NVIDIA reports Q3 Results, with record 39.1% Gross Margins

Arun

Unknown.
Moderator
Legend
Apparently, this would easily have been an incredibly good quarter, income-wise, had they not suffered from $14.2M non-recurring charges due to the settlement of various litigation matters.

Quick handmade comparative:
Code:
                Q3-05    Q3-04    Q2-05    Q2-04
Revenue         $583M    $516M    $575M    $456M
Gross Profit    $228M    $167M    $218M    $140M
Income          $71M     $29M     $80M     $4M
Net Income      $65M     $26M     $75M     $5M
Per-Share       $0.36    $0.15    $0.41    $0.03
As this table clearly shows, NVIDIA managed higher revenue this quarter than in Q2, as well as $10M higher gross profits. However, their expenses have been rising slightly, and include a one-time settlement cost this quarter, of $14M. NVIDIA's Income would be around $85M without that, their highest since the January 2002 quarter.
Conference call is beggining in 15 mins or so, expect a summary from me within 2 hours or so (and if you don't get that, most likely when I'll wake up instead). Enjoy.


Uttar
 
These are my ENTIRE notes, taken in realtime, with no pausing, of the Q4 Conference Call. Enjoy :)
As for my own personal, much smaller summary:
- Very good market, exceeding the analyst's expectations by a bit.
- Very impressive outlook. Jen Hsun hoping for mid-fourty margins by Q4 next year is amazing.
- Felt very confident, but in a good-kind of way, as in they really knew what they were doing.

Product-wise:
- 6800GS. This is the big news in this conference call IMO. First, they systematically presented it as a $199 product. When an analyst asked if that was intentional when it's currently priced at $249, Jen Hsun responded that it really was the same segment, implying thus they could easily cut the price. Furthermore, he said more partners will sell it soon, just they weren't ready at launch time.
- And, even bigger news perhaps: Jen Hsun confirmed there would NOT be an Integrated Graphics Motherboard for Intel processors, although I distincitvely remember him saying they were aiming for that a few quarters ago. This is because NVIDIA feels Intel is already very nicely covering that market, so it'd be hard to compete (-> focus on improving their position in the AMD market instead, which is 25M integrated motherboard units/year)


- Non-recurring charge of $14.2M for litigation matters
- 70% decline from XBox revenues
- Gross margins increased further
- $19M SM3.0 GPUs shipped

- Considers the AMD Integrated market as a significative growth factor
- Total MCP revenue increased 100% year-over-year.
- More than 90% of the AMD business (could have heard the number wrong?)
- MXM reduced time-to-market

CFO
---
- Made up for the lost XBox revenue, "plus some".
- Core business, year-to-year, is up 40%. 14% qurter-to-quarter.
- GF6/GF7 families represent 80% of the core revenue
- Memory and other revenues were relatively flat.

- No single factor in the improvement.
- Increased margins even though the XBox had high margins and is now gone.
- Ongoing operation costs up $5M, well within guidance.
- Settlement costs are associated with 3DFX and [couldn't hear]
- This is an estimate, but feel confident it is all it will be.

- Tax rate constant at 16%
- $16M of own stock repurchase.
- 73 days of inventory.
- Increase in goodwill due to 3DFX settlement. Also increased liabilities.

- IN THE PROCESS OF ACQUIRING SOFTWARE COMPANIES.

- 5 and 8% increases in Q4 expected.
- Expected growth to continue.
- GPU and MCP to grow as expected.
- Some believe their gross margins have peaked.
- They believe their gross margins will continue to improve.
- And they believe 40% gross margins in Q4 is possible (OMFG)
- Settlement costs are expected to increase $5M in Q4 (?! whut?)


- Near-term growth expectations: increase in discrete & notebook marketshare.
- Completely different position now that ATI announced,
- Expecting to still have the best lineup, for both retail and OEMs.
- Misc. referrals to great GF7 and NF4 reviews.
- "SLI has become a must-have feature for enthusiasts".

