NVIDIA revenue growth sustainable?

So for the umpteenth time on this board I want to revisit NVIDIA. This time to make sure I’m not too in love in the company, blinding me, being a detriment to market-beating gains. Saul’s point and reason why he’s out of NVIDIA is very very valid: “being hardware, the only way it can grow revenue 60% is to sell next year all it sold this year and 60% more. To do that for the next 3 years it would have to quadruple its chip sales. To do it for 4 years would require well more than a sextuple (655% actually). I think that is simply unlikely, while for one of our little SaaS companies it is much less unlikely, because last year’s sales stick there and don’t have to be redone. In fact they grow almost on their own. And, as far as multiplying stock price by six, it’s a lot easier for a company with a market cap of $5 million than for one with a market cap of $165 million; a LOT easier!:grinning:
Sure NVIDIA is not like other hardware companies, is relatively asset-light with high margins, but it still has no recurring revenue. So the question remains, for how long can they maintain their growth-rate. Hopefully this sparks discussion and research that will better help us decide what to do with NVIDIA.

This is not:

Looking at seeing whether NVDA is a better investment than the likes of MDB or PVTL. If MDB and PVTL are a success, the stock appreciation of a 4 and 7 billion market cap company is clearly going to eat NVDA’s appreciation for breakfast. However, I am more than happy with a 40% a year increase - that’s a double in 2 years.
Going to discuss NVDA’s CAP. I’m taking it as a given that NVIDIA’s Competitive Advantage Period is, for the time being, seemingly unassailable. Rumours have abounded for years that AMD and not Intel were/are coming out with an NVDA killer. Time and again NVDA has not only demonstrated being able to keep up, being ahead of the game, they’ve increased the lead. Unless anyone can provide something new in regards to competition, I’m not interested. Just wait for benchmarks to do the talking.
Insert point 3 here

NVIDIA
Market Cap: 160 billion
Stock Price: $270
TTM EPS: 7.04
Guided EPS: $8.00 (ultra ultra conservative)
TTM PE: 38.35
2019 PE: 33.75

Money Making Machine


       **yoy revenue growth	yoy EPS growth**
Q1 2017	  13.38%	        39.39%
Q2 2017	  23.85%                55.88%
Q3 2017   5.37%	                104.35%
Q4 2017	  55.10%                117.31%
Q1 2018	  48.43%           	84.78%
Q2 2018   56.16%          	90.57%
Q3 2018   31.54%                41.49%
Q4 2018	  33.92%            	52.21%
Q1 2019	  65.72%	        141.18%
Q2 2019	  39.91%	        92.08%

Cursory glance at just the numbers here, you can see how NVIDIA makes money. OpEx increases are minimal vs revenue increases. Very similar to ANET here. Additional revenue is falling directly to the bottom line at a remarkable rate. This demonstrates a well-run company that knows how to make money. I would also like to add a comment about NVIDIA being just another chip company. I would venture to say that NVIDIA is different. They have massive pricing power, their margins aren’t coming down due to competition, they’re extending their lead and people want their chips. Either because it saves them money (data centers), or because the hype generated around them (gaming), much like apple and the iphone.

Back of the napkin math here. Let’s model for 50% EPS growth for the next 4 years.


	**EPS	PE=35	PE=25	PE=15**
2019	8	$280	$200	$120
2020	12	$420	$300	$180
2021	18	$630	$450	$270
2022	27	$945	$675	$405

If the autonomous market never takes off and they reach stagnation/saturation in their current markets, a P/E of 10-15 would be realistic I imagine. In which case we’d probably be out of it long before then. Not sure why I did this table now but there you go.

