NVIDIA CC at Citigroup conference

[Arun]

NVIDIA just had a public CC from the Citigroup Technology Conference, and CFO Marvin D. Burkett was the one answering questions so it had an obvious financial focus. Here's a summary and the usual log - I did my best to try to give correct quotes, but that's often not quite the case, and there is mild to heavy paraphrasing sometimes. I claim no responsibility for the accuracy of my live log.

Summary
- "There wasn't a shortage of capacity - we just didn't order enough wafers, primarily on 90nm."
- PortalPlayer: "From a people standpoint, the acquisition went very very well. They're still there and very happy."
- CPU-GPU integration: "We don't have a x86 license, but it is an interesting concept."
- Still generating tons of cash, will likely exit the year with $2B+...

Log

- Introduction

- Demand: "So far it's just a continuation of what we've seen in Q2".
- Guidance/Inventory:
"Q2 was so strong that you had to wonder how much of Q3 was in Q2".
"Q3 is usually up double digits while Q2 is down, but this time Q2 was up double digits!"
"Because Q2 was so strong, inventory was down to 47 days because we didn't expect that."
"The issues regarding inventory had nothing to do with the foundries - we just didn't order enough wafer."
"We quickly remedied that in May and June, but because of cycle times, that'll only help in September and October."
"In the back-end, issues have been resolved. Assembly etc. is tight but will not be a problem."
"On the fab side, TSMC has been very responsive. They're great partners."
"They have been able to reduce the cycle time [one week?] as much as they could."
"I am not here to change the guidance, we are only 3 weeks in the quarter. But so far so good."
"There wasn't a shortage of capacity - we just didn't order enough wafers, primarily on 90nm."

- Share gain vs PC Growth
"Significant share gains in notebooks/Santa Rosa."
"You saw that in Q2 and you'll see significant further growth in Q3 there."
"The surprising part in Q2 was that the whole PC desktop business grew substantially."
"Tons of possible factors, from the dollar to CPU and memory prices, to channel inventory, etc."
"So far, continuation of that strength in Q3."

- Moving towards discrete?
"The notebook market is spending more on discrete GPUs."
"In the desktop market, CPUs and memory cost less..."
"So for a given pricepoint, you can invest more in GPUs as a differentiating factor."

- Percentage of IGP vs Discrete?
"In the last year or two, the gap has widened."
"So that depends on how fast the gap reduces again."

- CPU-GPU integration?
"I call that a Fusion Product."
"What's interesting is that the graphics is bigger than the CPU."
"From our standpoint, that's interesting."
"We don't have a x86 license, but it is an interesting concept."
"From an economic POV, CPU/GPU/Chipset have very different revenue per wafer."
"So what do you sell your single-chip for?"
"[single-chip] Fusion Products are probably a good option for the low-end."

- Expanding to GPGPU
[insert Tesla marketing from Marv]
"Intel has a part which they will name Larrabee."
"I understand it will do the same thing."
"GPUs are very good at this kind of market [so that's normal.]"

- Marketshare
[too much unhearable, sorry - but basically said OEMs weren't fearing a GPU monopoly]

- PortalPlayer acquisition
"Two justifications: Application Processor, and specific Low-Power technology."
"From a people standpoint, the acquisition went very very well. They're still there and very happy."
"They've been working since January as a single team [for a product by the end of the year.]"

- Where are you NOW for the next product?
"It's in the last stages of design. Doing a very good job."
"The PortalPlayer acquisition allows us to go after the [iPhone[-like] part of the market.]"

- Handheld margins?
"Everybody thinks that when we acquire somebody, their margins will go up *snickers*."
"40%+ margins ala Portalplayer was dictated by what the market is willing to pay and competion..."
"We believe that will probably about remain the case."
"I don't know what the competition will have in that timeframe, so it will depend."
"Lot more competition in that marketplace than in the GPU marketplace."
"If you have better low-power than the other guy, that's a great start."
"And the next question is performance, so we need better performance."
[marketing about application processor and NVIDIA media expertise etc.]