- "With the GF7 Family, we have already won the most notebook contracts in our history".
- AMD Integrated Motherboards market is expected to me 25M units next year, they expect a significant grab there.

- More than 7M SLI-capable Graphics Cards shipped, and another number I couldn't h
- 15% of the 3D Cellphone market captured, mostly through Sony-Ericson and Motoral.
- Very high growth expected in those markets.
- Excited about their

- RSX is going to take videorealism to a new level.
- License profts from RSX/PS3 are likely to be "very good".
- Mid-term gross margin expectations for the future in the "MID FOURTIES" (omg)


---
Q&A
---

- Notebook side, design cycle: 7800 a couple of months ago.
- What's the opportunity there before ATI reacts etc.?
Response: Did very good with the back-to-school period, etc.
Tons of opportunities late fall/early spring, expecting to do very well.
By the end of next year, expecting and/or hoping unit share to double in the notebook market.

- Revenue growth per segment in Q4?
- Inventory situation?
Response: High MCP growth expected in Q4 (C51 etc.)
See no reason to be losing GPU marketshare in Q4, so expecting increased revenue.
Very little FX inventory left. Now making transition from GF6 to GF7, monitoring inventory closely.
Right now, thinking they're doing a good job of balancing inventory, and "having enough versus too little".
Not thinking of inventory in terms of days right now, but more from a "response speed" pov.
While not wanting to have TOO much inventory either.
So they do NOT want to have a maximum inventory or anything, they feel having a fair bit of it is key to respond to the market properly.

[...]
Most important thing at the OEMs is to win the market long-term.
Channel business is more about strategic position, importance of brands.
As for developping countries, also an advantage.


Question: Great rebound in handsets in Q3, what about Q4?
Response: No decline in Q4 expected, just not a real growth driver in Q4, will remain flat.
As for growth, expected later.
Follow-up: How long do you expect to get to mid-forties?
Response: We expect continuous prograss. "Not gonna happen in Q4. [laugh] This Q4. [big laugh]"
"Obviously Jen Hsun expects we can do it by this time next year [laugh of CFO]"

Question: Competitiveness is a key part of margins in this business, so how come you feel so confident? Reference to write-off making things easier for them.
Response: "Hum. Well. I guess there was a mistake [laugh]"
I would never write them off at all. Frankly, when a competitor has that much inventory that was written off,
that was the biggest competition you can have. So very pround of having managed to do great against that.
Fundamental change for margins... Well we're not going to share any of it, but we've had a lot of business
and structural changes. Very fundamental changs. And this is what has allowed us such margins. Very long changes,
"our business is fundamentally different than businesses like, you know, design businesses".
Re-engineered almost everything they did to achieve this.
"Almost everything we do, we don't show. Execution is really really important. Look at the last 13 years of history, with the exception of GFFX, they won every generation's OEM contracts".
"There's absolutely no reason why we can't continue to win every single one of them. It's all within our control."

Question: 6800GS. How is it going to bridge the performance-mainstream etc.? Roadmap for it?
Response: "It's really one of the sweetspots of the marketplace". High ASP, and High Volume.
"Can now very confortably sell it at $199" (WTF? Isn't it $249?)
"We now have 2 killer products in this sweetspot of the marketplace" (referring to GT)
"It goes unnoticed from a press perspective, but it's also the most important market segment in our opinion."
"When the time comes for us to transition to GF7, we will, but for now we'll sell the 6800GS"

Follow-up: How would you caracterize the new integrated products on a gross margin POV?
Response: We like the integrated architectures, but they need to be designed from the grounds up for it.
You can't just take a discrete desktop GPU like a 6200TC and expect that to work in integrated.
You need to rearchitecture it completely. We need to aim for lots of features and VERY low cost.
"So we did two things. Through the course of the last 2 years, we built the nForce brand, number 1 core logic brand today
Secondly, we spent a lot of time to re-architecting to be able to really ramp up etc."

Right now, gross margins still lag behind. But opportunity to get them roughly on par.

