Whoa! Mot *should* sell their PPC division
I just read this article posted at Yahoo! Remember Iridium, Motorola's $5B mistake? Well, it was salvaged from bankruptcy court at the last minute, and is now run by new owners at 1/10th — one tenth — the cost that it incurred under Motorola's so-called management. They've launched 7 more satellites as backups, and they're on target to turn a profit by the end of the year. That's right: Iridium is about to become profitable!
Their clientele? The US Army, and shortly the British Army. The system has global reach and excellent security, and so the Army has not shied from the handset costs ($1500, half what they used to be) nor the calling costs ($1.50 a minute, down from $7). In fact, since the Army has banned less secure types of cell and satellite communications from the Iraq theatre, a number of broadcasters are becoming customers too.
The writer wonders what similar investors could do with Qwest and Global Crossing. I'm wondering what they could do with Mot Semiconductor.
Their clientele? The US Army, and shortly the British Army. The system has global reach and excellent security, and so the Army has not shied from the handset costs ($1500, half what they used to be) nor the calling costs ($1.50 a minute, down from $7). In fact, since the Army has banned less secure types of cell and satellite communications from the Iraq theatre, a number of broadcasters are becoming customers too.
The writer wonders what similar investors could do with Qwest and Global Crossing. I'm wondering what they could do with Mot Semiconductor.
Comments
As far as the semiconductor division, Moto doesn't really have choice. This will probably sell their fabs, at least, and outsource because they can't afford the investment needed for succeeding next-gen fab tech. It's good that Apple finally saw the light, assuming they are going with the 970.
One of the guys is still holding on to a load of Moto shares he bought into at a 15% discount, it's gonna be a while before he gets his money back on those. Poor sap, believed in the company line when he first started, and unlike an exec, actually had to put his own money into those shares.
Originally posted by THT
This is really unfair to Motorola. This new company would have never had gotten the satellites off the ground because they never had to pay for the huge R&D and technology development to get the Iridium constellation into orbit. All they had to do is recover their $25M investment. It'll take this company decades to recover the $5G that was needed to put it into orbit and to develop the phones.
That much might not be fair to Mot, since they have (or had) a lot of money and a lot of talented engineers. What I was focusing on was the management of the company. In other words, start with the infrastructure in place: Mot was still bleeding money just to run the company, and so their handset and service prices were outrageous, and so no-one bought in. The new guys are running a much leaner operation, so the service is becoming profitable and the handset prices are beginning (I emphasize beginning) to drop to something mainstream. And they're going to get 6 more years of life out of that $5B investment than Mot expected.
"Profitable" does not mean "debt free." It means that the company is making more than it's spending. The new Iridium would be able to start chipping away at that $5B by the end of this year. That's a better fate than it would have had under Mot.
So, while it's a given that the current owners could never have gotten the system off the ground for $25 million, the fact is that they're managing the company much, much better than Motorola did.
As far as the semiconductor division, Moto doesn't really have choice. This will probably sell their fabs, at least, and outsource because they can't afford the investment needed for succeeding next-gen fab tech. It's good that Apple finally saw the light, assuming they are going with the 970.
That's why I was wondering if someone else couldn't do a much better job running Mot Semi. The engineers are there; the fabs are there; the R&D labs are there. Great things have come out of Somerset. I'd really like to know how much of SPS' recent $1B loss was just inefficiency resulting from summary managerial incompetence.
Originally posted by Amorph
What I was focusing on was the management of the company. In other words, start with the infrastructure in place: Mot was still bleeding money just to run the company, and so their handset and service prices were outrageous, and so no-one bought in. The new guys are running a much leaner operation, so the service is becoming profitable and the handset prices are beginning (I emphasize beginning) to drop to something mainstream. And they're going to get 6 more years of life out of that $5B investment than Mot expected.
I still don't think it is fair. Yes, I think Motorola mismanaged Iridium. Yes, I think this new company is doing a better job running the business. But I think it isn't a fair assessment given the difficulties of putting up a 66 satellite constellation into orbit. I think it is more complex than the semi business going to 0.9 micron tech for instance. I'm giving them a fairer shake because the $5G needed to get off the ground and the chaos involved in $5G of business will bleed into Iridium operations.
If this company had to do the same thing, I don't think their current "profittable" ops business will be running as well. Mayhap, the company is also very lucky since they had 2 straight years of good business conditions (Afghanistan and Iraq) while when Motorola had control, all they saw was peace and only cell phone tech growing.
I'd agree that part of the problem was Motorola not figuring out that satellite phone costs had to be within reach of cell phone costs for them to get revenue. They should have known if they couldn't get there, they shouldn't have started in the first place.
