The Next NVIDIA: DRAM! Why I m Betting $10,000 On This Secret AI Memory Supercycle ETF

BizFinance Insights
3 followers
33 Video Views·May 22, 2026
AI Generated·May contain errors
Video Summary
AI GeneratedThe video analyzes the massive industrial expansion of AI data centers, identifying power and memory chips as the primary bottlenecks. It argues that memory has shifted from a cyclical business to a structural growth play, highlighting the Roundhill DRAM ETF and Micron as strategic investments to capitalize on the surging demand for AI infrastructure.
It's the hottest new ETF on Wall Street, and for good reason.
The biggest bottleneck in the AI build-out is actually memory chips.
DRAM ETF turned out to be one of the most successful ETF launches of all time.
Artificial intelligence promises to fundamentally rewrite the way that we think about modern economies.
I think that we're going to have a human-level performance on most, if not all, professional tasks.
So white-collar work, where you're sitting down at a computer.
Most of those tasks will be fully automated by an AI within the next 12 to 18 months.
That's Microsoft AI CEO Mustafa Suleyman's words, not mine.
When the CEO of a $37 billion a year AI business tells you that most of the tasks done
by 71 million white-collar workers in America will be fully automated within the next 12 to 18 months,
you don't call that a bubble, you call that revolutionary.
So revolutionary, in fact, that it's driving the single largest industrial build-out in the history of the world.
AI data center build-out has officially become the fastest, largest megaproject in the history of the United States.
With over $1 trillion of capex in under a decade, which is projected to grow to $7.6 trillion by the end of 2031.
And that's on an inflation-adjusted basis.
A project faster and larger than the U.S. interstate highway system, the Apollo program,
the space station program, and the build-out of the American railroads combined.
And memory remains the number one single largest bottleneck to the entire build-out.
Today, I will show you my $10,000 investment in the DRAM ETF.
The one and only ETF that properly tracks the entire memory sector.
I will show you how it has risen to be the fastest-growing ETF in the history of the United States stock market.
I will also show you my price target, my investment, my cost basis,
my in-depth thesis as to why I believe it is going to be one of the best investments over the next five years,
and we'll discuss the recent correction and see whether there's anything to worry about.
Moves like this are what has allowed me to more than double the United States S&P 500's performance year-to-date.
And in today's video, I'll show you how to do the exact same thing.
So without much further ado, let's go ahead and dive right on into it.
Our story starts with one name, Leopold Aschenbrenner.
You may not have heard of him, but you will soon.
This 25-year-old took a $220 million fund to over $5.5 billion in only a year.
In June of 2024, he published a 165-page essay titled,
Situational Awareness, The Decade Ahead, in which he made some very bold predictions.
First, artificial general intelligence, or AGI, would be achieved by 2027.
Second, he predicted that, quote,
by the end of the decade, they will be smarter than you or I.
We will have superintelligence in the true sense of the word.
And thirdly, he predicted trillions of dollars of industrial build-out in artificial intelligence,
saying, quote,
the most extraordinary techno capital acceleration has been set in motion.
As AI revenue grows rapidly, many trillions of dollars will go into GPU, data center,
and power build-out before the end of the decade.
The industrial mobilization, including growing U.S. electrical production by tens of percent,
will be intense.
I would like to remind you that this article was not written last week.
It was written almost two years ago in June of 2024.
Thus far, every one of his predictions, including the ones that we listed and the others in the paper,
have either already come true or are in the process of materializing.
His thesis of AI build-out is exactly what allowed him to propel his fund over 20X in one year.
And his new holdings just dropped.
In the most recent 13F filing, a striking change to his positioning took place.
First, the size of the fund grew in one quarter from $5.5 billion to over $13.6 billion worth of exposure.
Second, most of that jump in exposure has been led by massive purchase of put options
specifically on big names in the semiconductor space,
such as a $2 billion put on the SMH VanEck semiconductor ETF,
a $1.5 billion put on NVIDIA,
a $1 billion put on Avgo or Broadcom,
a $970 million put on AMD, and more.
