Our 2020 piece on Investing in Artificial General Intelligence talked about the Bugatti of computing – Microsoft’s AI supercomputer which was solely developed for OpenAI. At the time, it was a single system using more than 285,000 CPU cores, 10,000 GPUs, and 400 gigabits per second of network connectivity for each GPU server, making it one of the top five fastest publicly disclosed supercomputers in the world. Just over two years later, ChatGPT emerged from that monster machine and took the world by storm.
Our article proposed a scenario that’s being considered by the world’s greatest minds – the emergence of an artificial super intelligence that will surpass humankind in intelligence and quickly become dangerously intelligent.
Our ability as humans to perform recursive self-improvement suffers from a hardware limitation – the human brain. After we achieve AGI, the only limitation will be how fast the biggest semiconductor manufacturer in the world, NVIDIA, can churn out GPUs.
Should that AGI decide to pursue its goal of increasing intelligence by any means, it will then begin to focus on how it can obtain more GPUs as quickly as possible. It may start sending emails directly to Jensen Huang, suggesting that perhaps they begin working more closely together in designing better and more efficient GPUs. Soon, it will start corralling the world’s resources in that pursuit. Humans who aren’t falling in line with this new global initiative might need to be disposed of. Before you know it, another pandemic hits, and the only people who stand a chance of surviving are those who are helping NVIDIA realize its mission. Fortunately, shareholders fall within this classification and their lives are spared, not to mention they become filthy rich and are often invited to the AGI’s extravagant parties.
NVIDIA’s Gaming Revenues
There are two reasons we sell a tech stock – if revenue growth stalls or our thesis changes. Here’s a look at how NVIDIA’s quarterly revenue growth stalled in Fiscal Q2-2023 leading to flatlined revenue growth for Fiscal 2023.
To understand this drop, we need to evaluate revenue growth at a segment level. Doing so allows us to see that the largest segment, Gaming, dropped dramatically in Q2-2023 (the dark blue bar at the bottom).
Additionally, Datacenter revenues have flatlined since then, and even saw a slight decline last quarter. Let’s start with the Gaming revenue drops.
The decline in Gaming GPU revenue was sharper than anticipated driven by both lower units and lower average selling prices (ASPs). Macroeconomic headwinds across the world drove a sudden slowdown in consumer demand. As noted last quarter, we had expected cryptocurrency money to make a diminishing contribution to Gaming demand. We are unable to accurately quantify the extent to which reduced crypto money contributed to the decline in Gaming demand.
An aptly titled article by Nikkei – Crypto crash a boon for gamers as graphics card prices plunge – talks about just how dramatic the price drop has been. “The price for graphics cards — a computer component crucial for data-intensive activities like gaming — has dropped about 60% from a 2021 peak amid a slowdown in cryptocurrency mining,” says the piece. During the crypto boom, NVIDIA had developed hardware specifically for cryptocurrency mining, so they’re taking a hit there, but their entire gaming segment is now suffering from lower average selling prices. What the company calls a “channel inventory correction” is actually a lower demand which also puts downwards pressure on ASPs.
Consequently, we might consider their gaming segment to now be working from a “new normal” baseline where growth will come from the release of new products vs. a bunch of crypto miners throwing money at hardware so they could print cash. Counting on a resurgence in crypto mining is pure speculation, as was the entire space as we warned in many research pieces. The emergence of generative AI is where we should look for future growth to come from.
NVIDIA’S Data Center Revenues
AI adoption is at an inflection point; the Generative AI opportunity is significant and driving strong growth in Data Center
NVIDIA CFO, Colette Kress
We’ve been eagerly watching the growth of NVIDIA’s Datacenter segment as that translates to the growth of AI hardware exposure which is why we invested in NVIDA to begin with. Unlike with Teradyne, our “skate to where the puck will be” strategy has worked out quit well with NVIDIA’s Datacenter revenues now representing 60% of total revenues.
Last quarter’s drop in Datacenter revenues was explained by “a timing issue,” as opposed to the usual “macroeconomic headwinds problem, but the bigger story is that NVIDIA expects to see tremendous growth going forward. To see the potential, you need to look past the endless cacophony of “generative AI is the way forward” emails that now flood people’s inboxes.
The AI Opportunity
Elon Musk once correctly criticized people who immediately support whatever cause is being crammed down their throats by the mainstream media. In the same way, we’re extremely suspicious whenever some investment theme starts being pumped by literally everyone. The AI opportunity has always been here, and prior to the emergence of ChatGPT, Mr. Huang described a $600 billion total addressable market for his company – $300 billion in hardware and $300 billion in software. In their recent earnings call, an analyst probed those numbers to see if last year’s TAM estimate has increased based on the hype around generative AI. Mr Huang’s response was to say that the TAM will simply arrive sooner, and likened the availability of ChatGPT to the smartphone app store. Now, everyone will be able to harness the powers of AI.
Remember the supercomputer Microsoft solely developed for OpenAI – one of the top five fastest publicly disclosed supercomputers in the world? It’s computing power like that needed to train models like ChatGPT. With the generative AI cat out of the bag, companies across all industries will start building “AI factories.” Companies will no longer just manufacture physical goods, they will manufacture intelligence.