That's why I was wondering if someone else couldn't do a much better job running Mot Semi. The engineers are there; the fabs are there; the R&D labs are there. Great things have come out of Somerset. I'd really like to know how much of SPS' recent $1B loss was just inefficiency resulting from summary managerial incompetence.
Oh yeah. Someone could run Moto Semi much better. It's been said time and time and again that Moto's fabs are dirty which plays into them being late to 0.18 micron, not reaching 0.13 micron yet, and having poor yields on high clock rate chips. A competent management chain would never let that happen. The problem is that a 0.09 micron fab will need $5G of funding. Moto's market simply isn't there to support such an investment anymore. Maybe they could make the 0.13 micron fab profittable enough to fund 0.09, but suffice it to say, being late in fab in the semi business means death. A different company taking over Moto Semi would have to face the problem of getting $5G for HiP 8... so they face a more daunting problem than the Iridium people did.
Originally posted by THT
I'd agree that part of the problem was Motorola not figuring out that satellite phone costs had to be within reach of cell phone costs for them to get revenue. They should have known if they couldn't get there, they shouldn't have started in the first place.
Well, that levels the playing field, doesn't it? Part of the problem with Mot's management was making the commitment in the first place, and the new company shows how well they'd have done if they hadn't!
(Yes, I'm being facetious.)
Oh yeah. Someone could run Moto Semi much better. It's been said time and time and again that Moto's fabs are dirty which plays into them being late to 0.18 micron, not reaching 0.13 micron yet, and having poor yields on high clock rate chips. A competent management chain would never let that happen. The problem is that a 0.09 micron fab will need $5G of funding. Moto's market simply isn't there to support such an investment anymore. Maybe they could make the 0.13 micron fab profittable enough to fund 0.09, but suffice it to say, being late in fab in the semi business means death. A different company taking over Moto Semi would have to face the problem of getting $5G for HiP 8... so they face a more daunting problem than the Iridium people did.
Actually, they've already partnered for fab tech at .065 microns and down. That's what their deal with STM got them. So I don't really see that as a problem. They are already getting someone else to front a few billion for a fab.
The only issue, as I see it, is whether they'll diddle around in the sort of expensive, poorly organized muddle that's characterized them for too many years now, or whether they'll buckle down and make hay while the sun's shining.
Originally posted by Amorph
Well, that levels the playing field, doesn't it? Part of the problem with Mot's management was making the commitment in the first place, and the new company shows how well they'd have done if they hadn't!
(Yes, I'm being facetious.)
I don't understand what you mean by leveling the playing field...
Actually, they've already partnered for fab tech at .065 microns and down. That's what their deal with STM got them. So I don't really see that as a problem. They are already getting someone else to front a few billion for a fab.
Everyone seems to be reminding me of this.
Sort of funny how I advocate Apple to bring PPC design inhouse and outsource fab to someone else. However, I also advocate Apple going with IBM or Intel for fab.
The only issue, as I see it, is whether they'll diddle around in the sort of expensive, poorly organized muddle that's characterized them for too many years now, or whether they'll buckle down and make hay while the sun's shining.
2 years? It's been muddled since HiP5 circa Fall 1999 when they couldn't ship 500 MHz 7400s. Hehe, if the management structure stays the same, they really can't do anything about changing it. Many many bureaucracies grow old because the management structure becomes too like-minded. It's a natural tendency at those pay grades to only promote those who think like them. Comforting.
So, yeah, your idea of Moto selling off their PPC division could be a great boon, as long as none of the management structure follows it. But if they have fabs, I don't think they have much chance.
Originally posted by THT
I don't understand what you mean by leveling the playing field...
The new owners didn't have to shoulder the $5B investment, but the old owners shouldn't have made the investment in the first place.
Like I said, I was being silly.
Everyone seems to be reminding me of this.
They're not outsourcing, though. They have an investment in the plant, their own R&D will be used (and further developed), etc. It's an alliance with STM. They're only outsourcing fabrication for products made on older fab tech, which the likes of TSMC can handle easily.
It's a good way for them to keep current while they're trying to turn themselves around. They still own, or partly own, their most advanced fabs.
Sort of funny how I advocate Apple to bring PPC design inhouse and outsource fab to someone else. However, I also advocate Apple going with IBM or Intel for fab.
Maybe Apple should buy Mot SPS. Seriously. Shut the fabs down and turn them into a design house. Keep Somerset as a technological testbed. There's some real talent there for the taking.
IBM probably wouldn't like that much, now that I think about it.
So, yeah, your idea of Moto selling off their PPC division could be a great boon, as long as none of the management structure follows it. But if they have fabs, I don't think they have much chance.
Their alliance with STM is supposed to cover them through .09 and .065, and I've heard that, for some products, they'll skip .13 altogether and go to .09, since the engineering issues are largely the same and .13 is yesterday's news. If that's the case, they could surprise us. It might also explain a postponement of the 7457...