These puts give him massive leveraged downside exposure to the semiconductor sector,
the very same sector that longing allowed him to 20X his portfolio.
But why would he do this?
Why would he pull a complete 180?
That's because the third realization of his 13F filing explains his largest plays
and leads us to the DRAM ETF, power and memory.
You see, in keeping with the thesis in his paper,
value was always going to flow through the value stack
to whatever part of the value stack happened to be the largest bottleneck at the time.
He's shorting the hyped chips while going long the bottlenecks.
In 2024 and 2025, the bottlenecks were chips.
This is why NVIDIA, AMD, and Intel rallied explosively.
But it's also the reason for almost $8 billion worth of put options
against them in the new filing.
You see, after their meteoric ascent, they are nearly priced to perfection.
It's not that they won't perform.
They likely will continue to perform exceptionally well
now that they've achieved blue chip status.
However, most of their parabolic 10, 20, 50, 100X growth has probably already occurred.
But as we've discussed earlier, AI doesn't just hunger for chips.
It ravens for an entire cooperating orchestra of hardware.
And the two biggest bottlenecks in the AI buildout right now
happen to be his two largest plays, power and memory.
In the power department, the fund's largest current long position
is an $827 million position in Bloom Energy,
a company which focuses in on-site power generation.
Connecting a power-hungry data center to the power grid
can take years of buildout, generation expansion, and permitting.
Bloom Energy's modular solid oxide fuel cell system
allows for on-site power generation and can be brought online
in a matter of months, not years.
In the memory department, we also have a $250 million position in SanDisk,
one of the leading companies in high-performance storage and memory solutions
critical for the AI buildout.
We also have a nuanced position in Micron,
specifically a $584 million put and a $422 million call,
signaling cautious bullishness on the high-bandwidth memory producer.
So what can we gather from this fund, its performance, and its outlook,
and how does it have anything to do with the DRAM ETF?
To start with, a focus on bottlenecks is what will lead us to finding
the best investments.
For years, the bottlenecks to AI, the greatest companies at the best prices,
were the chip manufacturers like NVIDIA, AMD, Intel, TSMC, and more.
It's not that these aren't still amazing plays.
They are.
But new breed of great companies at good prices are being discovered.
It should be obvious at this point that the next greatest plays in the AI boom
will be found down the value stack.
Now it's not just NVIDIA that's selling next year's production.
It's power and memory.
We're going to talk about power in a future piece of content,
but for today, I want to focus in on memory,
and specifically how to capitalize on it.
Think about your brain.
It has two major executive functions, processing and recall.
For your brain to function, it also requires raw materials, like food.
And after it's built to operate, it requires an energy input
in the form of calories.
An AI data center operates much the same way that your brain does.
To allow the AI to think, the data center requires compute,
like the GPUs and the CPUs created by NVIDIA, AMD, and Intel, and more.
To allow the AI to recall information, the data center needs memory,
such as the high bandwidth memory for on-ship near-memory
and longer-term deep storage.
Companies like Micron, SK Hynix, Samsung, and SanDisk
are the largest providers of all sorts of memory,
and their production build-out is significantly behind
the compute chip manufacturers,
leading to higher prices as demand spikes and supply remains low.
This leads to massive operating margins,
minuscule forward PEs, and explosive charts.
Finally, we also have the materials and the energy components
that our brain needs.
These demands of data centers are why names like Caterpillar
are exploding in the construction department,
and why names like Bloom Energy are exploding in the power department.
Suffice it to say, as AI gets exponentially smarter,
its demand for larger and more powerful brains will continue as well.
This will undoubtedly lead to more demand
in all four major quadrants of inputs.
And as long as that demand stays exponentially above supply,
which it most likely will,
prices will continue to soar.
We also need to understand how much more rapidly
AI is able to grow its intelligence
compared to how quickly companies are actually able
to build out the materials that are needed for their brains.
