Brad’s Note: Today, we’re sharing an essay from Jeff Brown, founder of Brownstone Research. I’ve known Jeff for years. And what I always appreciated about his research is that he (like me) comes from the industry he covers. My experience as a real estate developer and investor gives me a unique understanding of this sector that you just can’t get from sitting behind a desk. In the same way, Jeff’s experience as a technology executive gives him a unique insight into the technology markets. Also like me, Jeff is very bullish on the data-center space. Below, he shares why Nvidia’s recent $100 billion investment into new data centers is just the beginning. Get the full story below.
Go big or go home.
That should be the motto of this single moment of history in the making.
This once-in-a-lifetime mashup of electrons, silicon, and unlimited capital. This supernova of technological breakthroughs that is happening right now.
Whenever we see a development that makes our jaws drop, which happens every month these days, we can be sure there’s an even bigger development right around the corner.
Awash in Cash
NVIDIA (NVDA) recently announced that it is investing $100 billion into OpenAI – which will be used to build and deploy at least 10 gigawatts (“GW”) of AI data centers – which means 4 million to 5 million more NVIDIA graphics processing units (“GPUs”).
The numbers are so large, they are hard to comprehend.
To put that in perspective, a single gigawatt of power can serve around 750,000 homes. NVIDIA and OpenAI are going to build the capacity for 10 times that.
This announcement makes NVIDIA’s $5 billion investment in Intel (INTC) from September look paltry.
As I shared with my readers at the time:
In a press conference this afternoon, it was stated that the two companies have entered a product development partnership to design two new collaborations between Intel’s 86 central processing units (“CPUs”) and NVIDIA’s GPU technology.
I can’t imagine what Intel’s board of directors is thinking right now.
They probably feel a kind of regret and embarrassment, the kind that gives you discomfort in your stomach.
They should have negotiated for a much larger amount… NVIDIA clearly could have done a larger deal with Intel.
Three years ago, a deal like this for NVIDIA would have been impossible.
In its fiscal year 2023 (ended January 29, 2023), NVIDIA had just $13 billion in cash and produced $3.8 billion in free cash flow. It was super healthy financially, but there was no way it could pull off a deal like this.
It simply didn’t have $100 billion to give to a customer to buy its own products.
But today, NVIDIA is sitting on $57 billion in cash and in NVIDIA’s current fiscal year (ending January 31, 2026), will generate about $100 billion in free cash flow… in a single year.
Next fiscal year, NVIDIA’s free cash flow will jump by almost 50% to $148 billion.
It is awash with cash, and it pays almost nothing in dividends, just $834 million.
I’m sure some of us are wondering, isn’t this deal a bit shifty?
Isn’t this left-pocket-right-pocket movement of money? I’ll give you $100 to buy my $100 product.
Doesn’t that mean NVIDIA will go broke?
I wouldn’t blame anyone for the double-take.
And you’re not wrong.
But…
Be the Kingmaker
It’s just the wrong framework for viewing the deal:
-
First off, NVIDIA has the cash to make the investment. The $100 billion will be given to OpenAI in tranches, as it is needed for building AI infrastructure over the next few years.
-
It is an investment in OpenAI. NVIDIA will be receiving equity in OpenAI’s for-profit arm. NVIDIA will share in the financial upside of OpenAI, and almost certainly get its money back, and then some, at some future date.
Its first $10 billion will result in owning about 2% of OpenAI. OpenAI’s valuation will naturally increase over time, so each $10 billion invested won’t be worth an additional 2% ownership; it will be less as OpenAI’s valuation increases, but once all is deployed, NVIDIA should own at least 10% of OpenAI.
-
Given NVIDIA’s current market position is – what would be considered – a monopoly in GPUs for training AI in data centers, and as the most valuable company in the world ($5 trillion valuation), it has become nearly impossible for NVIDIA to make acquisitions of any company of size and importance. It would never get through anti-trust.
-
NVIDIA has zero interest in making large dividend payments during the largest infrastructure buildout in history, for which it is perfectly positioned for more growth, and to become the world’s first $10 trillion company.
That leaves NVIDIA with only one option…
Invest.
Spread the money around. Pick the winners. Be the kingmaker.
And that’s precisely what it is doing.
In September alone, NVIDIA:
-
Invested $5 billion into INTC
-
Signed a letter of intent to invest $500 million into Wayve, an agnostic, U.K.-based autonomous driving software company
-
Made a strategic investment in Eleven Labs, the leader in AI voice models capable of producing ultrarealistic voices, now indistinguishable from a human voice, in more than 70 languages
-
Announced it would make $2.5 billion of investments in strategic “partnerships” in AI companies in the U.K.
These companies will be successful.
Not only will they have the capital needed to grow, they’ll have the guarantee of having access to NVIDIA’s most advanced GPUs, which have backorders that run out past a year, as well as access to cloud services that provide NVIDIA-based computational resources.
But this is just half the story…
13 Times More…
Last month, OpenAI sat down to reveal its own vision for what it is planning on building.
Below is the first AI superfactory in Abilene, Texas.
The company has been constructing it over the past year.
OpenAI’s Abilene AI Superfactory
Source: Wall Street Journal
What we’re looking at above is 1,100 acres and eight interconnected data centers that require 900 megawatts of electricity to power. Raw horsepower. Texas tough. Go big or go home. (Note the plots already cleaned out for the next data centers, just behind these eight.)
And here’s the gut punch…
OpenAI CEO, Sam Altman, said that he needs 13 times this new capacity that we’re looking at above. 13 times.
And Altman also announced that OpenAI will be building five more new AI superfactories across the U.S. with partners Oracle and Softbank.
These data centers are linked to the $500 billion Stargate project that was announced earlier this year with President Trump.
OpenAI, just one company, is now committed to at least $1 trillion in investment in artificial-intelligence infrastructure. And it has the money and backing now to make that happen.
And he’s not the only one…
Abundance in Texas
When Elon Musk was asked about the latest development with NVIDIA and OpenAI, his answer was simple and concise.
Musk, and his team at xAI, will be the first to 10 GW of compute, the first to 100 GW of compute, and yes, even a terawatt (“TW”) of compute.
Musk and his businesses are all now based in Texas to take advantage of its business-friendly climate and its abundance of energy.
And needless to say, Musk has the capital, the intelligence, and the ability to raise as much as needed to win the race.
It’s not a coincidence that the great state of Texas just announced that it is investing $350 million to help facilitate nuclear energy projects in the state.
Texas’s energy wealth is abundant, natural gas is everywhere, and Texas is smart enough to realize that it will need even more than natural gas.
Roughly 100 GW of compute will be needed to achieve artificial superintelligence. Nothing, and I mean nothing, will slow down until that happens.
And trillions of dollars of additional investment will be needed to upgrade the U.S. power grid to support this hyper-acceleration that is unfolding before our eyes.
Opportunity is everywhere.
Go big or go home.
Jeff Brown
Editor’s Note: As Jeff says, the state of Texas recently made a critical change to its laws… In a quest to become the AI data-center capital of the world. Now, these giant buildings are popping up everywhere in Texas… And it’s creating a massive opportunity in the energy sector. Get the full story right here.
|
