't think making a socket available on a motherboard (Torrenza) is in anyway comparable.

Perhaps some more nursing off the IBM teat?

gb

Bryant is a *major* beancounter. I wouldn't worry about the beans.

Bryant did state at the Jan CC that the 45nm startup costs would be overweighted into 1H07 which implies some production this year. This is probably finishing D1D 45nm and getting the next fab (AZ or NM?) 45nm complete in Q3 which could ramp then.

D1D could probably pump out some wafers in late 1H07 sufficient to maintain or even extend the highend lead while AMD fumbles around.

Israel comes on line in 1H08 so the remainder of 45nm startup costs would be spread in 2H07 and 1H08.

The concern about 65nm is misplaced. The majority of CPUs even in 08 will still be on 65nm even if 45nm starts in July 07.

45nm this year is all about leadership and showing that AMD is falling further behind.

gb

I wonder if Intel, though simulation, was able to show that current incumbent appproaches would hit a wall and this approach would be more scaleable?

Intel's internal simulation capability is likely unmatched anywhere.

Further they certainly have better insight into targetted process and production capacity than the incumbents.

Intel's combination of design, process and manufacturing applied to this kind of problem might be quite formidable.

And their internal compiler, debugger and profiler tools are excellent. The only complaints I ever heard about them was price.

gb

Some thoughts about Intel discrete graphics.

I see a lot of posts about how difficult it will be for Intel to compete with Nvidia and the former ATI for highend graphics. I don't buy it.

It's not like Intel doesn't have some idea of where graphics ought to go. Intel Architecture Labs have done a lot of work in graphics. Certainly the chipset guys have plenty of ideas that don't get implemented due to cost concerns on integrated graphics.

Simulation? Lots of simulation capability

Architecture? Lots of smart folk who would love to lead. There's only so many lead CPU architecture slots.

Once an architecture direction is set implementation of the blocks shouldn't be that much different than doing the same for a CPU.

Circuit design? Plenty of those folks.

Fab, process, OEM customers? Doh...

I think the only thing holding Intel back until now was a concern about what the FTC might think about trampling ATI and Nvidia. AMD solved that problem for them.

The only question in my mind is when the first product hits the street.

gb

32bit

no need to ask them...the info is at their websites. and i'm not asking how to make the software run.

i'm suggesting that msft's vista marketing means that even home users may delay a pc purchase until vista software compatibility improves since the only pcs they can buy are loaded with vista.

it's intc-investment pertinent.

gb

Interesting Vista problem:

My daughter is an artist and is uses Photoshop regularly. She has begun adding 3DsMax to her toolbox and is taking a class on it.

In order to do work at home, rather than in a computer lab at school, we've purchased the program and are running it currently on my wife's "hotrod" C2D machine under WinXP MCE. This machine was purchased this past Christmas.

We're now considering getting my daughter the same machine to replace her 4 year old notebook. But when I go to the Dell site I see that the same machine is only available with WinVista. Unfortunately neither PS or 3Ds are supported on Vista. PS apparently runs after a fashion but 3Ds doesn't run at all.

For us this means foregoing a PC purchase until these programs run. I wonder how many other PC purchasers are similarly stalled?

Might make for an ugly couple of quarters until MS marketing and software reality align...

gb

it would be really interesting to see who ibm would go to for fab of power and mainframe cpus.

any thoughts on candidates?

gb

two coffee spews in two days!

this is getting to be too much

gb

sounds like the rabble are cooking medieval stew.

"last one to sleep, throw a rat in the pot"

gb (i wonder what the king is eating...)

