NVDA: math for 10x price increase, maybe...

I would suggest that the CEO’s prediction of growing to be 10x bigger is based on sales dollars being 10x greater, not market cap. We can expect NVDA’s PE to drop over time like any other successful company from its current 35 or so (forward PE) and a 10x market cap may not be realized if/when sales are 10x the current level. Not a certainty, just pointing out that is what happens over time.

Hi Rob,

Yes, I think Hwang probably meant revenue. So the question then becomes what a 10x increase in revenue might do to the share price:

  1. Let’s assume for argument’s sake that revenue can continue to grow at at the same full year revenue growth rate. If you average the past 5 quarters then you get about 45% growth per year. At this rate, they will grow revenue 10x in a little more than 6 years. I think that is a reasonable assumption given that:
    a) revenue is constrained because they have not been able to produce enough product and NVDA will likely solve this issue. Gaming GPUs are out of stock, Titan V is out of stock.
    b) data center revenue is growing like mad and 3 quarters ago NVDA had a transitionary period where they were priming the pump by signing up the all the Cloud providers and the system integrators. Now that they have all of this in place, they can really get cranking.
    c) the professional visualization business has been a slow grower for a while with the past with CAGR over the past 3 fiscal years (2015 - 2018) at 5.8% growth. I think that their RTX (real-time ray tracing) technology may well juice this segment’s growth.
    d) Jetson is for devices at the edge. This area hasn’t really stared taking off yet. When it does, I think it will be bigger than data center business.
    e) The automotive business line is about 1/10th the size of the gaming business. The automotive is mostly infotainment. Soon (in a few years but probably well before 6 years from now) the autonomous automotive revenue will kick in for NVDA.

  2. Yes, you are correct that if the growth rate slows then the P/E will probably get compressed. This would contribute to a compression to the valuation.

  3. However, if revenue is growing at 45% then earnings will grow even faster. The E part of the P/E would work to offset the multiple compression. So how much has earnings been growing the past 5 quarters when revenue was growing at 45%? The answer is 83% (that’s the average 1 year growth rate over the past 5 quarters). So can this rate continue if revenue continues to grow at 45%?

Now for the calculations: TTM EPS is $4.91 and the TTM P/E is currently 45. If we grow the earnings for 6 years at 83% then we get TTM EPS of $184. If we compress the multiple from 45 to 20 then we get a share price of $3688 which is 16.7x of today’s price of $221.05.

This may seem a bit outrageous but we will see what happens…



I’ve been looking at NVDA recently and boy, that’s some nose-bleeding valuation you got there.

I’m struggling to find good data on NVDA’s crypto-exposure.
Notably, how much of NVDA’s 2017 revenues were due to crypto-sales, and how many GTX 1070 and 1080 are in the hands of dedicated crypto-miners?

It appears that the crypto-bubble is ending this year, and that would mean that many/most of those cards will end up on ebay.

GeForce cards are approximately 60% of the company’s revenues. It seems likely that around half of that (and mostly the high-margin cards) are from miners. This is just a reasonable guestimate on my part, though.

Imagine what Nvidia’s 1Q18/2Q19 numbers will look like if those sales disappear while half of last year’s GTX 1070 and 1080 make a re-appearance on Ebay.
I know there’s pent-up demand among gamers who’ve been frustrated by high prices, but it’s probably not going to absorb THAT kind of supply.

Given the sky-high valuations, such a development, potentially set against a generally bearish market environment, would be devastating for the stock price.


Do we know that the market for used video cards is a good one? I personally would never buy a used video card. You have no idea what the person used it for before, and I don’t know if you could get it replaced under warranty.

I suppose there’s always some kind of market but given how quickly consumer video cards become obsolete I just don’t see a signification portion of their GeForce related revenue being affected by the secondary market.

Maybe it’s the ultra high end cards, but you just don’t need a Titan card to play most games.

Millions of iPhones hit the secondary market every year yet the new ones still sell ok.


Do we know that the market for used video cards is a good one? I personally would never buy a used video card. You have no idea what the person used it for before, and I don’t know if you could get it replaced under warranty.

I have both bought and sold used video cards on Ebay. Ebay currently has 751 listings for used GTX 1070 and they appear to sell at between 350 and 400 dollars.


Millions of iPhones hit the secondary market every year yet the new ones still sell ok.

Millions of video cards also hit the secondary market every year. This is usually not a problem, but it appears likely that the number of high-end cards sold on the secondary market will go up by multiples later this year as crypto-miners quit the business in droves.


There were several factors for NVDA’s growth on the gaming side.

  • 28nm to 16nm transition

  • AMD losing focus on discrete GPUs

  • Crypto market

  • 28nm to 16nm transition
    Before 2016, NVDA (and AMD) had been stuck on the 28nm process for several years. It just wasn’t cost effective to build bigger dies with more transistors. Once 16nm (14m, for AMD) became available at the merchant fabs, new GPUs were released. Transistor counts were up 2X-3X, with the same die size. Performance was up roughly 50%. This was the best price/performance jump in over five years.

  • AMD losing focus on discrete GPUs
    AMD was in dire straights from 2008 to 2012. Fabs were sold, HQ was sold, multiple rounds of layoffs, a couple of CEO changes and R&D cuts. AMD’s share of discrete GPUs dropped down to 20%. AMD did well in the mid-low end. But the high end, where most of the profits are, was left for NVDA.

  • Crypto market
    There was a small bubble (by today’s standards) in 2013-2014. AMD’s cards with a $400 MSRP briefly spiked to $1000. Wafer starts to shipping cards is about six months. AMD ordered a bunch of GPU wafers to meet the expected demands. The bubble burst, and AMD wrote off some $200M in unsaleable GPUs. The current bubble went much higher and lasted much longer than before. We’ve experienced a six month period where every GPU has sold well above MSRP. It’s not clear what will happen when the bubble bursts this time.

Note that “Gaming” still contributes to over 50% of NVDA’s revenues. However, the FY17 to FY18 gaming revenue growth was only 36%. Gaming has been a drag on NVDA’s overall growth rate. Total revenue growth was 41%, and all of non-gaming was 47%. I expect that non-gaming will be 50%+ of NVDA’s revenues by FY2020. And generally, the gross margins are currently better on non-gaming devices.

Disclamer: I am still long on NVDA. i reduced my holdings by two-thirds in January. The next 10x in stock price will be much harder than the last 10x.

Source for NVDA’s revenue breakdowns