NVDA: What's happening


NVDA’s GTC conference started this week. The above link is to the recording of CEO Hwang’s keynote presentation. If you are an NVDA investor or are considering investing you must watch this 2.5 hour video. All I can say is WOW, HOLY MOLY!!

There were so many product introductions and announcements that it left my head spinning. This company is innovating so fast that I can hardly believe it. The recording is a must watch, but here are a few highlights that I can recall:

Introduced the next version of Volta 6 MONTHS AFTER VOLTA WAS INTRODUCED. this new version is twice as powerful.

Demonstrated real-time ray tracing technology that can make photorealistic images at 60 frames per second compared to the current one frame per 10 hours that the industry currently can do. Hwang said that this has been the holy grail for 40 years and is now possible. There are many applications that will use it.

Announced that they now have 370 partners in the automotive sector that are developing on NVDA’s chips for autonomous vehicles. This is up from 220 last year and 320 recently!!! They also announce the next 2 versions of chips that will be used in autonomous vehicles. This is just amazing progress.

They demonstrated video from autonomous vehicles drive around which is pretty amazing.

They introduced Constellation which is how they will test billions of miles of AVs in a virtual world. Twenty cars can only test a million miles in a year in the real world so virtual testing will be required to test safety. There are 10 trillion miles driven per year and it will be impossible to test enough miles by actually driving. There are 770 accidents per 1 billion miles. AV will be much safe than humans driving around.

I have little doubt that NVDA will enable very fast development and introduction of level 5 vehicles in a few years. With 370 partners who is left to use other hardware/software? Intel has little chance in my opinion.

NVDA introduced CLARA a cloud based AI system that can be connected to the install base of older medical instruments such as ultrasound and many others. Hwang said that there are 3-5 million of these legacy instruments and that only 100,000 are replaced every year. CLARA will automatically “upgrade” these old system and instantly provide them with additional capabilities. Amazing to watch the demonstration.

New version of TENSOR was announced.

NVDA uses 660 petaflops of commuting power internally and they will be upgrading that to more. The world’s most powerful supercomputers have about 100 petaflops. NVDA basically has the world’s most computing power.

They demonstrated a holodeck with a driver that was driving a car in the real world from the holodeck. This was like the movie Avatar! Incredible!

They demonstrated distributed computing where 4 GPUs running “on-prem” where connected to the cloud to speed up the processing by a huge amount.

They announced that Titan V is out of stock and they are try to more build ASAP. A few days ago that announced supply issues with GeForce. It looks to me that NVDA is moving so fast and that their products are so popular that they can’t make enough or they underestimated what was need to satisfy all the demand. NVDA will solve this, but it tells me 2 things. First, NVDA’s technology is advancing at a blistering rate and maybe even they are underestimating how fast they are changing the world and directly influencing progress. Second, it appears that NVDA’s Q1 results which will be announced in early May will blow away estimates (my opinion).

There’s probably more stuff that I left out. The video is a must watch!

I am now more confident that NVDA will continue to grow. How much I don’t know, but I’m more confident than before that NVDA may well become the world’s largest company by market cap within a decade. And during this keynote the stock started falling. Sure it fell with the rest of the market, but still. All these announcements and it dropped? I saw this as an opportunity and I couldn’t resist adding to my position.




I have a strong feeling that the irony of the NVDA share price drop coinciding almost perfectly with the start of this keynote could be quite laughable within the not-too-distant future.

Given the tinderbox that the match of the report about NVDA’s temporary halting of on-road tests was thrown into, the drop wasn’t wholly illogical. http://discussion.fool.com/nvidia-and-pedestrian-death-33025899…

That said, I could see the drop seeming particularly ironic in the future (despite likely fading into a rather small blip).

long NVDA
amused by irony

Introduced the next version of Volta 6 MONTHS AFTER VOLTA WAS INTRODUCED. this new version is twice as powerful.

Very interesting. I think it is appropriate to share a Ray Kurzweil observation here.

Most people on this board are probably familiar with Moore’s Law. Gordon Moore was a previous CEO of Intel and he famously stated that we could expect computing power (determined by the number of transistors on a chip) to double roughly every 18 months.

Kurzweil went on to say that the dawn of the computing era actually dates to the late 1800’s when the US had its first automatically tabulated census. It was mechanically done, but it was an early computer nonetheless. From that time until the late 40’s or early 50’s, a span of roughly 50 years, computing power doubled every three years.

Over the next twenty five years or so computing power doubled roughly every two years.

Next came Gordon Moore and his postulate.

At some time in the late 90’s or early 2000s it started doubling every year.

Now it is every six months.

Nvidia’s incredible 6-month gain is by no means an endpoint, but just an example of a trend. I expect these increases in processing to go from six months, to three months, to 9 weeks, etc…

This will be incredibly exciting to watch.