ROFLMAO!!!

that was good for a coffee snort through the nose as well!

gb

captain a few first mates may be stormproofed but the galley slaves are not.

gb

all this babble about die size and process leveling ignores the fact that smart people are continuing to work on architectural improvements and integration opportunities.

i don't expect any slowdown in what can be achieved and i don't expect average die sizes to get smaller either.

this kind of talk comes around periodically, most often when amd is down and its fans are hoping for the next saving grace.

ain't gonna happen, and amd has a mountain of debt this time around. they have yet to ramp 65nm in fab36 and will have not started converting fab30 by the time inlel starts shipping 45nm for revenue. their cash position may force them to stall the fab30 conversion which will put them at a severe disadvantage.

gb

are you looking for historical data? if so amd has all their back quarterly reports for several years on their website.

i don't have the time but it would be interesting...

gb

i'm not so sure that amd can raise money at acceptable terms. my guess is they don't stand a chance of doing financing until they can demonstrate *oems shipping* barcelona; not some press release, but boxes.

they already stated at the q4 analyst meeting that they'd be *negative* $500M cash flow for '07. and i suspect most analysts consider that wildly optimistic at this point.

they only have $1.541B in cash so they'll burn through at least a third of that in '07 even by their own december claims. right now i'm guessing they'll burn through more like half of it unless they kill their capex.

a knuckle a week for juice comes to mind...

gb

was there ever a time in amd's history, pre-fab30, where they were about to enter a downturn with this much debt?

gb

i haven't kept track of the chipset process roadmap but that seems to indicate that the new chipsets for 2h07 will be on 65nm?

gb

so theoretically they could continue to produce current 65nm cpus as the transition to 45nm occurs then start producing 65nm chipsets or graphics.

what's happening to the 90nm fabs now producing chipsets?

gb

although i worked at intel for 18 years i was never privy to amd/intel licensing terms.

i therefore take with big grain of salt any assertion about what others say about those terms. and i doubt intel and amd will issue a joint press release anytime soon clarifying it for us.

i remember a lot of legal wrangling around patent laundry by cyrix and ibm that employed a lot of lawyers.

i certainly wouldn't take amd's side if they chose to pursue this but i wouldn't rule out them trying it.

gb

as intel moves in '08 to 3 45nm fabs i wonder how many 65nm fabs will be left behind?

what will happen to the current 90nm fabs producing chipsets?

as intel's chipset mss continues to grow the addition of discrete graphics will add a new opportunity to fill those fabs along with 65nm fabs left behind.

this should be interesting.

gb

"can't" is an interesting word.

i don't know what they would do with their existing fabs.

what is ti doing with theirs?

gb

i think if the current intel tick-tock 2 year cadence is maintained through the current public roadmap it will put amd as we know it out of business.

i think intel used to worry a bit about what happened to amd but i think that is over.

i think the end game is that amd gets taken private below $10/sh, the buyer dismantles the business and maintains the lawsuit in hopes of a big payout.

finis

gb

by ti path i meant fabless, i.e. turning to foundries.

gb

i'm sure you're right.

still, i see few other options...other than the ti path...

gb

that's what i suspect as well. which is why i said "frantic."

they would need to select library equivalents or design new library equivalents for the blocks in the existing designs. i suspect there are a heck of a lot of late nights ahead.

failing to do that means they pay someone else to build ati's power hogging graphics while amd produces celeron fighters in their own underutilized, debt ridden fabs.

not pretty...

gb

i'm guessing there's a lot of frantic activity going on inside amd to tape out some existing ati design(s) on 65nm for production this year.

amd can't afford to let those 65nm/300mm fabs get delayed given intel's plans to have three 45nm fabs in operation in '08.

nor can they afford to fill them with cpus that compete with the bottom third of intel's product line.

rock and a hard place...

gb

given amd's embarrasing and frankly nervous responses to questions posed i expect a lot of downgrades in the next few weeks. i don't think amd @$17 had this amount of trouble priced in...

gb

i thought henri's response to the question of "what is your sales pitch for 1p servers?" to be embarrasing. i think it was "better product."

laughable...

gb

analysts pretty concerned about negative cashflow impacting cash on balance sheet for full year.

sounds like they're planning on bleeding > 1/3 of cash on hand and analysts think it's going to be considerably worse than that hence requiring additional financing.

gb

i remember seeing some activity on it before retiring from intel in 2000 so they've been working on the serial interconnect for a while.

there's definitely some surprises coming for most folk.

gb