Market Segments


QUARTER	TOTAL	GROWTH	GAMING	GROWTH	DATA CENTER	GROWTH	VISUAL	GROWTH	AUTO	GROWTH
Q1, 2016	1.2 B	5%			79 M	57%			77 M	121%
Q2, 2016	1.15 B	5%	661 M	59%	62 M	-15%			71 M	76%
Q3, 2016	1.3	8%	761	44%	82 M	8%	190 M	8%	79 M	51%
Q4, 2016	1.4 B	12%	810 M	25%	97 M	10%	203 M	7%	93 M	68%
Q1, 2017	1.31 B	13%	687M	17%	143 M	63%	189 M	4%	113 M	47%
Q2, 2017	1.43	24%	781M	18%	151 M	144%	214 M	22%	119 M	68%
Q3, 2017	2 B	54%	1 B	63%	240 M	193%	207 M	9%	127 M	61%
Q4, 2017	2.17 B	55%	1.35 B	66%	296 M	205%	225 M	11%	128 M	38%
Q1, 2018	1.94 B	49%	1.03 B	49%	409 M	186%	205 M	8%	140 M	24%
Q2, 2018	2.23 B	36%	1.19 B	52%	416 M	175%	235 M	10%	142 M	19%
Q3, 2018	2.64 B	56%	1.56 B	25%	501 M	109%	239 M	15%	144 M	13%
Q4, 2018	2.91 B	34%	1.74 B	29%	606 M	105%	254 M	13%	132 M	3%
Q1, 2019	3.21 B	66%	1.72 B	68%	701 M	71%	251 M	22%	145 M	4%
Q2, 2019	3.12	40%	1.8	51%	760	83%	281	20%	161	13%

Okay so here is the crux of the matter. What are the growth predictions for their current markets?

Data Centres
The fastest growing segment. Buy more GPUs, save more money. That is the CEO’s message. They require less hardware, consume less energy, and saves the end-user money, whilst completing more advanced functions. NVIDIA’s growth of this segment has been amazing and needs to continue. A gartner forecast predicts 2019 IT spend in data centers to be about 1.1%, and another predicting 9% growth annually over the next 5 years. Terrible! .
NVIDIA says it’s TAM in this market will be $50 billion by 2023. So there’s some disconnect there (which I’m sure is because of me). But at at run rate of 2.5 billion, there’s a lot of room for growth here. Near-term/mid-term, data center growth is absolutely vital. Can NVIDIA maintain 60% growth annually for the next 5 years? Well, I’ll be looking for 70-80% for the next year at least.

Also, CUDA

Gaming

Average launch price of new GPUs has had a gradual increase since 2011, compared with AMD being flat. The previous top tier GTX 1080 ti launched at $700 in 2016. Now the RTX 2080 sold out in pre-orders in just a few days at $1,200. When fully released it will be $999. To me, that’s like apple and their iphones. $1,000 for a phone? Ridiculous. $1,000 for a gaming GPU? Crazy!! And yet both sell out. Apple didn’t need to sell more phones to show growth, they just sold them at higher prices.

NVDA are making a custom GPU for Nintendo’s switch. This is an important optionality for NVDA. PC gaming has seen a recent uptick in recent years and NVDA is benefitting from it. Consoles aren’t going away and I’m sure the battle between PC and consoles for market share will continue. NVDA need to ensure their GPUs are in the consoles as well. They’re not making the same mistake intel did with ARM and mobile phones.

Gaming growth. Difficult to find predictions, but I’ve seen some ranging between 10-20%.
Not great at all! Especially as they already have a dominant 70% market share in this market. Extra growth has to come from increasing that share and increasing pricing power. With the current architecture upgrade, rapid selling out of preorders (RTX 2080 TI and 2070 are sold-out), I am expecting a big uptick in Q3 and Q4. As the gaming industry starts making RTX games, the NVIDIA cards will become must haves. Market share will increase unless AMD comes up with something fast, which is not how things are done! They’ll be minimum 1 year away, if not 2 or 3.

NVIDIA also works closely with game developers, providing software/openCL, that helps game developers optimise their games for NVIDIA GPUs.

New specs for the RTX are out (from NVIDIA) claiming the new GPUs are between 50-75% faster than the 1080 without utilising ray-tracing. This is great news as there was doubt about how big a different the new chips will be if RTX was off. This helps clear that up! A lot better!

Autonomous
Not going to say much here. Currently a story. Big hopes for it to takeover growth if the others start to flag. They’re creating the biggest partnership network in this industry, but we’re still a couple years away from realising it. If NVIDIA takes this market, then all bets are off and woaah it’s massively undervalued. But for the time being that’s just a story, and we’re not here for story stocks.

Visualisation

So this fledgling segment, I think, is quite exciting. With the Quadro RTX cards only available in Q4, I don’t expect a significant uptick here until next year. These are NVIDIA’s highest margin units, and opens up the massive visual effects industry.