Data comes in. That data center does exactly one thing and one thing only. It cranks on that data and it produces a new updated model. Where raw material comes in, a building or an infrastructure cranks on it, and something refined or improved comes out that is of great value, that’s called the factory. And so I expect to see AI factories all over the world.
Credit: Jensen Huang
NVIDIA isn’t just a hardware company, they’re a full-stack software company that will have their own AI models where AI is used to recursively improve itself. Given their leadership position, NVIDIA should continue to be a the forefront of AI innovation and continue to build bigger and better things to meet industry demand. It all sounds good on paper, but the proof is always in the revenue growth.
Stalling Revenue Growth for Fiscal 2024
There’s a concern that NVIDIA’s Datacenter growth won’t be enough to offset the decline in Gaming this year, such that the firm sees minimal growth this year again. For example, the company expects next quarter’s revenues to grow sequentially by about 7.5% to reach $6.5 billion. The black bar below shows that number along with that same sequential growth rate plugged into the remaining three quarters of Fiscal 2024 (in green bars).
The above would translate into year-over-year growth of just 7% for NVIDIA this year which means we’d then have two consecutive years of subpar revenue growth. However, this is where generative AI should start coming to the rescue with outsized Datacenter revenue growth carrying the entire company on its back. In other words, if we don’t see strong Datacenter by the end of this year, then Mr. Jensen’s vision of AI factories will be called into question, and the hype around generative AI might be just that – hype.
Going Long NVIDIA
While NVIDIA’s revenue growth stalled last year, our thesis has only strengthened as AI hardware exposure continues to increase and the sudden emergence of generative AI is expected to drive strong demand for AI hardware. Consequently, investors are pricing in this future growth as shares have soared +85% year-to-date compared to a Nasdaq return of +15% over the same time frame. Since we trim any position in our portfolio that exceeds a 10% weighting, we’ve been continuing to sell shares of NVIDIA and harvest gains from one of the most successful holdings in our portfolio. Once the market catches up to NVIDIA, or NVIDIA reverts to the mean, the weightings will adjust accordingly, but people on the sidelines are probably asking a more difficult question. Is now a good time to invest in NVIDIA? While some portfolio managers say, “how can you not own this,” that’s because they already do, and it doesn’t mean you should go backing up the truck.
The failed ARM acquisition aside, Jensen Huang has demonstrated that he’s capable of anticipating market trends successfully and reacting to them accordingly. If generative AI is expected to be the next big growth driver for artificial intelligence, NVIDIA is well positioned to reap the rewards. The demand for AI hardware isn’t going away anytime soon, so it’s reasonable to believe that NVIDIA could continue to be benefit from their leadership position well into the next decade. However, buying into the face of hype is never a good idea, and our simple valuation ratio shows just how overvalued NVIDIA has become.
We don’t invest in stocks with a simple valuation ratio greater than 20, and that’s where NVIDIA was sitting when we last looked in last year’s article on The Right Time to Sell NVIDIA Stock. Today, that ratio has soared to around 28 compared to October of last year when it dipped to around 12. Here’s how that compares to other larger stocks in our tech stock catalog using (the average across all 195 stocks we calculate this for is 6).
|NVIDIA CORPORATION (NVDA)||666,505||28|
|SNOWFLAKE INC. (SNOW)||44,207||20|
|MOBILEYE GLOBAL INC. (MBLY)||33,480||15|
|SAMSARA INC. (IOT)||9,811||14|
|INTUITIVE SURGICAL, INC. (ISRG)||87,830||13|
|CROWDSTRIKE HOLDINGS, INC. (CRWD)||30,568||13|
|SYNOPSYS, INC. (SNPS)||57,119||11|
|PALANTIR TECHNOLOGIES INC. (PLTR)||17,273||9|
|UIPATH, INC. (PATH)||9,225||9|
|ILLUMINA, INC. (ILMN)||35,081||8|
|OKTA, INC. (OKTA)||13,579||7|
|ADVANCED MICRO DEVICES, INC. (AMD)||154,838||7|
|TESLA, INC. (TSLA)||613,456||6|
|TERADYNE, INC. (TER)||16,758||6|
Of the 68 tech stocks in our tech stock catalog with a market cap of greater than $5 billion, NVIDIA is the most richly priced behind the king of rich valuations – Snowflake. If you’re looking to go long NVDA stock, simply set your own simple valuation ratio target and let that determine at what price you’ll go long. At a simple valuation ratio of 20, that means you’d only buy shares if at $193 or less. Set a target and stick with it. No FOMO.
NVIDIA’s Gaming segment may have had some overreliance on cryptocurrency mining embedded within which caused the entire company’s revenue growth to slow last year. The effects might be permanent moving forward, which is why it’s critically important for the Datacenter segment to take up the slack.
Mr Huang’s vision of AI factories being built across all industries to harness the power of generative AI offers promise that we might be at the cusp of the next big driver for AI hardware. Furthermore, NVDIA is building their own AI factories that will enable them to better harness the power of AI for their clients and democratize access generative AI through cloud offerings. NVIDIA may be overpriced right now, but that’s just a reflection of investors pricing in their future growth prospects. Prudent investors will set valuation targets and wait for the hype to subside. It always does.
Tech investing is extremely risky. Minimize your risk with our stock research, investment tools, and portfolios, and find out which tech stocks you should avoid. Become a Nanalyze Premium member and find out today!