Question: What about fab capacity? Constrained?
Response: "We would like to be less constrained, but the truth is the whole world is constrained."
"Need to do the best we can with the capacity that we do have."
"I would love to have more, and we'll try to get as much as we can."
[...]

Follow-up: Handset. How many different customers are you shipping to? How many expected?
Response: "Three of the Top 5"
"Working hard to get all of them." - "Always focused on the multimedia-rich segment." "Focus on 3G"
"Without the content, who needs the media processing capabilities that we provide?"
"That focus has been a good one, I think"

Question: Marketshare? Holiday segment?
Response: "It depends on segment. And it depends on product."
"Obviously our market share is higher in the high-end; 7800 but 6800 too. Expecting share in that market to INCREASE."
"SLI is really very important, getting more important."

"In integrated, AGP low-end, and even PCI-E low-end, our marketshare is lower."
"So that's an opportunity for us there." - "Growing quarter-to-quarter there."

Follow-up: Intel's roadmap for the integrated?
Reponse: "We have no plans for Intel Integrated".
Shortage created, but will be fixed soon by Intel
"Well served by Intel", so basically no reason to enter there and compete
(Seriously, WTF?)

Question: ?
Response: It was 33M. Substantial but not a real contribution to EPS yet.

Question: On gross margins, integrated, etc.
Response: We haven't commented on the integrated corelogic margins, and we usually don't.
Usually below corporate average, but OK.
"We should be very very competitive in the marketplace with that product [C51]" - "So that's really the point of it."
"There isn't one bullet that has increased margins." "I would have fired that bullet years ago."

Follow-up: 2006, Channel Market Growth in PC Graphics?
Response: "We serve the GPU market. And we serve the integrated graphics market. And we also serve the core logic market."
"That includes notebooks, workstations, etc."
- Core logic is $10B, GPU is $5B market

"The experience of Vista on better GPUs is just better". So more people will buy [discrete] GPUs.
"I think core logic is going to continue to be very important".
"Not sure how much of that $10B we can go after, hopefully a lot."

Question: Workstation, what growth this quarter? What about next?
Response: Workstation business is getting more and more important, as digital content is getting more important.
So it's the MARKET's that's growing. Implying the Quadro just has a majority in a growing market.
Amazing growth of a lot of different digital content companies, "and where there's digital content being created, there's a Quadro".
Recently introduced SLI for Workstations. Importance of "multiple GPUs driving multiple Displays" in the Workstation market.

Follow-up: PCI-Express influence in the Channel?
Response: Increasing quarter-to-quarter. Making progress, but not quite exactly there yet.
Only real issue now is the pricepoint of PCI-E vs AGP motherboards, but no reason for PCI-E price to be higher so good long-term.

Question: Expense growth?
Response: About 3-4% quarter-to-quarter growth, considered reasonable. "Did a formidable job" (in keeping it moderate)

Question: 6800GS partners?
Response: All of our core partners will be selling the 6800GS. All the ones selling the 6800GT will sell it.
The important thing is not all had it ready at launch time.
$199 vs $249: "We started out a little bit higher, but it's basically the same segment" (-> expect $199 to be possible for 6800GS)

Question: Longer-term new applications?
Response: Personal media players. Not a lot of opportunities with MP3 players etc., but as richer music & games appear,
then portable devices over the next 10 years are going to get really amazing. So that can be a good market.
On the other hand, the very small devices are unlikely to be more important than the server market, which is also very important.
"We have tons of computing expertise".

Question: Handset business has been irregular. How many more do you need "smoother" revenues in that?
As for Sony PS3, obviously that'll increase the gross margins since it's 100% margins, how would margins be without that?
Response: [big laugh regarding handsets, about"letting Jen Hsun answer"]. "The segment of the market we are serving is 3G"
"Every single season we get more and more models, every single season the 3G market expands."
"As we increase the number of models, revenue will increase"
"Not really focused on the irregular bit, focusing more on the actual opportunity". "Devices becoming more important part of our lives."