A render farm of 240 Dual 12-core Skylake Intel CPU servers costs about $2 million, according to Huang, compared to a comparable setup of four Nvidia RTX 8-GPU servers that would cost $500,000, take up about a tenth of the space and require one-eleventh of the power to operate.

This continues the theme of Buy more GPUs (NVIDIA), save more money. They’re making intel look old and obsolete here. We have to wait to see how intel are going to fight back here, but this is/will steal intel’s lunch.

From a personal point of view, I’m excited at what the movie and animation can create for us to consume. You don’t need to be a big production company to create something cool - this opens the industry up to more content creators. Remember when Toy Story first came out? It looked amazing, but it took them over 4 years to create. Render times were over 4 hours per frame. When they rerendered it in 2011, it still took a render farm, but they could do it in the background at 30 minutes per frame. Well, NVIDIA are working with Pixar to develop a programming language to fully take advantage of these new cards. Yup, NVIDIA differentiating themselves again with software.

Now it’s not just the movies that will support the pro vis segment. Catalogues are created/rendered. Big enterprises such as IKEA use render farms to create them. With the new Quaddro RTX, with real-time ray tracing, not only will it be cheaper, but faster!

I’m quite excited for this segment. Prior to this month I really didn’t think much of it all. Focus has been on gaming and data center growth, with autonomous cars a dream for the future. Clearly it has a tiny base to grow from, but Q1 and Q2 will be interesting in this segment.

Conclusion

So I’m not sure if that was as helpful (to me) as I thought it would be. I started this with, admittedly, biased feelings hoping to present the bull case. I was hoping to do this by finding sources predicting data center and gaming growth will be in the high, let’s say 40s, for years to come. The sources I found barely make them double-digit growth. Terrible, terrible! And yet I’m currently not swayed away from NVIDIA. Clearly I’m still in love with the stock and will need more to sell it. It’s currently 10% of my portfolio (60% is in fast, high-growth stocks like MDB), and I do see it as a steady-eddy anchor to my port.

Hope you find my ramblings somewhat useful.

64 Likes

First of all, NVDA is not just another chip company. Or for that matter, a board or system company. They also have quite the software stack going for them. This is a big plus. Intel and AMD cannot compete on the hardware side, but also not on the software side. This is a big deal.

My only concern with them, and I’ve mentioned this quite a bit on the Stock Advisor NVDA forums along with Starrob, that the GPU will eventually be eclipsed by something else. That will probably take 5-8 years to do, however. The GPU is very energy intensive is the primary issue with it. So much so that process shrinks and architectural tweaks won’t help much. But, as Turing has shown, with NVDA including Tensor cores and new ray tracing AI technology, I have optimism that Jensen is not dumb enough to ride the GPU into eternity.

For now I’m comfortable with an 8% stake in them. I consider them a “growth anchor” stock, if that makes any sense. But honestly, I understand why this board in particular is not overly in love with NVDA. They are a great stock, but I really do expect some of my picks from this board to out-perform NVDA over the next couple years. I’m just not willing to go all-in with a full-on growth concentrated portfolio myself.

14 Likes

It is not quite true that there is no recurring revenue since, as the technology improves and the games take advantage of that technology, the user has to upgrade in order to get the best game experience. It is not that the old board wears out and needs replacing, but that the standard changes for what is an acceptable board.

3 Likes

A gartner forecast predicts 2019 IT spend in data centers to be about 1.1%, and another predicting 9% growth annually over the next 5 years. Terrible!


Go look at these numbers and buy more TWLO, I guess, as Communication services is about a 1/3rd of the 3-4T IT global spend!
https://www.gartner.com/en/newsroom/press-releases/2018-01-1…

For Datacenters specifically, keep in mind that NVIDIA’s DC category includes cloud-based GPUs in AWS, etc…and should include a proliferation of edge/IOT devices over time (Jetson-based versions of their drive platform) and more.

Also, while global DC growth in the world may not be huge, the manner in which that budget is spent is the key. NVIDIA cutting into legacy CPU spend, as an example.

The analogy may be with TTD and advertising. Even if advertising market didn’t grow, if the dollars/budgets are shifting from traditional (newspapers and TV, etc) to digital/programmatic, then TTD stands to reap benefits.

