Hello everyone,
Taking a look at NVDA today, I see company's GPU segment makes up around 84% of the company's total revenue base, and EBIT contribution is over 100% because its "other" segment is EBIT negative.
Semiconductor company's operate in a very cyclical and competitive industry, and usually booms like this are followed by the inevitable bust as competition ramps up, production capacity overshoots, etc. etc.
I believe NVDA's massive run-up has a lot to do with GPU revenue growth, and according to an industry expert I happened to connect with, the recent crypto boom that we all witnessed was a great contributor to NVDA's massive revenue ramp...my question: is this sustainable??
GPU prices have fallen a great deal per many internet articles...yet the stock continues to hit new highs.
NVDA's other segment, "Tegra Processor (~16% of total sales)," seems to mainly include SOC modules used in the Nintendo Switch, along with automotive systems. NVDA just reported that this segment grew 40% in the recent quarter, with the automotive portion growing just 13% y-y (auto grew 75% in 2016). Auto makes up just over 5% of total sales. Is the majority of this segment's recent growth due to the Nintendo Switch?? I have no idea, but NVDA's disclosures seem to indicate that to be true.
This is just a bird's eye view. I haven't dug super deep into this idea, but from what I've seen, NVDA is starting to look like a short.
Appreciate any thoughts.
Thanks,
burrym
Hey burrym, the following topics might be helpful:
Calling relevant professionals! Jeffde23 BoutiqueButtMonkey madcat112
Hope that helps.
UPDATE: My thesis has evolved from the cryptocurrency overhang to one that is more focused on competition from ASICs, FPGAs, etc.
I am short NVDA, but right now my position is more delta neutral, because the bull thesis is still in force. When momentum shifts after the market starts to question the bullish sentiment, I'll position myself to be more delta negative.
UPDATE: AMD missed revenue guidance and most semis are getting taken out, with NVDA getting thrashed . I'm still short, although it's getting harder to justify my very conservative DCF valuation, but on a historical TEV/sales basis, downside is well below $100. I try to trade under the premise that markets overshoot to the upside and the downside, and technical support is still further down. Gonna ride the money wave, and see where I wash up.
Check the 8-k for the segment breakdown. You will see datacenter, gaming, autos, professional and other.
The notion that semiconductors have cyclical qualities beyond broader GDP/industrial growth has its roots in three issues, all of which are waning.
First, big product cycles drive semi content. Start with desktops, move to laptops, then to smartphones, all had their booms and busts. This will not be an issue in the future. Semi content is in "everything". Autos, industrial (factory sensors), all sorts of consumer hardware (smart home stuff), cloud servers and then all of the old stuff (computers, smartphones).
Second, the industry distribution structure was generally opaque. You had the actual semi companies producing chips 3+ months ahead of time, and shipping to local distributors, neither of whom had much visibility from the end customers. End customers are akin to consumers. They ordered what they could until they didn't need it, and then they cancelled if required. More semi companies now are more closely tracking inventory at distributors and/or limiting capacity additions even when demand is strong.
Third, the memory space is still cyclical. DRAM and NAND, while in high demand, also have significant fluctuations in supply. Pricing moves around a lot. We seem to be in the midst of a downturn right now.
Long story short, none of these issues really affect NVDA, which is a datacenter/ML/gaming secular growth play. The GPU distribution network is very different than for analog/digital ICs.
The key to the NVDA long is the innovation. They are producing massive performance gains from generation to generation for the most important applications (ML, gaming, virtualization), and they can price for it. That's what it's all about.
Yes, I agree with you that a large part of the bull thesis surrounds NVDA being the early entrant into ML/DL, but NVDA is definitely not the only player in this space, and the future of GPUs in this space seems to be in question.
Both Google and Facebook are developing their own AI ASICs. According to the CEO of Cerebras Systems, Google's TensorFLow has essentially commoditized NVDA's exclusive GPU ecosystem.
Intel, AMD, Microsoft, and a bunch of start ups are also trying to make headway into this nascent market. It is hard for me to think that NVDA's current 30%+ ROIC is sustainable with competition around the corner.
There is also chatter surrounding the massive build up of GPU inventory due to the boom and subsequent bust of crypto-mining. NVDA's management claims that "crypto-specific" revenue won't have a large impact going forward, but this disclosure seems to exclude all of the GPU's used normally for gaming that were also used in crypto-mining.
NVIDIA has brand name recognition and will always be considered a top tier item in their products (video cards).
Other companies can go on and develop their own products, but their brand name is not there (nor will I think it will ever get there). What wso23 is spot on about the company.
Some of Citron’s reasons included that Nvidia is a high multiple stock that is no longer beating Wall Street expectations by a wide margin, that current customers are developing their own AI chips, that tensor processing units are beginning to edge out Nvidia’s graphics processing units in cloud computing, and that short interest on the stock is at an all-time low.
Yes, I question the sustainability of the bull case that Nvidia will take over the AI space. Sell-side analysts are going crazy over the GPU and CUDA, and yet Nvidia's own customers are introducing their own ASICs and FPGAs in their own AI cloud services.
Backing out the current TEV via a reverse-engineered DCF implies some unrealistic revenue and margin assumptions.
With that said, NVDA has a clean balance sheet and is a good company. I've structured my current position largely delta-neutral with vol being pretty cheap just in case the sell-side continues the circle-jerk.
Still short. The stock seemed to mostly down with the broader markets, although news came out with the announcement that Huawei has it's own competing AI chip. China is around 20% of NVDA's sales, so this announcement is not immaterial.
Jesus. Had I put my entire portfolio into these put options, I'd be at the Porsche dealership tomorrow, and on a first-class flight to Tokyo the day after. It's never enough. Never, ever, ever.
Saw it's down 17% and came straight here. Really good call, maybe you should share your picks more often.
michael, is that you?
trolling value investors club or sumzero makes sense, but WSO?
Repellat nihil eveniet quia repellendus. Voluptas deleniti minima dolorem et sit tenetur. Similique doloribus qui placeat aspernatur suscipit beatae. Exercitationem corrupti enim est corporis. Voluptatem consequatur quisquam accusamus odio quo cupiditate.
Dolores est et dolores consectetur ut error repellat. Veritatis necessitatibus voluptate ut sed molestiae eum in. Rerum et delectus sint atque nobis. Praesentium consequatur ut odio laudantium quod et nihil voluptas. Recusandae et itaque doloremque sapiente laborum fugit aut. Et aut illum incidunt iure.
Tenetur possimus explicabo odit rerum et aliquid. Nesciunt totam dolor quos nesciunt quod. Sed fuga eveniet modi eius iste molestiae est occaecati.
See All Comments - 100% Free
WSO depends on everyone being able to pitch in when they know something. Unlock with your email and get bonus: 6 financial modeling lessons free ($199 value)
or Unlock with your social account...