A frontier AI model might double its need for compute
within a matter of weeks,
but to build twice as much compute
might take a matter of years.
If you've studied economics for more than five minutes,
you'll understand that price is nothing more
than a reflection of the market's supply and demand.
And when supply is massively constrained
and bottlenecked by physical, real-world limitations,
and the demand for compute is able to double
in the matter of just weeks,
we have potentially one of the most
bullish stock market setups that humanity has ever seen.
This is exactly why Ashenbrenner
took out positions in energy and memory plays,
and it's why our topic of today's video, DRAM,
will likely perform so well.
But what is the DRAM ETF?
Why does it exist?
Investors, to some degree,
understand everything we just discussed
and are desperate for exposure to memory producers
as their operating margins continue to explode.
Enter the Roundhill DRAM ETF,
launched on April the 2nd of 2026.
Since its inception,
the ETF has exploded in value
from an original share price of $27
to an all-time high of $56 a share
just a little over a month after launch.
This meteoric growth has been driven
by its three largest holdings.
Number one, SK Hynix, with a 28% weighting.
Number two, Micron Technologies, with a 26% weighting.
And number three, Samsung, with a 21% weighting.
These three companies make up roughly 73%
to 75% of the entire ETF.
It's essentially a three-stock bet on memory.
Smaller holdings include Keoxia at 5%,
SanDisk at 5%,
and Seagate and Western Digital Technology,
each at about 5%, respectively.
In this one fund, you effectively have exposure
weighted largely by market dominance
to the entire memory sector,
not just in the United States,
but globally as well.
This allows investors not only to get access to Micron,
but also to the topic of last video,
which is SK Hynix,
and the other big name in memory,
which is Samsung.
Total exposure to the recall quadrant
we discussed earlier, all in one ETF.
AI demands a brain,
and not just a brain,
a continually evolving and ever-larger brain.
And the two biggest bottlenecks
to building at the brain that it demands
are power and memory.
Memory demand, as I'm about to prove to you,
is expected to astronomically explode
over the next few years
as AI data center build-out
continues to accelerate.
Michael Dell, the founder of Dell Technologies,
is now projecting a 625X demand explosion
for memory within three years.
Michael Dell, CEO of Dell Technologies,
spoke at a Bank of America event on the 7th,
stating,
As both per-accelerator memory capacity
and system scale expand simultaneously
in AI infrastructure,
total memory demand is forming a structure
where it increases roughly 625X.
He continued,
Expanding memory supply takes years,
but current AI infrastructure demand
is showing no signs of slowing.
We are still in the early stages
of technological adoption.
Dell explained that per-accelerator memory capacity
has grown from 80 gigabytes
with NVIDIA's H100 in 2022
to 2 terabytes by 2028,
while the scale of accelerator deployment
within data centers
is also expanding roughly 25X,
driving total memory demand
between 2022 and 2028
up to 625X,
25 times times 25 times.
625X from 2022 to 2028.
That level of growth, mind you,
is completely and totally unprecedented.
We have never seen anything like that before,
and it shows us that enormous memory growth
is likely to sustain
at least through the end of the decade.
But if that isn't convincing enough,
let's hear from Jensen Huang,
the CEO of NVIDIA,
who recently said this
to DRAM manufacturers
like SK Hynix, Micron, and Samsung.
He said less than a month ago,
Go ahead and build memory fabs.
However much capacity you add,
NVIDIA will consume it.
Those memory manufacturers
are the exact companies
that the DRAM ETF encompasses.
And that is exactly why
we have taken out a $10,000 position in DRAM
and also a $10,000 position
in one of its major holdings,
Micron,
which I'm going to show you right now.
Jumping over to the brokerage account here,
you can see that Micron and DRAM
are actually two of our largest positions
right now
betting on this memory supercycle.
So first, let's break down our position in DRAM
and take a look at the actual DRAM chart.