- Revenue per wafer?
"Margins is directly proportional to revenue per wafer."
"Each generation has better reconfigurability etc."

- Channel/OEM patterns and inventory?
"I think the channel went out of Q1 with a very low inventory."
"OEM patterns are not unusual[, but they have been good for notebooks.]"
"We think the inventory the inventory in the channel is [still] lean."

- OpEX growth?
"We're disciplined to spending in that 23-25% [OpEX range]."
"It's sufficient."

- Backlog?
"The normal GPU business, you have no backlog."
"Different with OEMs, you do have a decent backlog."
"This situation is slightly different right now, we also have a backlog in the channel."
"We are probably in an unique backlog situation for a GPU supplier."

- You're building a lot of cash. What do you want to do?
"Don't give it away."
"I think cash will increase again this quarter, and Q4..."
"I expect to go above $2B by the end of the year."
"The board is continually evaluating [stock repurchasing]."
"We still want to acquire companies for cash."
"So beyond that, I have no other comment."

 

[Geo]

$2B is a lot to keep around. Maybe we'll see them go to dividends.

It appears his capacity issues comments were regarding the drop in inventory for 2Q. I'm not sure they change their previous comments about seeing the potential for capacity constraints in 2H.

 

[Arun]

$2B is a lot to keep around. Maybe we'll see them go to dividends.

Heheh. Well, that'd be midly ironic for a high-growth tech stock indeed! I think large acquisitions are in order... Potential targets I can think of include:
- VIA, for its x86 expertise as well as all the other stuff they have.
- SiS, for the same reasons minus x86 (so much less attractive)
- Companies specializing in wireless, especially Wireless USB or Bluetooth/FM.
- Companies specializing in wideband wireless, although that depends on the direction they want to take.

It appears his capacity issues comments were regarding the drop in inventory for 2Q. I'm not sure they change their previous comments about seeing the potential for capacity constraints in 2H.

I am indeed not sure it changes much for Q4, but he did seem to imply that there shouldn't be any real problem in Q3 AFAICT.

 

[Geo]

Wireless would indeed be interesting, but really starts to broaden their tech footprint well outside their traditional areas of expertise, it seems to me. At least if its a major play.

Of course many of us have been expecting VIA and SiS to fall for the better part of two years now, and I still think the day is coming near. . .but I'm less sure NV would be willing to plop down large sums for either of them

 

[AlexV]

Heheh. Well, that'd be midly ironic for a high-growth tech stock indeed! I think large acquisitions are in order... Potential targets I can think of include:
- VIA, for its x86 expertise as well as all the other stuff they have.
- SiS, for the same reasons minus x86 (so much less attractive)
- Companies specializing in wireless, especially Wireless USB or Bluetooth/FM.
- Companies specializing in wideband wireless, although that depends on the direction they want to take.

I am indeed not sure it changes much for Q4, but he did seem to imply that there shouldn't be any real problem in Q3 AFAICT.

AMD is missing from that list(I'm joking)

 

[Geo]

That's only ~3Q of losses for AMD at current burn rates!

 

[Sound_Card]

I would like to add Texas Intruments to that list.

 

[AlphaWolf]

I would like to add Texas Intruments to that list.

Sure, Microsoft and apple too right?

$2 billion in cash is a far cry from the 50+ billion it would cost to acquire TI if you were at all serious.

 

[Sound_Card]

Sure, Microsoft and apple too right?

$2 billion in cash is a far cry from the 50+ billion it would cost to acquire TI if you were at all serious.

And General motors.

No, really, I checked Yahoo, I did not think they were worth 50 bill. (hit head on rock).

 

[Davros]

$2B is a lot to keep around. Maybe we'll see them go to dividends.

Or maybe they'll do something crazy like lower prices

 

[Geo]

You forgot your smilie!

 

[dizietsma]

Anyone want to guess on the statement :-

- Where are you NOW for the next product?
"It's in the last stages of design. Doing a very good job."


??

 

[Mortimer]

That was about the new portalplayer stuff I think.