Don’t blink

Explorer Supernaut
You can see all my holdings here: http://my.fool.com/profile/TMFJebbo/info.aspx


I have found that for investing, getting stuck in all the minutia of new products, technology, cool stuff, etc. is fun, but not helpful in actual investing. Same thing I do for a living, I take a data center load of data and toss out 90%+ of it, and put on a case with the only things that matter.

Investing is like that as well.

The smallest gist is earnings. Looking at forward earnings, if you go out to 2019, I am getting P/Es for ANET and NVDA as low as 20 or below, and as high as 24. Of course forward looking prognostication all depends on the forward looking events prognated upon coming to happen.

But, this being said, anyone here thinks NVDA or ANET will be selling for 20-24x in 2019 if growth continues as modestly expected within analyst parameters?

Reason I am rather not anxious about the events of late. The market goes from looking at things 6 months out or shorter to looking at things 5 years out or longer…currently, despite the market being “frothy” as some would say, it is nothing like that for stocks like Nvidia or ANET this point in time, one just needs to look out to 2019, which I guess is an eternity for anyone trying to trade in and out of stocks.



The smallest gist is earnings. Looking at forward earnings, if you go out to 2019, I am getting P/Es for ANET and NVDA as low as 20 or below, and as high as 24…But, this being said, anyone here thinks NVDA or ANET will be selling for 20-24x in 2019 if growth continues as modestly expected within analyst parameters?

I don’t know where you continue to get your erroneous valuation numbers, Tinker, but they always seem to be 50% lower than the standard fwd estimates.

Bloomberg EPS adj for ANET is 31.5x for 2019.

For NVDA it’s 34x EPS adj for 2019.

Not anywhere near 20x.

Needless to say, GAAP is higher.

Hope is not a strategy.


Seems rather that my numbers end up being far more accurate however.

Hope may not be a strategy, but living your life and investing forward using fiction is even less of one.



To get into more details of analyst earnings estimates. The present analyst average estimate on Yahoo is $6.91 for 2018 for Arista (29 analysts). That would be a whopping!..suspense, suspense, suspense…17.82% YoY earnings growth. Thus Arista is expected to grow earnings at approximately 50% of the rate of its revenue growth (and less than half of expected revenue growth for Q1). And remember Arista very conservatively estimated their revenue growth for the year, but yet far higher than the analyst earnings estimates, despite how profitable Arista is.

Last year earnings estimates (from memory) were about $3 a share, some higher, some lower perhaps $3.50 or $2.50, or perhaps even $1.91 and it ended up at $5.61.

But if you want to live with Arista growing earning at 17.82% for this year…I’d rather look at the real world.

Analysts have to know Arista is going to grow earnings faster than this absent some horrific things happening, but yet they keep it as it is. You can ask them why.



Just to run the numbers.

Arista $5.61 earnings 2017. Nothing one time or extraordinary happened other than margins did increase. Margins probably will not increase every quarter this year, but they won’t decline by that much either.

35% earnings growth from $5.61 is $7.57, and 35% earnings growth for 2019 is $10.22. That is conservative for Arista.

Cash should grow around $800 million in 2018, and even more in 2019. That gives them $2.3 billion cash at end of 2018, and more than $3 billion in cash at end of 2019. The enterprise value continues to decline while the earnings continue to go up.

So let me know what is more probable, 17% earnings growth or 35% earnings growth when Arista is projecting faster than 40% growth for next quarter and 25-30% for each quarter for the rest of the year, with long-term growth of 25%. Usually quite conservative, but even if not.

Not being real aggressive here, just being conservative based upon actual numbers. Analysts have to believe that either Arista’s margins will be cut in half, or the Arista is being so aggressive in their forward guidance that they are overstating it by 2x.

We shall see, as time always moves forward.

There is one analyst with $7.90 estimate for 2018. What I think is happening is that not all 29 analysts seriously cover Arista, or they have not updated their earnings, or some such thing, as absent some awful stuff that no one has prognosticated, even the “resurgent” Cisco, no competent analyst can think that Arista will only grow earnings by 17% (or less - yes, 17% is the average, meaning a good chunk of analysts have Arista growing earnings at much less than 17% - in fact, the bulk of analysts have to be coming in much lower than 17%, because the analysts estimating higher growth are estimating far higher growth than 17% and mathematically it does not work without the bulk of analysts estimating less than 17% growth).

So analysts ----> non-analysts. Whomever you trust more you should invest with.



So analysts ----> non-analysts. Whomever you trust more you should invest with.

Hi Tinker,
Here were my thoughts on that (taken from my end of the month summary).

Did you ever hear of an economist who got rich investing in the stock market? Or an analyst either, for that matter? If the analyst could do it successfully, he’d be investing for himself instead of working for a salary for a brokerage company.