Same here for NVIDIA…the total IT spend globally isn’t the issue…it is how those dollars are spent moving forward, and how much of that pie the GPU-based solutions can capture.

Dreamer

12 Likes

50% growth and 15x 2022 pe is a 10.7% CAGR, fwiw.

4 Likes

It is not quite true that there is no recurring revenue since, as the technology improves and the games take advantage of that technology, the user has to upgrade in order to get the best game experience.

That’s not recurring revenue, that’s an opportunity for a competitor to step in with a better/different product. Just ask any automotive OEM.

5 Likes

Well, it isn’t guaranteed recurring revenue, but at this point Nvidia doesn’t really have a competitor that can “step in”. If you want state of the art, you get it from Nvidia.

My point is that it isn’t like selling someone a car where not only can they replace it with any one of a large number of choices or even keep it for 10+ years. To stay on top of state of the art gaming or other GPU applications, you need to keep current with the latest product on a relatively frequent basis.

6 Likes

50% growth and 15x 2022 pe is a 10.7% CAGR, fwiw.

Definitely worse bear cases for a lot of other stocks, right?

1 Like

NVDA’s growth prospects are not that much of an enigma. It should be easy. They have multiple growth avenues and lots of greenfields. And, in fact, are even creating markets from thin air. Use cases that seemingly no one thought of before.

In Data Center, they own training and are making huge progress (as in doubling sequentially) in the presumed larger Inference market. Inference is currently overwhelmingly CPU. This WILL get disrupted by the lower latency more efficient GPU.

Gaming. They got this one locked up. They just changed to game for good, pun intended, with Turing and RTX. Top end gamers will need NVDA to play top end games.

Pro vis like Inference has farms of CPUs to displace. NVDA technology is so vastly superior you have visualization companies changing the name of their product from RayRT to Ray GPU, as an example. Sleeper segment poised to ramp quickly IMO. About 11% sequential growth this last quarter.

Automotive. AI cockpits and autonomous vehicles. Based off their partners if the race were today they’d have the vast majority of worldwide market share in AV. Only realistic “competitor” is Intel. AI cockpit is likely to be a huge win too. With the newest products just hitting the market Auto grew 12% sequentially this last quarter.

With those last two segments starting to show signs of life with promising markets in all segments I think that at some point over the next year or so (by end of 2019) you will have a quarter where all 4 segments will hit stride all at the same time. I’ll define that as all 4 segments with greater than 30% yoy revenue growth.

Having been invested now for greater than 3 years, Nvidia does this as new products and innovations propel them to the next big thing. They consolidate and trade in a range for awhile before driving higher again. Maybe the recent breakout is the start of the next trend? We’ll see I guess, but staying invested here is an easy decision. I understand some who want to chase other stocks. Having another stock to buy that you think will be much better is always a valid reason to sell. But the thesis for Nvidia is better than ever right now.

Darth

30 Likes

Automotive. AI cockpits and autonomous vehicles.

Autonomous could/should be a lot more than our cars/vehicles on the roads today.
Drones, robots…anything that can move, and would benefit from being able to move without human assistance. Military applications could be staggering.

There was some mention of a combo of Jetson and Xavier:
https://www.forbes.com/sites/moorinsights/2018/08/24/nvidia-…

NVIDIA will productize the Xavier SOC in at least three form factors: The Jetson Xavier for drones and robots, the Drive Xavier for apps such as Level 3-4 driving assistance, and the flagship Drive Pegasus, with dual Xavier SOCs and 2 Volta GPUs to support fully autonomous Level 5 vehicles.

Dreamer

long: NVDA, TTD, IQ, BZUN, AYX, TWLO, and NTNX. Oh, and the tiniest bit of expensive ZS.

5 Likes

But the thesis for Nvidia is better than ever right now.

Darth

At a 40x PE it’s better than when it was at a 17 PE [and $19] just a few years ago?

I don’t think so.

2 Likes

Jensen Huang, the CEO, describes Nvidia as a software company. They also make the hardware that their software runs on. A few years ago, he said Nvidia employed more software engineers than hardware engineers. CUDA is the best gpu tool available. Nvidia has a substantial moat. A competitor would need to make better hardware and software AND get support added to dozens of rendering applications.

