Editor’s Note: Today, we share a special essay from Jeff Brown, founder & CEO of Brownstone Research, a corporate affiliate. Prior to publishing investment research, Jeff cut his teeth as a technology executive for firms like Qualcomm, NXP Semiconductors, and Juniper Networks. This experience gave him a unique insight into the technology markets. It also led him to recommend investments like bitcoin in 2015 and Nvidia in 2016. Today, he shows why Nvidia’s dominance could soon be challenged and gives details on a little-known company that could do it.
On February 6, 2016, I sat in the conference room of a resort on the Nicaraguan coast. Above, ceiling fans spun lazily (it’s a tropical climate year-round). Through the open windows, you could hear the faint roar of the Pacific.
I had been asked to present my best idea at an investor conference put on by a family office. Inside the room, there were a few dozen attendees.
They were about to get a whole lot more than their money’s worth…
After the host introduced me, I approached the podium and said (emphasis added):
[This company], at its core, is a semiconductor company.
[… ]
They have four primary verticals: gaming; pro-visualization, which is really, think about design or CAD-CAM, computer-aided design; data centers, which really encompass high-performance computing and things like artificial intelligence; and then the automotive industry.
I went on to explain how the company was governed by Moore’s Law – the gold standard of exponential growth. I showed how the market was dramatically underestimating its potential, how its share price would soar in the years ahead, thanks to mass adoption of cloud computing and artificial intelligence (“AI”) and not from gaming, which was at the heart of the company’s core business at the time.
Remember, this was nearly 10 years ago. AI was little more than science fiction. Hardly anybody expected it would become the technology it is today.
But in that same presentation, I said:
Here, we’ve got a billion-plus dollars’ worth of venture investment in artificial intelligence. If you add in the large corporations like Apple and IBM in 2015, there was more than $10 billion invested in artificial intelligence last year. When you have that much money, things move very quickly.
“Very quickly” turned out to be an understatement.
The $10 billion investment from 2015 looks quaint today. The large tech firms are now spending tens of billions of dollars quarterly to pursue their AI ambitions.
And the stock I recommended that day did fabulously – up some 18,000% since.
That company was, of course, Nvidia (NVDA).
Nvidia changed the landscape of the semiconductor industry. It became the workhorse for AI. And it became the most valuable tech company in history, now worth $3.6 trillion today. And, yes, it made many fortunes in the process, some of which now belong to my subscribers.
The attendees listened politely, but I don’t think they really grasped what I was trying to tell them. If I remember, some guy got up to use the bathroom halfway through.
Please don’t make that mistake…
Take just five minutes to read what I’m about to tell you. I believe it will be worth your while.
You see, a new company is set to rival Nvidia’s products. It could very well dethrone Nvidia as the go-to company for AI semiconductors.
And, just like in 2016, hardly anybody knows about it.
How Did I Know?
My name is Jeff Brown. For 25 years, I worked as a technology executive for the likes of Qualcomm, Juniper Networks, and NXP Semiconductors. Today, I’m the founder and CEO of Brownstone Research.
This hands-on, global experience in high tech, as well as being a prolific angel investor and investment analyst, gives me a unique insight into these companies. And it’s how I was able to pinpoint Nvidia’s potential all those years ago.
Back then, most analysts didn’t understand the company’s technology. And by extension, they didn’t understand its potential. How could they? Most analysts have never worked a day in the technology industry. And they don’t understand semiconductors.
Most analysts looked at Nvidia as a company providing graphics processing units (“GPUs”) for high-resolution video, mostly for video games. That’s how the company got its start. Most on Wall Street saw it simply as a gaming technology company. But they didn’t see that the parallel processing made possible by these GPUs was also suitable for machine learning and artificial intelligence.
Again, from my presentation (emphasis added):
One of the interesting things about the market, about Wall Street, is that they often misunderstand Nvidia. The reason is, they still have this old perception that it’s just a gaming company. But today, it’s very different.
[… ]
Google had a project called the Brain Project. They used 2,000 CPUs. And the task that it was given was to watch YouTube… So, this is an artificial intelligence watching YouTube. And its job was to classify cats and people, just by watching YouTube and videos. Nvidia ran a similar task, same thing, and they were able to do the same type of artificial intelligence, the deep learning, with only 12 GPUs in the same amount of time.
The pieces were there for anybody to put together. But to invest in the technology markets, you really have to understand the core products.
And after following one company for years, I’m convinced it’s Nvidia’s most worthy competitor.
A New King?
I have little doubt that Nvidia will be an important company for years to come. But with a market share in the AI semiconductor space of 70% to 95%, competitors are eager to chip away at that dominance.
I believe I know the one company that could actually do it. And it will do it by addressing a key weakness in Nvidia’s products.
You must remember that Nvidia’s products – like its current H100 GPU – are general-purpose semiconductors. They’re capable of training large language models, but there is room for improvement.
Let me show you what I mean…
OpenAI’s ChatGPT model is far too large to run on a single Nvidia H100 GPU. To run the model, OpenAI has to string hundreds of Nvidia GPUs together, which creates bottlenecks in the system and results in inefficiencies. And inefficiencies result in much higher costs for training.
It looks something like this:
But, Nvidia’s worthy competitor doesn’t have that problem because the scale is so different. Not only can ChatGPT fit on a single semiconductor, but much larger models – and future foundation models – will also be able to run on a single semiconductor from this company.
Again, it would look something like this:
This has dramatic advantages over Nvidia’s H100 in every category. And it gives the company making these products a large competitive advantage on a product-to-product basis. We can loosely think of this company’s products as roughly equivalent to about 50 Nvidia H100s. And this wafer-scale integration is what provides the business with radically better performance on a cost-per-unit-of-compute basis.
And when you’re spending billions of dollars on training an advanced large language model, there are very material benefits in terms of both costs and training time.
That’s what leads me to believe that this company has the potential to dethrone Nvidia as the leader in AI semiconductors. And, yes, I expect early investors could do well as these products are adopted.
If any of this sounds interesting, I’d invite you to join me next Wednesday.
On Wednesday, May 21, at 10 a.m. EST, I’ll be hosting a special presentation alongside Whitney Tilson, former hedge fund manager and editor of Stansberry’s Investment Advisory.
On that day, we’ll share our prediction for what we see coming to the markets on one specific day – June 2.
I’ll also give you more details on the company I mentioned above, Nvidia’s most worthy competitor.
You can get all the details right here.
I look forward to seeing you there.
We have so much to look forward to…
Jeff Brown
Founder & CEO, Brownstone Research
P.S. As a thank you for signing up for the event, I’ve also put together a special report on a next-gen artificial intelligence stock I call “Nvidia’s silent partner.” You can receive the report, completely free, right here.
|