And then we'll also break down
our position in Micron
and break down some other AI and memory plays
that we've got like Microsoft, EWI, Soxel, Sandisk,
NVIDIA, and OVGO.
All right, so we have taken out
a $10,000 position in DRAM,
specifically a $10,390 position in DRAM
at an average cost basis of $51.38.
You can see that breakdown
in all of the entries and entry dates right here,
along with all of our total gains and total losses.
As you guys know,
one of the things that we like to be on this channel
is extremely transparent with what works
and with what doesn't work.
And there are times
that we're going to make bad investments
that aren't gonna pan out.
And there's something that can be learned from that as well.
But so far, I'm actually very happy with this position
and it's performing very well.
So far, as you can see,
our DRAM is actually down by a total of $132.
It's down about 1.28%.
And that's because of a correction
that has been taking place
over the last couple of weeks.
As you can see, DRAM has had a meteoric growth
from its opening around $27 a share
all the way up to a recent all-time high
of about $56 a share.
And over the last few days,
we have seen a major correction
right back down
to the 20 daily exponential moving average.
This is a very important moving average
that markets typically will correct down to,
and it's actually a very good sign.
Because so far, as you can see,
the DRAM ETF is actually bouncing off
of this technical support.
This means that the range around $46 to $50
is acting as a major support on DRAM
and likely will continue to act as major support
into the future.
Now, we're about to talk about Micron and SanDisk
and SK Hynix and Samsung
and many of DRAM's actual components,
but I want to take a second
and actually talk about
what I think the price target is on this ETF.
Because ultimately,
that's what everyone wants to know.
And right now,
the most important thing
that you can understand
about the Roundhill DRAM ETF
is that it has a forward weighting
of about eight.
And this makes sense
because the overall memory sector
And that's for two major reasons.
Number one,
earnings have exploded
along with operating margins,
which has compressed PE ratios
to be very, very small.
But two,
you also have a market
that still believes the myth
of memory being cyclical.
Memory was cyclical
when it was mostly used for consumer hardware,
such as in personal computers.
But that is not the business model anymore.
Micron, for example,
has gone from 15% of its business
being in data centers
just a couple of years ago
to over 65% of its business
being in AI data centers as of today.
AI data centers,
as we've discussed so far in this video,
are likely not going to be extremely cyclical.
Instead,
they're probably going to allow
for consistent, sustained,
and rapidly growing demand
over the course of many years,
if not decades.
When we talk about a super cycle,
this is probably not a cycle
that's going to slow down anytime soon.
And because of the massive supply crunch,
we're probably not going to see
a decreasing in prices
or a decreasing in operating margins
anytime soon either.
That means that this forward PE
is honestly probably only going to go down
if price stays the same.
In other words,
if profits continue to increase
and price doesn't move,
the forward PE is going to get smaller,
making today's prices
an even more attractive entry.
Which is why,
when we talk about the DRAM ETF
and we start talking about price targets,
they might sound insane,
but they're really honestly not.
And we're also going to pass
these price targets
that we're going to talk about
through the sniff test.
Does it actually make sense?
Is it actually possible?
We're going to be realistic
here in this video
because that's just the way we roll.
I personally believe
that in the next three years,
before the end of 2028,
this ETF is going to go
to well over $200 a share.
And no,
that's not as crazy as you think.
And yes,
that chart would look something like this.
That's because its constituent holdings
are tiny as compared
to what they should be valued at.
Micron's forward run rate
gives it an estimated revenue
of approximately $134 billion
based on Q3 2026 guidance
of $33.5 billion.
And revenue is actually
probably going to be
significantly higher than this.
And as you'll recall,
we are dealing with
almost 70% operating margins on Micron,
which is again,
one of the major holdings of DRAM.
And this is the exact same story
for the other two major components
of DRAM,
Samsung and SK Hynix.
Within just a few short years,
it's not just going to be
the hyperscalers in this top 10 list.
It's most likely going to be
memory providers.
