Diving into NVDAs Trends

Investment Analysis Clubs Saul’s Investing Discussions
1

If you ever read my posts you know I’m a big fan of NVDA I feel their product offerings across all lineups are far superior and in some cases competition prohibitively superior. That goes for gaming, data center, provis, and maybe autonomous.

This quarterly report is worrisome for many reasons. We know the headline numbers and they were awful. Very much so.

Gaming awful. Can work through that but more on that later. But what’s worrisome is data center. Its not because Nvidia is getting disrupted by somebody. by any stretch. But maybe we are approaching what the market will support per quarter in accelerating hardware. Approaching, not reached. Like Saul says, in order to grow you have to sell the same as last year plus that much more.

Look at data center dollar add last couple of quarters. This is from previous quarter

Q1 18 +$113M
Q2 18 +$7M(what happened here?)
Q3 18 +$85M
Q4 18 +$105M
Q1 19 +$95M
Q2 19 +$59M
Q3 19 +$32M

So yes growing 58% yoy this quarter over year ago quarter, but if that trends into Q4 the growth rate comes down sharply. I loves me some Turing T4. I think it will do a lot of inference sales. But will Volta sales continue to ramp? You have to have same Volta or Volta next sales plus the T4 sales.

Second concerning thing is from 10-Q and they talk about it but the number is shocking. Inventory.

Well look for yourself

Start of 2017 inventory - $794M
Start of 2018 inventory - $796M
End of July Inventory - $1.09B
End of Oct Inventory - $1.417B

That is a 79% increase in 9 months. When 0 is normal. Huang. Huango. Huangenstein. Huangino. Buddy, what the heck happened?

I can understand mistakes. But Nvidia management has been near infallible. And that is part of thesis. And that is a HUGE problem. The $200M glut was talked about in Q2 during CC and Huang said we’d be good in a quarter. Then we found 40% more not 20% less. And a devistatingly short forecast.

This one hurts to admit. It will take more than a quarter to make that up. And the most promising short to mid term data center is MIA on the save.

Darth

Eating Crow Flakes for Breakfast tomorrow

51 Likes
2

Darth,

Nvidia missed by $29 million in its fastest growing and largest TAM market. They dominate that market. I do not recall them missing anything by more than $2-$3 million before. That is a very large miss. Did FPGA’s start to take more market share than anticipated? I don’t know.

The stock’s relative strength is also not real good this year, and it has become so cheap that the market (that got it right) really thinks it is just a bit better than Intel.

It is time to get back to the rules that work. You do know that for the last 2.5 years all the money was made in Nvidia when it was (1) expensive, and (2) had a 98% or higher relative strength from IBD. I just bought as it was going up (unlike lately when you had to buy because it just became too dang textbook cheap). Same with ANET, being as it went up. Sold it and did not buy it back when it got textbook cheap. All the money is lost when Nvidia is now textbook cheap and has a low relative strength. Pure might be the exception to the rule on Monday (they were last quarter as well and had a nice 30% rally - all of it and more given back since then). But the rules have proven their worth - an exception to them does not disprove them, but rather shows their worth.

In the end, even Nvidia, with all its dominance (and it is dominant), that dominance was not enough to overcome the markets indicators when it makes a stock (1) textbook cheap, and it trades it so that (2) its relative strength is quite low.

That is my lesson learned. Look at how simple Saul keeps it. My rules are that simple as well. But over the last few months (other than putting my largest holding Mongo away and not touching it - and it is at a near 52 week high despite the crash!), and now having TTD put away and not touching it, I became fickle with the rest. I played earnings twice in 3 months, and we just had to be right about Nvidia despite all the other indicators…I mean I doubt there are few of us who believe that Nvidia would not show the market the folly of doubting them! Expectations were set so low. I mean I don’t recall a stock like Nvidia where expectations were ever set that low. Even analysts were saying ignore this quarter, because Nvidia has such a bright future (And it still will have over time - but the speed of creating that future is material).

Nvidia will be back. And when it is no longer textbook cheap, and it has a much better relative strength, I’ll give it another look perhaps.

Until then, Saul has shown how simple it is, and it is pretty much what I have done over the last 2.5 years until the last few months. I put one stock away, and then a second, and got fickle with the rest. It was a 10% loss, so enough of a blow to be a good lesson, and sting a bit, but no biggie. We can all use a lesson in humility like that from time to time.

Tinker

15 Likes
3

I think NVDA is pretty new to this shift in demand. Look at their stock price over the 20 years…almost flat until 2016, then straight up based on hitting the market at the right time with the right products. But in the 2 years of growth, haven’t seen “bad” at all…

It appears to have caught them completely by surprise in how deep a slow down can be and how fast it can happen. Should they have seen it? Maybe. They SHOULD have known more about the markets they were selling into and what could happen if those markets turned off the demand.

I thought I remember NVDA talking about only a small 10-20% portion of their sales tied to crypto, but this sure looks like either it was a larger percentage or the slow down started mid Q3 and they made NO change to supply because the “orders” were still in place…or they simply started more wafers to build more inventory to support larger demand just as the demand dried up.

JI.

2 Likes
4

End of July Inventory - $1.09B
End of Oct Inventory - $1.417B

Does this inventory buildup break down the gaming vs DC inventory? The gaming inventory should resolve itself in 1 or 2 Q but I am more interested to understand if DC can maintain a high growth rate. A high inventory in DC will make it that much harder. What is the bull thesis for DC to maintain a 50%+ growth rate? I recall seeing a Barron article saying the cloud titans were reducing spend, from very high levels of course, but still a reduction in growth rate. And this seems to be confirmed by Nvidia’s DC growth reduction as well.