But then, I'm being optimistic here. This is Motorola. I know they're trying to turn around, I know they're succeeding, but still. Institutional inertia is unbelievably difficult to over come.
Originally posted by Amorph
They're not outsourcing, though. They have an investment in the plant, their own R&D will be used (and further developed), etc. It's an alliance with STM. They're only outsourcing fabrication for products made on older fab tech, which the likes of TSMC can handle easily.
It's a good way for them to keep current while they're trying to turn themselves around. They still own, or partly own, their most advanced fabs.
They don't have any advanced fabs.
Considering the state of their management and their fabs, I think the eventuality is that Moto will sell their fabs. Look at the previous alliance with AMD. AMD was presumably the small fish in that alliance since it looked like they were in it for SOI and 0.13 micron help. Both AMD and Moto had HiP 7 fabs in development, yet, AMD started shipping last Spring (6 months behind Intel) while Moto has yet to produce. Moto supposedly qualified HiP 7 in July of 2001, nearly two years, and still haven't shipped!
The moral of the story here is that Moto is still responsible for its own fabs in their new alliance, and I don't think they'll catch up at all. I'd love to be wrong, but I think your subject title is right. Moto will probably sell PPC fabs and become a design and IP house.
Maybe Apple should buy Mot SPS. Seriously. Shut the fabs down and turn them into a design house. Keep Somerset as a technological testbed. There's some real talent there for the taking.
IBM probably wouldn't like that much, now that I think about it.
Buying Somerset is a good idea. I don't think IBM will have a problem as long as Apple shares IP and uses their fabs.
Their alliance with STM is supposed to cover them through .09 and .065, and I've heard that, for some products, they'll skip .13 altogether and go to .09, since the engineering issues are largely the same and .13 is yesterday's news. If that's the case, they could surprise us. It might also explain a postponement of the 7457...
But then, I'm being optimistic here. This is Motorola. I know they're trying to turn around, I know they're succeeding, but still. Institutional inertia is unbelievably difficult to over come.
I don't know where get they are succeeding? The most important products they have are the 8540 and the 8560, both HiP 7 parts and both have yet to ship. From the 7457 technical briefs, it's also a HiP 7 part. Maybe the recent $400M injection in Moto Semi was to get HiP 7 product shipping, but it's simply won't make a dent into getting 90 and 65 nm fabs up and running.
Originally posted by THT
They don't have any advanced fabs.
I did say their most advanced fabs.
I know they don't have anything better than .13 micron.
The moral of the story here is that Moto is still responsible for its own fabs in their new alliance, and I don't think they'll catch up at all. I'd love to be wrong, but I think your subject title is right. Moto will probably sell PPC fabs and become a design and IP house.
The new alliance is different from the old alliance. AMD was interested in Mot's process tech, just as they're now interested in IBM's. Mot was also interested in some of AMD's. That was the extent of things.
The alliance with STM is resulting in an actual fab being built for both STM and Mot to use (and one other - Philips Semiconductor?). The three are explicitly pooling their resources to build a next-gen fab, and provide the research to keep it going. It's a lot closer than the AMD/Mot alliance was, and the fab is nowhere near Austin.
So Mot is not dependent on their own fabs. Really, there would be no reason for them to get involved in this operation if they were. The problem is not that they don't have the engineering capability to design and build parts at .09, or .065. The problem is that they don't have the fabs to build those parts in any quantity. Their main interest here is to get a clean fab for much less than it would cost them to build their own, with the tradeoff that they have to share it with two other companies. Still, it has to look attractive to them where they are now: If they can get anything rolling on 90 nanometer any time soon, they can catch up to the industry, (hopefully) make a profit, and then begin rebuilding themselves. They might stick with this partnership, too, if it turns out to be in everyone's interests to.
Buying Somerset is a good idea. I don't think IBM will have a problem as long as Apple shares IP and uses their fabs.
Of course, Apple buying Somerset would pretty much be a stake in the heart of Mot SPS.
I don't know where get they are succeeding?
This is Mot in general, not Mot SPS in particular. The information I have came on an article on the shareholder-mandated management shakeout at Mot (basically, the shareholders threatened to throw out the CEO, the board, and upper management if they didn't start fixing the company now). The result looks a lot like what happened to Apple's product line after Jobs: They hugely reduced the number of boards in their various cell phone models (there used to be a different board for each one), they've gotten rid of a lot of excess management layers, and otherwise tried to turn the ship around. I figure if they keep the pressure on they'll be back on target in two or three years. Apparently, given Eskimo's account of AMD's work with Mot, they still have quite a ways to go.