 

[Arun]

Wireless would indeed be interesting, but really starts to broaden their tech footprint well outside their traditional areas of expertise, it seems to me. At least if its a major play.

Well, with Wireless USB, you could see that as a long-term investment in making sure their chipset business doesn't become a commodity. But then they'd need to integrate RF in their single-chip solutions, which would be rather unusual for the PC market to say the least!

But still, getting WiFi 802.11n/Bluetooth 2.1/Wireless USB integrated on the (non-low-end) chipset would certainly be very interesting. That doesn't mean they want to do such a thing for either financial or engineering reasons, but it's worth pondering upon IMO. Of course, if they only want the baseband, they could always design that themselves.

And then there's the fact that if they want to have that functionality in GoForces too, they need that IP too (although in a very low-power-variety!) - but then again, I think they don't want to do that anymore. PortalPlayer did, but AFAICT, they decided to change strategy and not integrate wireless. Ugh. This kinda reminds me of how MediaQ started working on a basic application processor before being acquired by NVIDIA, and then they decided to focus exclusively on GPUs post-acquisition...

Of course many of us have been expecting VIA and SiS to fall for the better part of two years now, and I still think the day is coming near. . .but I'm less sure NV would be willing to plop down large sums for either of them

Well, they both have substantial reserves. So they won't *die* overnight, but as it is, they're in a bad position and I can't see it getting much better. SiS is actually recovering slightly so I wonder if they could reach break-even soon and actually do midly OK on their own once Mirage 4 is out... But while SiS improved in August, VIA's sales fell slightly. heh.

http://www.via.com.tw/en/resources/pressroom/pressrelease.jsp?press_release_no=1507
http://www.via.com.tw/en/resources/pressroom/pressrelease.jsp?press_release_no=1508

Probably a good way to consider their situation is to look at Shareholder's Equity in the balance sheet. They can sustain themselves for quite a while, but not forever - so I think it is definitely in their shareholders' interests to sell off.

The question really is whether NVIDIA thinks they're worth ~$1B in cash... And I'm not quite sure about that. Their balance sheet pretty much puts that at a $300M discount, but $700M is still a lot for a company that keeps losing money. Maybe if the net cost lowered a bit more, it would become a lot more attractive. Perhaps once their Intel license expires completely and they lose further share in the AMD market...

 

[INKster]

The question really is whether NVIDIA thinks they're worth ~$1B in cash... And I'm not quite sure about that. Their balance sheet pretty much puts that at a $300M discount, but $700M is still a lot for a company that keeps losing money. Maybe if the net cost lowered a bit more, it would become a lot more attractive. Perhaps once their Intel license expires completely and they lose further share in the AMD market...

A VIA without a x86 license, that's been "hinting" at a discrete exit from the desktop PC chipset market, that's been bleeding money every quarter, etc, doesn't strike me as a wise buy-out target anymore...
S3's tech isn't exactly earth-shattering, just like the dedicated VIA-branded micro-controller stuff (IEEE1394, networking, sound, etc).


SiS, on the other hand, looks increasingly interesting.
They've won that sweet Intel D201GLY deal, they still do the southbridge for the Xbox 360, they've landed plenty of low-end OEM desktop and wireless chipset orders and finally, they don't bleed nearly as much money as VIA's.
Oh, and they would probably be a cheaper purchase.

 

[Arun]

A VIA without a x86 license

The x86 license dynamics probably aren't what you think they are. What NVIDIA would want is the CN design and the engineers, not the license. They can get that anyway if they need it - Intel can't force companies out of the market if it's obvious they have the money to compete. There are antitrust laws at play here, most likely...

that's been "hinting" at a discrete exit from the desktop PC chipset market

How is that negative? Then they would fire part of their chipset team, and that'd only make the acquisition more attractive since NVIDIA doesn't need all these guys.

that's been bleeding money every quarter, etc, doesn't strike me as a wise buy-out target anymore...

NVIDIA nearly always buys non-profitable companies, or at least companies not making much money relative to their size. I doubt that's a factor here - that'd really just be part of the price of the acquisition. Which, indeed, needs to be considered when pondering how much NVIDIA would be willing to shell out.

