Nvidia’s recent earnings shed some serious light regarding questions I’ve had with respect to the narrative around the stock, and how management has been running the business throughout the crypto craze. What I have discovered is quite interesting, and raises some serious issues which gamers, miners, investors in GPU manufacturers, and cryptocurrency developers need to start tackling head on. In one respect, and Nvidia bulls should love this, as management has executed flawlessly over the last 12 months. As a shareholder, you probably couldn’t have asked for more as far as creating near-term shareholder value by capitalizing on a new business driver. That’s the good news.
The bad news, well, I’ll let you guys be the judge of that. But I will say one thing before I get started, any professional investor reading this should really ask themselves how, with an army of analysts covering the stock has, nobody tackled these issues yet.
My views around the longer-term threats facing Nvidia’s ML/DL accelerator datacenter business are well known.
That clearly controversial topic is not the focus of this piece. In fact, acknowledging the spectacular success of the datacenter business over the last two years is the starting point of this analysis.
From Q4 F2016 to when they reported on Q4 F2018 in early February of this year, the stock appreciated 715%. In that time, gross profit grew by $3 billion. As Nvidia is nice enough to break out gross margins on their reporting units, you can accurately estimate the contribution of each units GP growth to overall business.
This is what a rough estimate breakdown of GM growth looks like over Nvidia’s bull run.
..
2015-2017 ($mil)
% of Pie
NVDA Gross Profit Growth(x licensing)
$2704
100%
Datacenter/Prof Vis GPU GP Growth
$1422
53%
Tegra(Auto/Switch) GP Growth
$350
13%
Gaming/Crypto GPU GP Growth
$932
34%
As expected, Datacenter, with their robust gross margins was the main driver, and what’s even more clear is that the ‘new’ businesses of auto, Nintendo switch, and even crypto specific in 2017 contributed roughly 2/3 of all gross profit growth. The core gaming, and that’s assuming no crypto contribution in 2017, accounted for just 25% of the pie. A more realistic assumption is 20%, which by the way is consistent with the stable growth we’ve come to expect from this business.
These slides are from Nvidia’s last two analyst days.
.
There are some very clear takeaways from these slides. Notably, since Nvidia’s gaming growth spurt started five years ago, units have grown at 15% per annum and ASP at 11%. Also, when you factor in the recent surge in gaming notebooks you can safely conclude that the high-end desktop rig market has been a pretty stable single digit unit grower.
Nvidia gaming GPU revenue was only up 21% in 2017, and that’s with them disclosing that their GPU ASP increased from $180 to $200 during the fiscal year. This implies high single digit unit growth at best, which says a lot about organic growth in the high-end PC gaming market.
Here are some of the key things that happened in 2017:
-Nvidia launched the 1080ti, 1070ti, and the Titan X
-Crypto alt-coin demand arrived in April and we had two short lived price spikes in Jun/Jul and Dec/Jan, but relatively stable pricing for rest of the time
-AMD was completely out of the high-end card market for ¾ of the year with no competitive offerings to compete with Nvidia’s $350+ Geoforce cards.
Now you’d think several new high end SKU’s, lack of AMD gaming competition, and a new demand tailwind would have contributed to above average unit growth or a much sharper rise in ASP’s. That clearly wasn’t the case in 2017.
So how does one reconcile this disparity?
Here are five possible explanations for what transpired:
1)In the early stages of alt coin mining lift off, when difficulty levels were low, Nvidia’s gaming card could not compete on ROI vs AMD RX Radeon series.
2)As miners cleaned out AMD’s cards, Nvidia was quick to respond with 1million units of competitively priced P106-100 dedicated mining cards delivering quick volume to large farms.
3)Pascal architecture has clearly matured.
4)A big theme regularly emphasized by Nvidia’s CEO is that their Geoforce series is better value than buying a console. But with Nvidia GPU’s essentially doubling over last five years, that argument is no longer as compelling. A solid entry-level PC gaming rig in 2017 averaged almost 50% more than a Xbox1 or PS4.