AMD’s roadmap has their next-gen chip, NAVI, taping out soon, for a 2H-2019 delivery. For it to include ray tracing functionality, AMD would have needed to start designing it 12-18 months ago, which I find unlikely. It might be 2H-2020/2021 before AMD can match what Nvidia is offering today.

I do worry about the cap on growth rate. But I don’t see a competitor taking significant share any time soon.

3 Likes

Thanks all for the replies and comments.

We’ve been over numerous times on these boards how NVIDIA crushes their competition. Jensen was smart enough to build the lead in software and with partnerships. Everytime a competitor appears to catch-up, NVIDIA straight away releases a new, upgraded model that is years ahead of the competition, almost like they’re teasing or waiting to see what the competitor does first. They have a substantial moat. I’m not worried about the moat. NVIDIA is completely and utterly dominant in their field and it will take a new paradigm to disrupt them and GPUs (assuming they themselves, utilising their AI supercomputers, don’t come up with the something new first). So let’s not discuss that.

Revenue growth is what matters. I’ve tried to present the case that the data center and gaming industry is booming so incredibly fast that growth in the 40s + % is essentially a given in their two largest revenue segments. I failed to do this - growth is, and has been, barely double digits in these industries. And yet, over the last couple years, NVIDIA has posted astounding growth, triple digit growth in the case of data centers. So it’s not the tailwinds of total spend increasing, it’s a Dreamer stated, the tailwinds of GPUs eating more pie. And who doesn’t like pie!

So they’re stealing market share.

  1. GPUs are stealing intel’s thunder in data centers. The more you buy, the more you save. With a huge market left to penetrate, growth is not over here. I will be looking for growth to be consistent high growth over the next few quarters. I hadn’t even considered IOT/Jetson platforms - thanks Dreamer.
  2. Gaming - raytracing will fundamentally change the industry here. There won’t be a competitor here for a couple years at least. 70% market share may increase in the next year. Plus pricing power has increased. I will be looking for an uptick in growth here for Q3 and Q4. Especially considering how fast pre-orders sold out - they’ll sell them as fast as they can be made.
  3. Pro-vis: contributes hardly anything to revenue at the moment, but looking forward to Q1 and Q2 results - massive uptick expected next year as they steal even more of intel’s pie.
  4. Autonomous - of course not just cars. Drones/amazon drones? Farming machines…etc? Great future but until we have something of substance that can sell like hotcakes, I’m not going to factor that in.

NVIDIA - a category crusher in a rather moderately expanding market (gaming), a disruptor in a normal market (data centers, pro-vis), and a category leader in a new market (autonomous).

I think I’ll be happy sitting on NVIDIA for the next year. To be reviewed again. High revenue growth and even higher EPS growth for the next year looks set to continue.

13 Likes

At a 40x PE it’s better than when it was at a 17 PE [and $19] just a few years ago?

My statement about Nvidia’s thesis is in regards to the company and the company’s growth prospects. Valuation is a separate issue. Where the stock price goes is typically in line with the company but who knows.

Historically the TTM P/E for Nvidia is near a 2 year low. Check the chart below.

https://ycharts.com/companies/NVDA/pe_ratio

On a forward looking it is even lower, maybe a 30-35, though a realistic case could be made, and has been made on this board, that is more likely about a mid 20’s. The price around that time was around $250. Recent appreciation would make those comparisons a few points higher.

So as far as NVDA CAP and TAM and how the company is performing, the thesis is as good as it has ever been. Better than it was in 2015 at 17x PE, which is why it is now earning the 40x PE multiple.

Darth

Expecting to see Q3 growth lower than what we have been used to (25%ish) followed by a monster Q4 on new product launch and typical seasonal yumminess.

6 Likes

Better at 40 pe than 17x, yes. It is pure luck to find turnaround stories, or stocks that are about to take off when you just look at low P/e ratios. For a decade Nvidia was at a low P/E with little share growth. Then all of a sudden it started to take off and the valuation metrics rose as well, but earnings rose even faster.

It was at this point that Nvidia began to distinguish itself from all the other “cheap” P/e stocks. That was ~200% or more ago for me.

However, feel free to buy into the 17 P/E stocks with the hope that one might suddenly take off at some point in a decade or more.

