Cramer: Why investors can trust that Nvidia’s stock is bottoming
CNBC’s Jim Cramer ticks down the reasons why he believes Nvidia’s call that its weakest end markets are turning positive.
Nvidia’s stock ended the fourth quarter of 2018 as the S&P 500’s worst performing stock.
But its latest earnings report was so upbeat that Cramer thinks the worst could be over.
From the article:
7:22 PM ET Fri, 15 Feb 2019 | 00:57
Investors can trust that Nvidia’s stock has bottomed following its upbeat earnings report for a few key reasons, CNBC’s Jim Cramer said Friday after the chipmaker’s shares gained 1.82 percent.
The first is CEO Jensen Huang’s own outlook for his industry, which he put quite simply on the post-earnings conference call: “The world needs more computing.”
“That simple statement is the main reason why I believe Nvidia can ultimately turn things around,” Cramer said on “Mad Money.”
AND
Add on the fact that Nvidia’s newest graphics cards, powered by its proprietary next-generation technology called Turing architecture, haven’t yet seen monster sales because Nvidia’s customers don’t seem to need that level of processing power, and the company’s outlook did seem quite grim.
But “the biggest theme on last night’s call was that, yes, things are about to normalize, first in gaming and then in the data center,” Cramer said. “There will be no comeback of cryptocurrency mining — that’s the chief reason Nvidia’s business fell off a cliff in the first place. But anything high-powered like artificial intelligence or machine learning or autonomous driving continues to percolate, and I think Nvidia’s the leader in these markets.”
Nice post. Unlike you, I did dump my shares, and at a 30% loss after a rather long holding period. Frankly I’m a bit surprised they have bottomed quite this quickly but I agree with Cramer on this. I sold at $144, they are already at $158. Sometimes… sometimes… buying and holding works. I’m just trying to figure out if I get back in.
Not sure about upbeat. It went from a crypto mining issue last quarter to every segment down Q/Q on this one. There are a lot of companies performing better than that.
NVDA traded at the same price the day after both earnings. So no loss holding from then until now. Except that during the same time:
I’m not quite as nimble as most others on the board, so my outlook tends to be a bit longer term. Most of my core holdings in NVDA are from 2015/2016 so looking at a broader view it’s more like a stock that got way ahead of itself (because of crypto) and is now back to kind of where it should have been in the first place.
I’m not able to follow along so closely to quarterly reports, press releases, etc. I bought NVDA thinking the same thoughts that Huang mentioned, the world needs more computing. Perhaps my thought process fits more with what Cathy Wood of Ark Invest describes with long term trends – data/AI, autonomous vehicles, robotics, genomics, etc.
Things don’t always fit a quarterly or yearly time frame. AAPL was able to release and sell a new iPhone consistently every year almost in the same month which was convenient. But NVDA doesn’t release new products every year. I know I never looked at a calendar and decided September would be my computer upgrade month. Gaming/AI/auto development doesn’t progress on a calendar based schedule (I see technological advancements as more of a punctuated equilibrium), so I can’t really worry about having consistent percentage growth quarter after quarter or year after year. The market of course plays by its rules but with this kind of investment I just look at the longer term trends.
If we could hypothetically take the crypto impact out, what would revenues (an stock price) look like? I suspect it would just be a gradual upward slope.
Of course it’s easier to say that when you’re sitting on a 400% gain (albeit down from 600%) rather than a 40-50% loss. I’ll be monitoring how things are progressing, particularly autonomous vehicles, especially given Musk and Huang’s previous comments that autos with “robo taxi” capabilities will be available this year (though I’ll give them until 2020). But I’m more likely to read through sequential year end reports than quarterly reports.
IRdoc Of course it’s easier to say that when you’re sitting on a 400% gain (albeit down from 600%) rather than a 40-50% loss.
No matter how folks say they can be objective in their investing decisions, it’s always psychologically easier to think, mentally, I’m working with the ‘house’s money’. The same thought process works here as well. (Personally, I was up ‘a lot’; now, after 27 months I’m up only 65%.
(Still, a lot better than the money sitting in the bank at a rousing 0.00001% interest.)
Going back many, many decades, virtually every single super successful company on the stock exchange went thru severe periods of up and down as they achieved their preeminence. Every single one. My thot’s are that this is one of the gap down times for Nvidia. Their technological moat remains intact.