The most important products they have are the 8540 and the 8560, both HiP 7 parts and both have yet to ship. From the 7457 technical briefs, it's also a HiP 7 part. Maybe the recent $400M injection in Moto Semi was to get HiP 7 product shipping, but it's simply won't make a dent into getting 90 and 65 nm fabs up and running.
One thing at a time.
Originally posted by Amorph
[B]I did say their most advanced fabs.
I know they don't have anything better than .13 micron.
They don't have HiP 7 working just yet.
I won't consider HiP 7 a fab until they actually start shipping some important products on it.
The problem is that they don't have the fabs to build those parts in any quantity. Their main interest here is to get a clean fab for much less than it would cost them to build their own, with the tradeoff that they have to share it with two other companies.
So, it's semi-outsourcing?
Still, it has to look attractive to them where they are now: If they can get anything rolling on 90 nanometer any time soon, they can catch up to the industry, (hopefully) make a profit, and then begin rebuilding themselves. They might stick with this partnership, too, if it turns out to be in everyone's interests to.
I would hope that they can get things rolling too. But in this business, falling behind is usually a death sentence. Even when they knew they were behind, they couldn't even put in the work to make the 7457 bus a DDR bus.
The alliance would typically mean that Moto has to have a partner to develop next-gen fabs. They are essentially stuck with a situation like this because succeeding fabs always cost twice as much as the previous one to build.
One thing at a time.
Hmm... so the $400M was to get HiP 7 working? Their most important processors, the 7457, 8540 and 8560, are waiting on it. Even RapidIO is waiting on the 8540 and 8560 to ship.
The 90/65 nm fabs online? I'm hard pressed to believe that, unless you mean online as "prototype". If they are good, maybe they'll be shipping product in Summer of 2004.
Originally posted by THT
They don't have HiP 7 working just yet.
I won't consider HiP 7 a fab until they actually start shipping some important products on it.
I can see that. Heh.
At any rate, they still own their fab-to-be, and they've not given any signs that they're going to get rid of it.
So, it's semi-outsourcing?
Not really.
I would hope that they can get things rolling too. But in this business, falling behind is usually a death sentence. Even when they knew they were behind, they couldn't even put in the work to make the 7457 bus a DDR bus.
Their main customers are apparently smitten with MaxBus as it is, so why would they? They've been making noises about getting out of the desktop-performance CPU business for a while now. They missed the DDR bus, so to speak, so I can see why they'd rather just jump to RapidIO.
Hmm... so the $400M was to get HiP 7 working? Their most important processors, the 7457, 8540 and 8560, are waiting on it. Even RapidIO is waiting on the 8540 and 8560 to ship.
That would be my guess, since as you point out it's nowhere near enough money to build a fab. $400M to clean out MOS13 and pay the power bills to keep it clean long enough to do a production run and make some money.
The 90/65 nm fabs online? I'm hard pressed to believe that, unless you mean online as "prototype". If they are good, maybe they'll be shipping product in Summer of 2004.
They're not going to be making anything at 65nm yet. Like Fishkill, the plant's geared up to move down to that process eventually. They were planning to have 90nm product available soon, though. Mot wanted at one point to skip 130nm altogether and beat the industry to 90.
I wish them luck, but this is Motorola. I'm not sure how capable STM and Philips are of making up for their shortcomings.
Originally posted by Amorph
Their main customers are apparently smitten with MaxBus as it is, so why would they? They've been making noises about getting out of the desktop-performance CPU business for a while now. They missed the DDR bus, so to speak, so I can see why they'd rather just jump to RapidIO.
I don't buy the line that Moto's customers are so smitten with MaxBus that they don't need to change it. You can make a DDR MaxBus backward compatble to a SDR MaxBus, just like they made the G4 support the 60x PPC bus. On top of that it's performance limiting, expecially on AltiVec ops. Implementing a SDR/DDR MaxBus would have minimal changes, if not identical, to G4 packaging and footprint. It is evolutionary and the cheapest way for faster performance for Moto's embedded customers. It's not a decision about desktop-CPU performance; Moto will implement on-chip DDR mem controllers afterall.
The only reason I see Moto not improving MaxBus is because they see it as a competitor to RapidIO, which is what you say, I'm just a lot more cynical about it. The 8540/8560 architecture is a sea change for all of their embedded customers. Brand new boards and supporting ASICs. Nearly everything will be brand new. A SDR/DDR MaxBus represents a drop-in replacement for existing systems, and minimal changes in supporting ASICs for those who want to use DDR signaling.
It's just really bad that they had a 1 to 2 year window to have DDR Maxbus in play before transitioning to RapidIO. They were probably thinking they could move to RapidIO early last year, now it looks like 2H this year. So maybe they thought there wouldn't be time to have a DDR MaxBus before it would be phased out, but it turns out they had plenty of time...