So, for all the talk of Fortnight on conference calls, clearly GPU rig price competitiveness in mass market is no longer a major value proposition driver.
5)The ultra-high-end PC gaming market where you are spending 2k plus on a 60fps high performance rig remains a niche.
Thus, outside of the last month and half of the quarter, 2017 was for the most part characterized by a very clearly muted gaming market for Nvidia complemented by a steady mining tailwind. You could argue that if it wasn’t for crypto mania explosion clearing out supply of all cards at the end of Q4, that 2017 gaming GPU demand would have been viewed as rather disappointing year.
But despite the explosion in crypto demand and Nvidia’s failure to quickly respond with much in dedicated crypto boards, Q4 gaming revenue was only up 16% yr/yr which makes you think that organically gaming was quite weak x crypto and asp boost.
Which bring us to Q1 of this year……
In Q1 Nvidia’s gaming revenue surged 72%. This was quite surprising considering what I just pointed out, and the more muted relative trajectory gaming was following throughout the 3 preceding more stable alt coin quarters.
Q2 2017
Q3 2017
Q4 2017
Q1 2018
Gaming GPU Yr/Yr
36%
13%
16%
72%
Gaming Gpu+Crypto Yr/Yr
56%
19%
22%
107%
(source: NVDA SEC filings )
And this surge flipped the economic narrative that’s been behind the stock the last 2 years.
In a quarter with 125% y/y operating income growth, an estimated 70% of all gross profit growth came from Gaming/Crpyto GPU business.
Q1 2018 yr/yr ($mil)
% of Pie
NVDA Gross Profit Growth (x licensing)
$961
100%
Datacenter/Prof Vis GPU GP Growth
$262
27%
Tegra (Auto/Switch)
$33
3%
Gaming/Crpyto GPU GP Growth
$666
70%
(Source: Nvidia 10-q and segment margin disclosures)
Now one wonders how you explain this.
Q1 is usually seasonally down, and there had to be a decent degree of channel fill. So, what did Nvidia CEO have to say on this topic?
Just take a look at the Q1 transcript ….
Stacy Aaron Rasgon – Sanford C.
Bernstein & Co. LLC
Hi guys, thanks for taking my questions. First, I had a question on gaming seasonality.
It’s usually down pretty decently in Q1. It was obviously flat this time as you were trying to fill up the channel.
Now that’s done. I was just wondering what the supply demand dynamics as well as like any thoughts on crypto might mean for typical – the seasonality into Q2 versus what would be typical where it would usually be down – or usually be up pretty decently. How are you looking at it? And this is a question for Colette.
Jen-Hsun Huang – NVIDIA Corp.
Okay. Hi Stacy, so let’s see. Q1, as you probably know, Fortnite and PUBG are global phenomenons (sic) [phenomena] (18:51). The success of Fortnite and PUBG are just beyond comprehension, really. Those two games are a combination of Hunger Games and Survivor has just captured imaginations of gamers all over the world. And we saw the uptick and we saw the demand on GPUs from all over the world.
Surely, there was scarcity as you know. Crypto miners bought a lot of our GPUs during the quarter and it drove prices up. And I think that a lot of the gamers weren’t able to buy into the new GeForce as a result. And so we’re starting to see the prices come down. We monitor spot pricing every single day around the world.
And the prices are starting to normalize. It’s still higher than where they should be. And so obviously, the demand is still quite strong out there.
But my sense is that there’s a fair amount of pent-up demand still. Fortnite is still growing in popularity.
PUBG is doing great. And then we’ve got some amazing titles coming out.
And so my sense is that the overall gaming market is just really, is super healthy. And our job is to make sure that we work as hard as we can to get supply out into the marketplace. And hopefully, by doing that, the pricing will normalize and the gamers can buy into their favorite graphics card at a price that we hope they can get it at. And so I think there’s a fair – I mean the simple answer to your question is Fortnite and PUBG. And the demand is just really great. They did a great job.