CFO: "From my side, Sony royalties will definitively add gross margins. Estimating a 2% increase compared to no PS3, so 40% margins would become 42%"
Jen Hsun: "Net Margins depend a lot on R&D. I would love to increase the increase income percentage, and yet I want to increase R&D [so we'll have to compromise]"

Question: Margins higher than competitor, what factors do you think improves your efficiency compared to your competitor?
Response: This is not a single factor. It's a lot of hard work in a lot of different areas.
CFO: "I wouldn't feel as good about it if I didn't feel it was sustainable." "I see at this stage no reason why it wouldn't be sustainable."

Follow-up: Mainstream/Value segments. What's your view of the competitive environment there?
Response: 2.5 years ago or so, we come off from GFFX, so we needed to get off that architecture.
"That part of our history has been pretty well documented, and pretty well [...]" (couldn't hear)
"It was just unfortunate that the PCI-E transition happened around the same time, so we missed almost everything, almost ALL the OEMs"
"The marketshare of notebooks that you see, we've only capture the performance segment, which is why our marketshare is 20%"
"This time around, it's a completely different picture. This IS status quo, this is the way it's supposed to be."
"This is the way this company is ought to performance" (implying GFFX was a
"I'm not surprised we have technology leadership. I'm not surprised we have market leadership."
"We are going to win a lot more of the OEM


Jen Hsun Huang closing remarks: "I appreciate all of you joining us today, and I look forward to let you know about our progress at, hum [short pause and hesitation], our next conference call".
 
No questions about why the 90nm GPUs are late? Why 6800GS is apparently stepping into the breach?...

Also they seemed pretty defensive about inventory.

Jawed
 
Damn, you do that in real time? I feel so inadequate. :LOL: But then this is why I like it when you do it. ;)

I'll listen to it later and add anything if I hear something.

But, yeah, that sounds like "the old NV", pre NV30, feeling cocky in a "this is our right and heritage" kind of way. Y'know, Divine Right of Kings. :LOL:

Which is not to say they don't bust their ass, because obviously they do. . .but it's kind of a New York Yankees outlook, if you get my meaning.
 
Jawed said:
No questions about why the 90nm GPUs are late?
I don't think they're necessarily late... yet

Also, they've got GeForce 6100, 6150 and GeForce Go 7300 shipping on 90nm already.

Why 6800GS is apparently stepping into the breach?...
I'm guessing that it is in relation to wanting to get rid of chips that are still cost effective to produce; try and get ATI to make the move first and then they'll jump in with 7600 series. ATI have got to beat the GS on price/performance first.
 
Response: 2.5 years ago or so, we come off from GFFX, so we needed to get off that architecture.
"That part of our history has been pretty well documented, and pretty well [...]" (couldn't hear)

". . .and. . .uh, heh. . .pretty well described."

Having listened to it now, yeah, very confident and feeling like they've climbed the mountain all the way back to the top. . .where they feel they've always belonged, and usually proven they are correct in that belief. Certainly more often than not.
 
wow 39+% gross margin, makeup of 70 million lost XBOX revenue and 8% quarter on quarter growth. Sounds like a decent quarter
 
There's little doubt that they want to communicate the idea that over the last two years NVidia has totally restructured itself internally for maximum efficiency.

And they're not going to tell us what they've done.

Jawed
 
Jawed said:
There's little doubt that they want to communicate the idea that over the last two years NVidia has totally restructured itself internally for maximum efficiency.

And they're not going to tell us what they've done.

Jawed

And why should they? I seem to recall Orton answering a question last time along the lines of "they can do their own R&D".

I'm not convinced that's all of it tho. Yeah, I know that's fuzzy.

To answer your question above about 90nm, predicting margins going above 40% would not give any hint that they think anything is behind schedule at all. Indeed it suggests they think they are executing to near perfection at this point.

I'll be interested to see RSX and what evidence there is of how much new work they did there above and beyond G7x. I'd also love to know who's making more money per unit on the console gpus.

