FairytaleDesign/iStock via Getty Images
Summary of the thesis
NVIDIA Corporation (NVDA) has, rightfully, garnered considerable interest following its latest earnings report. NVDA is leading the AI revolution with its best GPUs and related software. There isn’t much competition at this point for NVDA, but that will eventually change, as the opportunity presented is too great.
I believe Alphabet, Inc. (NASDAQ:GOOG),(NASDAQ: GOOGL) is in a great position to challenge NVIDIA, as it has the right incentives and the necessary resources. Google is already developing and using some products that will compete with NVDA’s chips.
It will take time and a lot of money, but the rewards for Google are just too great not to give it a try.
There is only one NVIDIA, for now
In a recent conversation with fellow SA Trading Places Research contributor, we discussed the implications of AI technology, how it works, and why NVIDIA is so far ahead of the curve. While NVDA has led many other chip makers in its rally, the truth is that NVDA sets itself apart by offering a unique product that gives it monopoly control over the market.
But before we get into that, we need to understand a little better what the product is here. Graphics processing units were originally designed to speed up the rendering of 3D graphics, but today they do much more than that. Today GPUs can break complex problems down into smaller, more manageable tasks and solve them simultaneously.
But the hardware (GPU) is only half the caution. NVDA’s success in this area is due to its combination of hardware and software. NVIDIA doesn’t just sell chips; it also has best in class software to go with it called CUDA.
CUDA is what allows NVDA to program its GPUs to go far beyond rendering graphics. CUDA enables its GPUs to tackle general purpose computing problems. In other words, the GPU (hardware) holds the power and the software (CUDA) enables it.
Nvidia starts making software and hardware that work together to solve all these big problems, and now they’re not just making these graphics-friendly quote gaming GPUs for their old purpose, they’re making data center GPUs that are very, very, very , very expensive and all the software that comes with it and it’s a complete system and it’s the only one because they’ve been doing it for so long it’s the only complete system of software and hardware that works together and does these things right, right.
Source: Podcast for Pragmatic Investors
NVDA is pretty much the only option for big players like Alphabet, Inc. and Microsoft (MSFT) to actually develop commercially viable AI, but it’s costing them a pretty penny.
Proof of this can be found in NVDA’s GPU specs. However, for those investors not so savvy in cutting-edge technology, the proof is also evident in NVDA’s earnings and guidance. NVDA is responding to a real surge in demand for its chips. And that’s not an increase that we’re seeing across the board, but a growing specific demand for NVDA’s chips.
Other chipmakers like Advanced Micro Devices (AMD) and Intel (INTC) won’t benefit as much from the growing demand from AI applications because their GPUs are simply years behind NVDA, which has been developing this technology for over a decade.
Google is ready to intervene
But if traditional chip makers won’t compete with NVIDIA, who will? Maybe someone with deep enough pockets and big enough interests. This is where Google comes into play.
Google has a lot to gain and lose from AI. The introduction of AI technology is the natural next step for both Google Cloud and Google Search and something that has already been introduced in the last year.
But as mentioned above, this technology requires expensive hardware and software, which mostly comes from NVIDIA’s incredibly expensive GPUs. That is why Google has already invested heavily in developing their own solution to this, which they have, and it is called TPU.
Tensor Processing Units (TPUs) are application-specific integrated circuits (ASICs) developed by Google and used to accelerate machine learning workloads.
Source: Google
These TPUs, also known as AI accelerators, can be used, as advertised, to accelerate the training of large machine learning models. Google is in its fourth iteration of TPU, and this is what Bard works on.
This is just one of two ways Google is fighting NVIDIA’s “AI monopoly.” At its latest I/O conference, Google unveiled Palm2, its latest large language model. The key here, though, is that Google presented the Palm2 as a family of models, varying in size.
What we’ve found in our work is that it’s not quite the model size type – that bigger isn’t always better,
Source: DeepMind VP Zoubin Ghahraman
There is even a PaLM 2 model that can run on smartphones. This is important because the size of the models is one of the limitations today. The bigger the model, the more GPUs. Cutting down on size while maintaining functionality is another workaround.
Market opportunity
Google is in a good position here to challenge NVIDIA. The company has the right motivation, is already showing increased attention in this area and has the capital to recover. To put things into perspective, Google’s R&D budget for 2022 was $39.5 billion, compared to NVDA’s $7.34 billion in the same period.
Google has here the opportunity to enter a segment that is expected to grow at a CAGR of 25% over the next decade, reaching a total market of 400 billion dollars, according to Global Market Insights.
GPU Market Forecasts (Global Market Insights)
And this is just an idea of what Google will earn if it commercializes this technology. However, the biggest benefit for Google would be to power its suite of AI applications in-house. At the very least, much cheaper and, in the end, with better technology than NVDA provides.
Other considerations
Having said that, we have to understand that there are other important players in this race. Microsoft is one of them, as is Meta Platforms (META), which recently unveiled its own AI chip.
Finally, I would like to conclude with a word of caution. While AI has a lot of potential, companies should also be careful not to give AI too much responsibility too soon. Companies like Google need to be aware of the limitations of AI, as well as its possibilities.
Large language models are very good at predicting the next thing to say, but still lack the reliability required in a commercial setting.
Final thoughts
There’s no denying the growth story behind NVIDIA, but that doesn’t necessarily justify today’s valuation. While the stock could easily continue to rise, this is a speculative move now. A more sensible bet, at this point, could be to focus on Google’s ability to disrupt the GPU market. As mentioned above, they have the necessary means and motivation. Artificial intelligence is a complex topic, so investors need to work hard to understand it before investing in it.
#Google #Stock #Real #Threat #NVIDIA #NASDAQGOOG