“The Simple answer to your question is Fortnite and PUBG”
Now, last I checked PUBG was a year old and Fortnite launched last summer. Yet, in a quarter where gaming exploded over 70% after growing 15% for the previous six months; we are expected to take this answer seriously?
In Q1 Nvidia did nearly $300 million in crypto board specific revenue.
To understand this business, you need to look closely at the GPU mining chip variants Nvidia has sold in the past year.
P106-100
P104-100
P102-100
Gaming Equivalent
GTX 1060
GTX 1070/1080
GTX 1080ti/Titan X
Launch Date
June 2017
Dec 2017
Feb 2018
Die Size(mm^2)
200
318
471
Ethereum Hashrate
20mh/s
34mh/s
47mh/s
Price for bulk buyers
$220
$400
$599
They started out with the P106 which was essentially a stripped down GTX 1060, and it seems they were able to quickly sell a million of these last summer. Now understand these cards have no video ports and are sold in bulk to huge mining farms. The margins on them are better than the gaming variants, especially on the higher end ones as they use significantly less DRAM.
Nvidia accomplishes two things immediately by selling these cards:
1)They capture more profit per unit versus the gaming variant.
2)They meet mining demand with cards that cannot be used for gaming in the future
These two simple points demonstrate that Nvidia is incentivized to shift production away from gaming cards if possible. They are not doing gamers a favor by doing this as they would lead them to believe.
And as the mining cycle extends, their incentives to produce more of these cards in ever increasing ASP varieties grows.
The best example of this is the p102-100 card, and the surge we saw in Crypto mining board specific revenue in Q1.
Investors need to realize that the high- end GPU Gaming market, cards that retail over $350, has been pegged by AMD at roughly a million unit a quarter market.
To put that in perspective for you, it would appear that Nvidia in Q1 was able to sell roughly a million units of their most expensive gaming GPU chips directly to large mining farms. That’s how their dedicated mining revenue in Q1 was equal to their total dedicated mining revenue of the previous nine months. The implications of this are quite significant. Nvidia is in the midst of a virtuous mining cycle.
NVIDIA’S VIRTUOUS MINING CYCLE
1)One you can very easily conclude that nearly 40% of the Ethereum hashrate increase in Feb/Mar was directly attributable to these new high-end crypto boards.
2)You can also conclude that doing such huge volumes of their largest gaming GPU (the p102 is 135% bigger than the p106) means they had far less capacity for lower end cards.
3)By rapidly selling huge volumes of their most powerful gaming chip in a mining variant directly to large mining farms Nvidia ensures that they will drive up the difficulty levels of the respective crypto coins.
4)As difficulty quickly increases, they are further incentivized to shift their gaming product mix to their highest end cards as well. This is because DIY miners who will be getting access to these cards later than the large mining farms will need that hashpower and thus bid up those cards.
5)The combination of 2-4 ensures that you hit both dedicated mining demand and get maximum ASP out of your remaining gaming supply, which largely is gobbled up for mining as well. These actions ensure the overall GPU market will be supply constrained, and that prices will overshoot to the upside.
6)The extremely elevated retail prices allow you to raise all GPU quotes just as your visibility into mining demand cooling off clears up.
These higher ASP’s offset your lower unit volumes as end markets start to normalize. (we got evidence of this here )
7)Since prices fall a lot slower than they rise, Nvidia achieves a higher base GPU ASP to meet pent-up gaming demand.
8)All the while they tell gamers they are doing their best to meet gaming demand and sit back and watch while DIY miners flip out over “Evil Bitmain ASIC’s” killing the small GPU miner, when in fact they have been squeezing them as well.