A very poor financial report from the last quarter simply has not vaporized that moat. Period.
(They will very likely have at least one or two more ‘well down’ quarterly financial reports, based on YOY results.)
I’m staying, and hovering to look for times to add when they ‘gap down’ further.
Yet Tesla’s internally developed autonomous driving chip is, according to Musk 500% to 2000% better than the Nvidia Drive PX system used by Tesla so far.
Musk estimated on Tuesday that the chip would result in a 500 percent to 2,000 percent improvement over the existing system. The chip, designed by former iPhone chip designer Pete Bannon, is expected to process around 2,000 frames per second with full redundancy and fail-over, as opposed to the Nvidia chip’s rate of around 20 frames per second.
Musk estimated on Tuesday that the chip would result in a 500 percent to 2,000 percent improvement over the existing system. The chip, designed by former iPhone chip designer Pete Bannon, is expected to process around 2,000 frames per second with full redundancy and fail-over, as opposed to the Nvidia chip’s rate of around 20 frames per second.
Does anyone believe that Musk’s chip is going to be that much better than NVDA’s? This just sounds like more hype. How many times have we heard of a chip coming out that is going to kill NVDA? I can’t wait to see what this chip does but Musk in that statement seems to be exaggerating.
Does anyone believe that Musk’s chip is going to be that much better than NVDA’s?
I’d caution focusing too tightly on the tech.Tesla has designed a chip specifically to accelerate its software algorithms, apparently mostly around matrix math and convolutions performed on images. OTOH, Nvidia is selling a more general purpose AI chip. So, it shouldn’t be surprising that a company with talented engineers could design something to do one specific thing better than a more general purpose chip.
Trying to figure out whether Tesla’ singular-purpose chip is 2000% or 500% faster or only 200% faster is interesting, but not the point investing-wise. The thing about a Technology Moat isn’t just the technology aspect (you can get a taste of that here: (https://www.teslarati.com/tesla-patent-application-hardware-… ), it’s the business aspect. I totally believe Tesla would have had a hard time doing a better general purpose AI chip than Nvidia, but that wasn’t their goal.
What’s concerning about Nvidia now is that Tesla a) would bother to make the investment to build a better single-purpose chip, rather than either use Nvidia’s or work with Nvidia to make a better suited chip for Tesla’s purposes, and b) has been successful in a relatively short period of time. It shows, perhaps, that volume purchasers are willing to go beyond just waiting for Nvidia’s more general purpose tech to get better. For me, when it was just Google doing their TPU thing it was easier to ignore, as Google invests in all kinds of things that may not make short/mid term business sense. But now we see a company selling a retail product going outside as well. While Datacenter applications need to be general purpose, this may be a sign that volume AI applications will need/want more highly tuned chips for their specific purposes.
So while there may be some concern about the Tech in Nvidia’s tech moat, the large issue may be whether people are still trying to get into that castle, or have just decided to build their own?
I’d caution focusing too tightly on the tech.Tesla has designed a chip specifically to accelerate its software algorithms, apparently mostly around matrix math and convolutions performed on images. OTOH, Nvidia is selling a more general purpose AI chip. So, it shouldn’t be surprising that a company with talented engineers could design something to do one specific thing better than a more general purpose chip.
Bam. That. Plus I’ll add that a GPU has lots of circuitry that is necessary for rendering frames but useless for doing matrix math. Think circuitry for discarding hidden polygons as just ONE example. Tesla’s chip will have none of that, freeing transistors for other uses.
But now we see a company selling a retail product going outside as well. While Datacenter applications need to be general purpose, this may be a sign that volume AI applications will need/want more highly tuned chips for their specific purposes.
Those of us in computer architecture know that the existing architectures (Intel, AMD, and gulp, ARM) have run their course. (Maybe Larry was right to abandon SPARC after all…). The GPU is a very, very different animal than the CPU. And even more different architectures are probably on their way. This here is a great link as to why, from one of the heavy weights in the industry.
When performance doubled every 18 months, people would throw out their desktop computers that were working fine because a friend’s new computer was so much faster.” But last year, he said, “single program performance only grew 3 percent, so it’s doubling every 20 years. If you are just sitting there waiting for chips to get faster, you are going to have to wait a long time.” For a computer architect like Patterson, this is actually good news. It’s also good news for innovative software engineers, he pointed out. “Revolutionary new hardware architectures and new software languages, tailored to dealing with specific kinds of computing problems, are just waiting to be developed,”
Software will eat the world, be the key factor of expanding use of computing, for most things the hardware does not need to get better. If that is so, we here seem to be on the right side of the future.