S3's tech isn't exactly earth-shattering, just like the dedicated VIA-branded micro-controller stuff (IEEE1394, networking, sound, etc).

Obviously, but that's still engineers and IP that can't hurt. It is indeed interesting that VIA makes their own analog chips for their motherboards. I can't help but ponder if NVIDIA would also want to do that - my guess is no, unless they could integrate it into fewer chips than the competition. Which would be interesting - for example, they could integrate Gbit Ethernet and sound into a single analog chip. But for that, they need enough engineers, and I'm not sure how many VIA has left in these areas - I wouldn't be surprised if they were mostly still selling existing products.

SiS, on the other hand, looks increasingly interesting.
They've won that sweet Intel D201GLY deal, they still do the southbridge for the Xbox 360, they've landed plenty of low-end OEM desktop and wireless chipset orders and finally, they don't bleed nearly as much money as VIA's.
Oh, and they would probably be a cheaper purchase.

SiS is a way worse acquisition target than VIA imo. NVIDIA's strategy never has been to buy companies for existing deals. And in terms of engineers, how is SiS any better than VIA? They're also much less diversified, so they'd get less IP out of the deal.

It is true that SiS would be cheaper, but unlike VIA, I don't really get what the point of the deal would be besides engineers, existing contracts and getting rid of competition. Now, if you want to argue that the latter point is important, you might as well claim that NVIDIA should buy both VIA and SiS.

And given NV's cash position, that's not even completely unreasonable. But if that was the goal, then we're probably not thinking of an acquisition happening relatively soon, but rather in 2H08 or even later. I do agree that SiS is a possibility in that timeframe: NV must be hoping MCP73 and, to a lesser extend, MCP78/MCP79 will have squeezed them a bit more by then. But NV+VIA+SiS is a truckload of chipset engineers, so they sure as hell better have a good justification if they want to do that rather than just getting rid of competition!

 

[Davros]

But then they'd need to integrate RF in their single-chip solutions, which would be rather unusual for the PC market to say the least!

My intel mboard chipset has wireless lan

The x86 license dynamics probably aren't what you think they are. What NVIDIA would want is the CN design and the engineers, not the license. They can get that anyway if they need it - Intel can't force companies out of the market if it's obvious they have the money to compete. There are antitrust laws at play here, most likely...

1. did a company once make a intel chipset without a licence ?
2. is it the case that both ati + nvidia's drivers prevent corssfire + sli running on certain chipsets ?
where are the anti-trust laws in this case? why are there no cf + sli capable boards ?

 

[Arun]

My intel mboard chipset has wireless lan

But that's a discrete chip.

1. did a company once make a intel chipset without a licence ?

How is that related to a x86 license?

where are the anti-trust laws in this case? why are there no cf + sli capable boards ?

Neither AMD nor NV is a near-monopoly right now, so I'm not sure how this is comparable to a company preventing others from competing in an entire multi-billion-dollars industry.

 

[Maintank]

If they were to invest. I'd think it would be in a company specializing in wireless broadband to the home or mobile devices. Hasnt PC growth been pretty stagnant the last 4-8 quarters? The growth is in laptops at the expense of desktops?

 

[Geo]

Well, with Wireless USB, you could see that as a long-term investment in making sure their chipset business doesn't become a commodity. But then they'd need to integrate RF in their single-chip solutions, which would be rather unusual for the PC market to say the least!

But still, getting WiFi 802.11n/Bluetooth 2.1/Wireless USB integrated on the (non-low-end) chipset would certainly be very interesting. That doesn't mean they want to do such a thing for either financial or engineering reasons, but it's worth pondering upon IMO. Of course, if they only want the baseband, they could always design that themselves.

Actually, I've very much been a proponent of getting 802.11n (and before that, 11g) wi-fi built-in on desktop mainboards. It should be as common as built-in ethernet, in my opinion.

My new lappie has Wireless USB. . . I don't think there's anything out there to use it yet, but it does look very interesting for wireless PAN at high bw.