Now isn’t this a better explanation than:
“The Simple answer to your question is Fortnite and PUBG”
Like maybe you can blame Jansen for suffering from founder CEO syndrome, but there is no excuse for all of Wall Street sitting back and not asking hard questions that should be patently obvious to people in the investment business. This a classic commodity boom type story here which doesn’t require much brains to figure out. But maybe I’m being a bit harsh as they have to contend with a CEO who says stuff like this…
“We’re working really hard to get GPU down to the marketplace for the gamers and we’re doing everything to advise retailers and system builders to serve the gamers. And so, we’re doing everything we can, but I think the most important thing is we just got to catching for supply.” (NVDA CEO, Q4 CC )
Really? You could easily make no mining boards thus ensuring all supply is gaming viable either now or at a later date.
You could also drastically ramp the number of units available directly from your website at 2 card limits. There are in fact a lot of options that simply would be great for gamers and not very capitalistic, but you are in fact saying all this gamer first nonsense while you steer the business to capitalize of mining as much as you can.
“And so now, an investor a developer could invest hundreds of millions of dollars and create something that is just completely for the realistic and emerged that just beautiful. And so when the production value goes up, the GPU technology that’s needed to run it well goes up, it’s very different than music, it’s very different than watching movies, everything in videogames is emphasized in real time.
And so when the production value goes up, the ASP with a technology has to go up.”( NVDA CEO, Q4 CC )
Somebody want to explain to the CEO of the world’s biggest gaming gpu company that video game prices have been declining for the last 25 years.
Mike Tyson’s Punch Out cost virtually nothing to make and its 1987 price is what you’d pay for $200 million plus development budget Call of Duty. Does he watch Netflix? How many billions worth of content do you get for $10 a month? I think you can pick out 2-3 quotes a conference call from him that just leave you dumbfounded.
Then you have this critical exchange on Q1 call :
Timothy Arcuri – UBS Securities LLC
Thank you. I actually wanted to go back to the question about seasonality for gaming in June.
Normal seasonal sounds like it’s up mid-teens for June in gaming. But obviously, the comps are skewed a little bit because of the channel restock and the crypto stuff. So does the guidance for June assume that gaming is better or worse than that mid-teens normal seasonal? Thank you.
Jen-Hsun Huang – NVIDIA Corp.
We’re expecting Q2 to be better than seasonality, if I understand your question. We’re expecting Q2 to be better than Q1.
And we’re expecting Q2 to be better than seasonality. Did that answer your question?
This was instantly perplexing answer because it didn’t remotely jive with the guidance. If Q2 is going to be better than mid teens seasonality, you are talking something around $300ml in extra revenue. If you adjust for crypto guidance, that’s a net $100ml gain. That mean’s according to their guide that $200ml in Q2 revenue will vaporize somewhere between DCG/PV/Auto. That’s obviously impossible, and thankfully an analyst quickly picked up on this and asked a question.
Here is the follow up exchange this time handled by the CFO:
Stacy Aaron Rasgon – Sanford C. Bernstein & Co. LLC
Hi, guys. Thanks for fitting me in for my follow-up. This is a question for Colette. I want to follow-up again on the seasonality.
Understanding the prior comments, normal seasonal for Q2 for gaming would be up in the double digits. Given your commentary on the crypto declining in Q2, given your commentary on just the general drivers around datacenter and the Volta ramp, I can’t bring that together with the idea of gaming being above seasonal within the context of your guidance envelope.
So how should I reconcile those things? How are you actually thinking about seasonality for gaming into Q2 within the context of the scenarios that are currently contemplated in your guidance for next quarter?
Colette M. Kress – NVIDIA Corp.
Sure, Stacy. Let me see if I can bridge together, Jensen, and then some comments here.
Unfortunately, they’re moving quite fast to the next question, so I wasn’t able to add-on.
But let me see if I can add-on here and provide a little bit of clarity in terms of the seasonality. Remember in Q1, we outgrew seasonality significantly. We left Q4 with very low inventory in terms of in the channel.
We spent Q1 working on establishing a decent amount of inventory available.