Instead I look for winners. Winners keep winning. Stock become winning stocks because something is happening with their businesses. Sometimes legitimate, and sometimes just momentum. Whatever. It is at that point that you have a systematic way to identify a great new dominating growth business. It is that way tat yo avoid Nvidia in 2005 and buy it instead in 2015 or 2016.

So yeah, Nvidia was FAR BETTER at 40x than at 17x, unless you happened to get lucky and have just bought it for some reason in 2014 because you had an inkling of something.

Tinker

18 Likes

NVDA with a GAAP P/E of 40 is cheap. Compare to growth rate.

NVIDIA revenue growth sustainable?

What is your timeframe? Short term yes. Long term, time will tell.

This is precisely what the “S” curve is about. The middle third, the fast rising part of the curve, goes from around 15% to 85% market penetration. Nvidia has multiple markets so you’ll likely see multiple “S” curves superimposed on one another.

Late in 2015 Mr. Market changed its opinion about NVDA, I haven’t explored why it happened but it must have been the recognition of the GPU architecture that is more IP and Software than chip hardware. The change in perception from chip to software is a reasonable cause for the fast rise since late 2015. Most Saul software stocks have GAAP P/Es well in excess of 40 if they have earnings at all.

Denny Schlesinger

6 Likes

Late in 2015 Mr. Market changed its opinion about NVDA, I haven’t explored why it happened but it must have been the recognition of the GPU architecture that is more IP and Software than chip hardware.
I think it was the recognition of the reality that Nvidia had found a niche to serve in graphics and eventually the coming AI and had gotten away from the dogged battle it was fighting with Qualcomm for supremacy in smart phones. The Tegra vs Snapdragon contest was really holding both companies back. When Snapdragon won in Smart phones and Nvidia cornered graphics then things got interesting.
AFAICR.
A

1 Like

When Snapdragon won in Smart phones and Nvidia cornered graphics then things got interesting.

Thanks for filling in the blank!

Denny Schlesinger

Still catching up on my reading (yes I have a life outside of investing, even in retirement). I have been pondering what to do about my NVDA position ever since I read Saul’s take on the potential for future growth. Saul is right a lot more often than he is wrong. But I just haven’t pulled the trigger on selling my position. My thoughts are, well, even if he’s right, the new Turing architecture has a significant growth path in front of it. I’ll at least wait a couple of quarters to see where that goes.

Then, today on SA I read this article: https://seekingalpha.com/article/4203387-nvidias-bullish-nar…

As way of encouraging you to read the article, here’s just a few quotes:

“CEO Jen-sen Huang sees the company growing 10-fold.” Well, he’s the CEO, what else would he say. But read on . . .

"Gaming: $1.81 billion, up 52% year over year, above the $1.75 billion consensus . . . gaming is the world’s largest entertainment industry with 200 million gamers. Even further, Nvidia’s GeForce Now, which is a brilliant cloud-based subscription service for gamers who cannot afford the expense of setting up their own at home rig, has been called a “Netflix for games” by the company. Looking for recurring revenue? No need to sell more an more chips when you can grow a subscription base in the world’s dominant entertainment industry.

"DGX-2 replaces 300 servers in the cloud, or several millions of dollars, all now in one little box for $399,000. And Nvidia isn’t just the hardware seller - every inch of it, every inch of this 300-server replacing beast, is created by Nvidia - from system architecture to system software . . . this is 90% less expensive, about 99% smaller, consumes 97% less heat than current cloud configurations, and is 500 times faster than prior AI learning systems. I haven’t actually seen it, but I assume this is not exactly a desktop system, but it brings AI development into the semi-personal domain by replacing 300 servers at a cost of $400K.

Just a few highlights. Someone in the thread mentioned that NVDA’s future still looks risky as eventually some disruptor will displace the GPU. I’m sure that’s true. It will probably be NVDA that does so. And, since they don’t own their foundry, when the world of computing move from silicon to optics and graphite, NVDA will not be caught by the legacy dilemma. It will be easy for them to move to the new technology.

NVDA is about a 6.5% position for me. I’m holding.

29 Likes

Nvda just broke out it is in nu hi ground y would u sell it now?

Pretty simple

1 Like