And as a result of that,
we are already seeing
and we will most likely continue to see
a major re-rating in price.
So what would Micron,
SK Hynix and Samsung
have to do to cause
this kind of price target?
Well, as of right now,
Samsung Electronics,
SK Hynix and Micron Technologies,
the three largest holdings
in the DRAM ETF
are worth approximately
$2.84 trillion collectively.
NVIDIA, on the other hand,
is nearing a $6 trillion valuation.
NVIDIA, one chip manufacturer,
albeit the biggest,
is worth over double
the entire memory space.
For DRAM to go to $200,
all we need to see
is for these companies to quadruple.
But can they quadruple?
Can SK Hynix and Micron,
two companies that individually
will probably be making
as much money as NVIDIA
relatively soon,
quadruple?
Absolutely.
Micron could easily be
a $2 to $3 trillion business.
SK Hynix also could easily be
a $2 to $3 trillion business.
Samsung Electronics
just recently crossed the bar
into the trillion dollar space.
Quadrupling on the DRAM ETF
is actually not as far-fetched
as we think it is
when you consider just how
massively undervalued
these companies are.
For example,
at today's prices,
with today's growth,
and today's size
of the AI data center build-out,
NVIDIA's forward PE
is almost four times larger
than the DRAM ETF's forward PE.
If the DRAM ETF
re-rated to a 26x forward multiple,
then you would see price
go to well over $150 per share.
And again,
the only reason
that these companies
don't have 20x plus forward PEs
is because of this idea
that memory is cyclical.
Well, it's not anymore
because AIs are not going to stop
demanding bigger brains
anytime soon.
This might be one of the most
important points of this entire video.
The supply of compute
is constrained
by the laws of physics.
It is a physical constraint.
The demand for AI
does not exist
in the physical universe.
It exists in the digital universe,
which means that
its demand for compute
can scale almost infinitely faster
than the supply of compute.
Let's talk a little bit
about some of our other holdings as well.
Micron, for example,
being one of our largest positions.
So far, we are up 13%,
and even today,
we're up about 600%.
The current price
is around $720.
And on the recent correction
following our recent video,
we saw Micron drop
all the way down to about $650.
This is something
that people were very scared of,
but not us
because we're not buying
Micron the ticker symbol.
We're buying Micron the company.
I don't care if Micron
comes down here
and fills a gap at 590.
I don't care if Micron
comes down here
and fills a gap at 540.
I consider these to be
excellent buying opportunities
if they occur.
If Micron drops and revisits
former all-time highs around 450,
I'm totally fine with that,
and I would buy more
because you have to remember
one of the fundamental principles
that we operate by here.
We do not buy charts.
We are buying great companies.
Please remember my tenure.
I originally started out
as Crypto Jeb.
That's what that silver play button
is over there.
We've got over a quarter million
subscribers doing technical analysis.
I've been a technical analyst
for almost a decade.
I love reading charts,
and I'd like to think
I'm kind of good at it,
but I couldn't care less
what the Micron chart does.
I want to see
what their earnings do.
I want to see
what their revenue does.
I want to see
what their profits do,
and I've got every reason
to believe that they're only
going to continue to moonshot.
In other words,
if Micron does fall down
to different levels
that people have talked
about it dropping to,
for example, 600 or 550
or even down here to 480,
I consider all of those
to be excellent entries
because all those drops do
is give us better entries
at better forward P.E. ratios.
But it is also important
to note that right now,
Micron is actually finding support
on its 20 daily
exponential moving average,
which is very nice.
I also want to talk
a little bit about
some of our other positions.
One that I have not
mentioned recently is Chippy.
This is one of our
high income funds.
It spits off about 40% yield.
It's based on the
SOX Semiconductor ETF.
We've got $8,400 here.
Our average cost basis
is about $60 a share,
and it's currently trading
about $73 a share.
This spits off about $3,000
a year in passive income,
which is very nice.