We wanted to concentrate on our miners separately. And then you can see we did that in terms of Q1 by moving that to OEM and moving that to cryptocurrency only boards. So we left Q1 at this point with healthy overall channel inventory levels as far as where we stand.
So that then takes you now to Q2. But if we overshot in terms of seasonality in terms of Q1, we don’t have to do those channel fill dynamics again as we get into Q2. But we do have demand out there for our gamers that we can now address very carefully with the overall inventory that we now have available.
So putting together, Q1 and Q2 together, yes, we are within normal seasonality, again, for a guidance. And we’ll see how we’ll finish in terms of the quarter. But you should be in that range.
So, yes, from a normal seasonality, at a year-to-date inclusive of Q2, yes, we’re on that overall seasonality. Always keep in mind, generally, our H2s are usually higher than our overall H1s, and that’s what you should think about our overall guidance.
Gaming is still strong. We have to comment that our overall drivers that have taken us to this place over the last three to five years with phenomenal growth and our ability to grow that overall market is still here and all of those things are together. We just had a few quarters in terms of making sure that we get the overall channel correct and put our miners separately.
I hope that clarifies in terms of where we are, in terms of gaming seasonality.
This answer is a complete 180. The CFO is now telling you that contrary to what Jensen just said, gaming q/q is essentially being modeled flat. This is significantly worse than normal Q2 seasonality. This of course makes sense considering the Q1 channel fill and mining explosion, but it’s still one hell of a headscratcher.
The CEO clearly understood the question and repeated his answer twice.
Possible explanations for this???
1)Nvidia is sandbagging and Jensen, being as bullish as he is, can’t help himself and slipped up. This is a plausible theory, but NVDA has zero history of sandbagging. If you look at the top line beats vs. guidance, the recent delta is almost all crypto. But with crypto visibility being what it has been going into this report, you’d have to believe they somehow sandbagged $350ml in raw gaming and DCG revenue.
And seeing as we are half way through Q2 with no new gaming launches, that now seems impossible.
2)Another explanation is Jensen is simply referring to gamers from the view they been locked out of the market, and thus his pent-up demand take means he expects them to be seasonally stronger than normal.
The offset here, which the CFO had to indirectly clear up, is this pent-up demand + increased ASP is expected to net wash the lost gaming classified revenue that was in fact mining driven. This means “gaming” is seasonally stronger than normal, but still not enough to completely offset lost gaming crypto demand. Personally, this makes the most sense.
3)A third possible explanation is that Nvidia is planning a new gaming GPU launch near the end of the quarter, but that it was not an absolute certainty.
Here you can say Jensen jumped the gun, and the CFO had to correct this. Possible, but very hard to believe considering production lead times in this space.
Then as if all this wasn’t confusing enough, at Computex this week, Nvidia’s CEO told us that new gaming cards would not be coming for “a long time”. I am hard pressed to recall a CEO of tech company of this size ever making such a vague public statement regarding a key product roadmap. Every sell-side model is factoring a late Q2 into H2 major bump from a new graphics card cycle, so how exactly are they supposed to interpret this statement.
No new launch as we lap the crypto boosted quarters? Or is the launch being pushed back to Q4/Q1 2019 to in fact offset those crypto comps in gaming? This does seem to make some sense. You can conclude Nvidia is banking on much higher ASP’s and pent up gaming demand to work their way through Q2/Q3. Or maybe they know something more about Etherum’s roadmap and want to time their refresh to match the proof of stake transition there. That way they further mitigate contending with 2nd hand mining Pascal cards flooding the market.
At some point we will get answers to these questions, but for now all we can do is speculate. But one area that clearly deserves a lot more focus is Nvidia’s crypto mining business.