Tesla has said the eyeballs (sensor) part of autonomous driving is already good enough, what is needed is more brain (chips and software) If GPU and CPU are nearing their limits improving the hardware part can only mean specialized chips.
What’s concerning about Nvidia now is that Tesla a) would bother to make the investment to build a better single-purpose chip, rather than either use Nvidia’s or work with Nvidia to make a better suited chip for Tesla’s purposes, and b) has been successful in a relatively short period of time.
Musk estimated on Tuesday that the chip would
That’s my point Smorg. Tesla hasn’t developed the chip, this is just Musk saying they are going to have it. How many times have we heard Musk over hype and not deliver? Until we see the chip we will not know if he has been able to develop it. Musk has done some pretty incredible stuff but he has also missed the mark a few times. I don’t get to excited about people coming out claiming they have a better chip then NVDA because so far they all have fallen flat. The future is hard to predict.
Tesla has said the eyeballs (sensor) part of autonomous driving is already good enough, what is needed is more brain (chips and software) If GPU and CPU are nearing their limits improving the hardware part can only mean specialized chips.
That is a good point Mauser, and if that is the case then maybe companies building their own chips to work with their software is the only way to go. That would be the end of the Generalization of chips.
Nevertheless, the point about a chip built for a specific purpose vs one that is broad spectrum is valid. I have thought for some time that the big challenge to GPUs in AI would be ASIC built for specific purposes.
The article also does a little editorializing like we have: With sufficiently deep pockets, anyone can create a custom ASIC that beats general-purpose hardware for a narrow application. The question is whether the strategy will pay off. With deep learning algorithms still evolving at light speed, it can be risky to lock down hardware functionality if you’ll need to change out the silicon a year later, when the algorithms refresh.
In the article: "Nvidia emphasizes the P40’s ability to accelerate both phases of deep learning.
Which gets back to how things will work. Tesla cars don’t train - they send data back to the cloud where the training is done. What Tesla needs to autonomous driving is image processing and inferencing. They don’t need to do both. This could be the scariest part of Nvidia - that the number of chips needed for training is orders of magnitude less than the number of chips needed for inferencing. Tesla doesn’t want each car to learn something different, they want to gather the learnings from all cars so that, for instance, a car that’s never been in the snow knows how to deal with snow the first time it sees snow.
If Nvidia’s chips are mostly for training, or the inferencing is harder/more expensive due to the training ability, Nvidia could be at a business model disadvantage. Tesla could buy a few hundred Nvidia chips for training, and then put hundred of thousands of its own chips in its cars for inferencing.
Tesla hasn’t developed the chip, this is just Musk saying they are going to have it. How many times have we heard Musk over hype and not deliver? Until we see the chip we will not know if he has been able to develop it.
Yes, Tesla has developed the chip. The new boards with new chip(s) are already in hundreds of employee cars being road tested. They are scheduled to go into regular customer new cars shortly (some claim they already are, but that has not been announced). They will also be retrofitted to all eligible Teslas (all cars built since October, 2016 have auto-pilot 2.0 or later hardware) that got the full self driving option. Also to all eligible cars that elect to upgrade to full self driving now.
This is way past “Musk saying they are going to have it”. Very soon, and very widespread, meaning most of the existing fleet, including all Model 3s. I expect my cars to up upgraded before summer.
Musk estimated on Tuesday that the chip would result in a 500 percent to 2,000 percent improvement over the existing system. The chip, designed by former iPhone chip designer Pete Bannon, is expected to process around 2,000 frames per second with full redundancy and fail-over, as opposed to the Nvidia chip’s rate of around 20 frames per second.
It’s hard to believe that anyone would put ANY stock into anything that Massah Musk would claim, unless he’s performing as a barker in front of a tent offering highly trained elephants, two headed snakes, and scantily clad women.
It’s hard to believe that anyone would put ANY stock into anything that Massah Musk would claim,
Except that it is highly likely that at general purpose GPU, much of whose power is directed at graphics, might be outperformed at the specific task of directing automated driving.