Frankly, they've kicked ATI's butt on a lot of levels the last year, and much of it was the result of several years of different efforts coming together to reinforce each other (think nforce & SLI as an example). I mean, really, gotta tip the cap.

If they can bring the 90nm mainstream & high-end ship into harbor on time and at their target clocks (whatever those are), I think it is going to be very, very hard for ATI to overtake them overall next year, tho obviously that would leave room for specific victories here and there.

But, hell, first ATI needs to get back to profitability! It's worth remembering that a lot of what we hear in these calls is backwards-looking in nature in the first place, and then there is a definite lag factor anyway when new product comes in. I mean, it took ATI a year or more after R300 came out before they looked like a monster in market share and financials. You don't dig the hole over night, and you don't climb out over night either.

But ATI is still in a better competitive position now, IMO, than they were before R300, so it is still a doable thing to do even without a major (I won't even whisper the name of you-know-what) NV f'up.
 
Last edited by a moderator:
I think ATI began digging the hole Winter/Spring of 2004 when they decided to focus on PCI Express so much, and reached bottom with bringing X850XT (and PE) to market "with good yields" right on time for everyone to say "we don't want that, we want R520" (or the NVidia equivalent).

It didn't help that they broke their spade at Christmas, either :LOL:

I'm seriously optimistic about ATI now - the headline, for me, is that Xenos is a clear pointer to DX10 and ... [thread-derailing stuff deleted].

Jawed
 
I start a little earlier, with Rialto slipping. And then trying to wedge a major unplanned engineering effort --CrossFire-- into an already packed schedule because of Xenos. You called it "domino" before, but I see it more like snowball turning into mid-size avalanche as it picked up speed and momentum coming down the mountain.
 
Looking at how they're emphasizing the profits they'll earn on the 6800GS, I wonder if ATI made a mistake of not going for a 256-bit bus card in this segment. I wonder if we'll see 12-pipe versions of the X1800XL soon. They've been binning this way since R300, so I don't think they'll stop now.

For comparison, note that ATI's X1800XL has the same amount of memory, same bus width, and same memory speed, but it sells for twice the price! I don't think the die size is much more than 35% bigger (has anyone measured NV42?), though I'm sure 90nm costs significantly more for the same area.

If ATI paired their new memory controller and FP blending with R430, they'd have a killer product at a very reasonable price. I'm really thinking all the R5xx technology just cost too much die space. Fast dynamic branching is cool, but a more traditional 32-pipe card would probably have easily fit into that many transistors, judging by R480. I'm thinking that 3 arithmetic units per ROP will change my mind, though. A card about equal to RV530 times 2 is badly needed (RV560?).
 
Fodder said:
There's no excuse for new products without SM3.0 at this point in time.
I think there is: performance/$. Since its introduction, NV40 has a feature advantage over R520 for one reason only: FP blending. There is little in PS3.0 that lets you practically get more advanced graphics save for two features: Dynamic branching, and unlimited dependent texture reads. The former will take a while for use in games (SC:CT claims to use it, though I'm skeptical), and is quite slow on NV40/G70. The latter hasn't been used in a game yet either, and even so I'm sure it could be accomplished without a radical pipeline redesign.

Anyway, I didn't necessarily mean PS2.0. I meant similar in architecture so that it wouldn't need double the transistors for the same number of pipes.

The only reason ATI isn't paying dearly for their decision to make a "real" PS3.0 part (i.e. very usable dynamic branching) is that their competitor isn't on 90nm yet. I applaud ATI for their decision because it does advance the field, but it'll cost them.
 
Mintmaster said:
Since its introduction, NV40 has a feature advantage for one reason only: FP blending.
Yet between Far Cry's HDR labelling and the Power of 3 marketing blitz, an awful lot of people equate that advantage with SM3.0. NVIDIA's spinners would have a field day with fresh SM2.0 competition. A traditional 16-pipe part on 90nm with basic (eg. NV40-level) SM3.0 might have done down well, but I think ATI are looking further ahead.
 
Back
Top