We also have a $3,200 position
in Microsoft
that we acquired
at an average price
of $402 a share.
This position is a little bit
tangential to the point
of this video,
but I do want to point out
that effectively
all you have to do
when you're buying
the hyperscalers is rotate
from the most overvalued
of the Mag 7
to the least overvalued
of the Mag 7.
Microsoft and Meta
being the two most obvious
plays right now.
Microsoft only drops down
to its 200 weekly
moving average
every once in a while,
and when it does,
it ends up being
an incredible
generational entry.
And Microsoft sits right here
at that exact level
right now.
A massive rally on Microsoft
is likely to start
over the next several months,
considering they are
one of the biggest,
fully integrated,
and most rapidly monetizing
hyperscalers in the game.
Microsoft's AI copilot
could be pushed out
to over 20 million
paying customers
in as little
as the next year.
An honorable mention
would also be Meta,
which may not bottom out
anytime soon.
It may take it a little bit
for it to come back down here
to its 200 weekly
moving average,
but there's probably going
to be a bounce
happening relatively soon.
Probably seeing the massive layoff
that's taking place
in Meta right now.
While about 10%
of its workforce
is being cut,
this is not because
Meta's struggling,
but because the very
artificial intelligence
that they're investing in
is, as we've talked about
earlier in this video,
already starting to replace
many of their white-collar jobs.
This is most likely
going to lead to
even higher net operating margins
over the next couple of earnings,
which will lead to
compressing the forward PE,
which will make
today's entry on Meta
even better
just a couple of months
from now.
Meta currently sits
at a forward PE
of approximately 20,
which is a very low
forward PE
compared to its competitors
in the Mac 7.
Definitely tell me
in the comments section
if you guys think we ought
to load up on some Meta.
I'm starting to think
this is a great time to do so.
It's a fantastic business.
I also want to talk a little bit
about some of these
other positions.
We've invested $1,000
into NVIDIA
and $1,000 into Broadcom.
These are long-term holds.
I don't plan on
liquidating them
anytime soon,
and I do expect them
to continue to perform
exceptionally well,
as they are phenomenal names
that are necessary
in the build-out
of artificial intelligence.
We also took out
a $1,000 position
a couple of weeks ago
in SanDisk.
Most likely,
the next video
on this channel
will be a video
on SanDisk
because I actually think
that SanDisk is still
massively undervalued
at a price of $1,384 a share.
SanDisk is a perfect example
of a company
that has gone
absolutely vertical
and is probably
not done
anytime soon.
Stay subscribed
because our next video
or at least a video
at some point
in the near future
is most likely
going to be on SanDisk
as we will be talking
about its competitor,
more so a compliment
to high-bandwidth memory,
which is their
high-bandwidth flash,
which I actually think
could cause
their earnings
to explode
over the next couple of years.
Don't forget,
this was the only
major memory play
that Ashton Brenner
took an actual
stock position in
in his most recent 13F.
So that's our investment
and our cost basis.
Approximately $10,000
into MU
at an average price
of $637 a share.
Approximately $10,400
into DRAM
at a price of
$51.38 per share.
$3,000 into Microsoft
at $402 per share.
And then several thousand
dollars of investment
in multiple other names
that capitalized
on the AI and data
center build out.
I personally believe
that the DRAM ETF
is going to perform
exceptionally well
because I believe
its holdings,
Micron,
SK Hynix,
Samsung,
SanDisk,
Western Digital,
and all of the others
are going to perform
exceptionally well.
And I want you to be
able to take advantage
of that.
So if you'd like some
help taking advantage
of that,
I would love to sit
down with you
and do some financial coaching.
So if you would like
to sit down
for a one-on-one
with me,
then make sure
to send an email
to MalachiPatz
at CryptoJob.com.
He's my executive assistant
and he'd be more than happy
to get you set up
on a consulting call with me.
If you enjoyed today's video,
make sure to hit that like button,
subscribe to the channel,
and I'll see you guys
in the next one.
Peace.