For all the talk we got earlier in the year about Bitmain making more money than Nvidia, reality has set in and we now know that’s not remotely true. Bitmain’s CEO recently disclosed that they did $3.5 billion in revenue last year (not 3-4bl in profits). Considering their gross margins are far lower than Nvidia’s, and what we now know about ASIC mining economics thanks to Cannan Creatives IPO filing, Bitmain’s 2017 profits were likely no more than $800ml. As alt coin mania started far later than Bitcoin, Nvidia’s TTM crypto profits likely matched or exceeded Bitmain’s.
This is somewhat ironic when you consider that the crypto critics on the stock have quickly capitulated, seemingly because Nvidia’s GPU biz didn’t unravel at the same rate as crypto coin prices. One is left to wonder if any of these analysts seen how a commodity bust plays out for the supply chain winners? Frack sand anyone?
The way I see it Nvidia currently has a two-fold crypto problem
1)Crypto GPU demand over the last twelve months has amounted to at least $1.
6bl in revenue, and now is a headwind if things just remain stable over the next 4 quarters.
2)GPU market ASP’s are very vulnerable to a sharp decline should crypto mining rapidly transition towards ASIC’s.
Issue number one is pretty self- explanatory, but issue two is a lot more interesting.
Every move Nvidia’s has made throughout this crypto craze is exactly the type of stuff that Bitmain fearing alt-coin crypto purists thought they were preventing with ASIC resistant/GPU friendly proof of stake algorithms. Nvidia has demonstrated the ability to quickly produce a lot of mining specific GPU’s that are delivered in volume directly to large mining farms. This is the same thing Bitmain is accused of with early mining batches. The difference here is that Bitmain is dependent on their mining customers, and thus is incentivized for them to earn a sufficient ROI. Nvidia could care less. Their R&D investments for their graphics card are a sunk cost that is recouped from gaming demand. They are free to shift production as they please to capitalize on lucrative mining demand, and then shift right back to gaming or GPU acceleration when that demand disappears.
Bitmain and other ASIC crypto mining manufacturers don’t have that luxury, so they are naturally more vested in the long-term success of miners.
For all the hoopla, Crypto mining is a manufacturers game.
They control supply, they know their chips economics, and they have perfect visibility on large mining farm demand. This means they know where the hash rate is going, and the impact that will have on the demand for their 2nd batch of ASIC’s or in the case of Nvidia, their broader market GPU’s. This is what happens when one manufacturer dwarves all others.
So, if decentralization is essentially an illusion in alt-coin land, why resist ASIC technology?
At this point, I can’t see any reason. It’s been clearly proven that ASIC’s can be made for all the alt-coins that will quickly render GPU’s obsolete. Technologically there is no fighting this.
It’s also been proven lately that ASIC’s can be made which retain enough flexibility to overcome repeated hard forks. Thus, it is inevitable that in a short-time anyone making money mining will be relying only an ASIC. But what I see developing in the crypto community based on what’s come out of thought leaders like the founders of SIA coin, ZCASH, and Ethereum is increasing evidence that the leadership gets this. That’s a notable shift, and this has occurred without any heat on the likes of Nvidia and how they have quietly cleaned up on mining while publicly adopting the position of gaming GPU starved consumer champion.
In fact, you can argue that more crypto coin developers are realizing there is upside in partnering with ASIC manufacturers.
Anyway, as you can see there is a lot to think about going forward in GPU land.
It will be interesting to watch what happens if Ethereum ASIC’s get going later in the summer, because this could really be a hard catalyst here for the whole GPU market. And I think the back and forth between AMD and Nvidia now deserves some more attention.
AMD has been grabbing headlines with their roadmaps and GPU demos, and for the most part Nvidia has essentially pretended like they don’t exist. Does Jensen’s vague time table around Pascal’s successor have something to do with a big surprise for AMD? Or have they hit a hiccup? Only time will tell but seeing as how many GPU nuts there are in both these stocks, I’m hoping that by sharing this note we get some answers in the subsequent debate.
Disclosure: I am/we are short NVDA.
I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.
Additional disclosure: long googl, vicr, amd, avgo.