00:00:00
artificial intelligence. It's
00:00:01
everywhere. It exploded, reshaping
00:00:04
industries, promising utopia or maybe
00:00:07
even chaos, depends on who you ask. And
00:00:09
for the past 2 years, two companies have
00:00:11
been driving that explosion. Nvidia and
00:00:14
Palanteer. Their stocks, well, they went
00:00:16
vertical. They became synonymous with
00:00:18
the AI gold rush. But now, well, the
00:00:20
markets are getting a little shaky, and
00:00:22
Nvidia has consistently been taking hits
00:00:24
all year on its stock price. And you're
00:00:26
probably sitting there, I don't know,
00:00:27
maybe looking at your portfolio and
00:00:29
wondering, all right, which one of these
00:00:31
is the real deal? If I have to bet on
00:00:33
the future of AI, where do I put my
00:00:35
money today? Look, they both have
00:00:37
massive potential. That's undeniable,
00:00:39
but I'm going to dig deep, and I mean
00:00:41
deep, into these two companies. We're
00:00:43
going to try to figure out which one
00:00:45
makes the most sense in the market
00:00:47
today. And yeah, I know that that means
00:00:49
that half of you might disagree with me
00:00:50
by the end, but stick with me. We're
00:00:52
going to look at different timelines,
00:00:54
different risks, and different stock
00:00:56
estimates. I want to pull out all the
00:00:57
key information. Forget the noise. What
00:01:00
actually matters for understanding where
00:01:02
these two giants are going because the
00:01:04
contrast, that's what makes this
00:01:05
interesting. It's like we're looking at
00:01:07
two fundamental building blocks of this
00:01:09
AI revolution, personified by two
00:01:12
companies. Nvidia, it's the engine, raw
00:01:15
pure power. They make the insanely
00:01:17
powerful hardware, those GPUs that
00:01:19
everyone, and I mean everyone, needs to
00:01:22
build and train these massive AI models.
00:01:24
It's the foundational muscle. And then
00:01:26
there's Palanteer. They're positioning
00:01:27
themselves as the operating system for
00:01:29
AI, the software layer that lets
00:01:31
companies actually use all that NVIDIA
00:01:34
power, connect it to their own messy
00:01:36
data, and do it securely and
00:01:38
effectively. It's the brain telling the
00:01:39
muscle what to do. Let's break it down.
00:01:42
Engine versus operating system. Nvidia.
00:01:45
For years, we knew them for this, making
00:01:47
games look realistic. But their big
00:01:49
pivot, the move that really put them in
00:01:51
the heart of AI, it came from a kind of
00:01:53
brilliant accident. They realized that
00:01:55
the way their chips handle graphics,
00:01:57
tons of calculations all at once,
00:01:58
parallel processing, was perfect for
00:02:01
training AI. But here's the real secret
00:02:03
sauce. The thing competitors are
00:02:05
struggling with. CUDA, or compute
00:02:07
unified device architecture. It's their
00:02:09
own software platform launched nearly 20
00:02:12
years ago. It lets developers unlock the
00:02:14
GPU's power for stuff beyond graphics.
00:02:17
Millions of developers know it. There's
00:02:18
a whole universe of tools built around
00:02:20
it. That's not just hardware leadership.
00:02:23
That's a software moat deep and wide.
00:02:26
Because there are a ton of developers
00:02:27
trained on CUDA and are less likely to
00:02:29
switch to a competing platform and learn
00:02:31
a whole new programming environment. But
00:02:33
the moat doesn't stop with CUDA. They
00:02:35
also have NIM or NVIDIA inference
00:02:37
microservices where NIMS are pre-built
00:02:40
optimized software containers that allow
00:02:42
enterprises to easily deploy AI models
00:02:44
from NVIDIA partners or even
00:02:46
open-sourced on NVIDIA hardware. NIMS
00:02:48
abstract away much of the complexity of
00:02:50
model optimization and deployment,
00:02:52
potentially accelerating enterprise AI
00:02:54
adoption. NIM agent blueprints provide
00:02:57
templates for building AI agents for
00:02:59
specific tasks like customer service or
00:03:01
data extraction. This layer aims to
00:03:03
capture value beyond the chip sale
00:03:05
itself. They're also creating Omniverse,
00:03:07
which is a platform focused on creating
00:03:09
and operating virtual worlds,
00:03:11
simulations, and digital twins,
00:03:13
targeting the industrial digitalization
00:03:15
market. It leverages NVIDIA's strengths
00:03:17
in graphics and simulation, and is
00:03:19
offered via subscription for enterprise
00:03:21
use. Then there's NVIDIA AI Enterprise,
00:03:23
which is positioned as an operating
00:03:25
system for enterprise AI. The software
00:03:27
suite bundles tools, frameworks, and
00:03:29
support to streamline the development
00:03:31
and deployment of production-grade AI
00:03:33
applications. My point is that they're
00:03:35
not slowing down at all. Granted, there
00:03:37
is that whole $55 billion loss from not
00:03:39
shipping their H2O chips to China, but
00:03:41
the momentum is still there. They're
00:03:43
building the whole ecosystem, their own
00:03:46
processors, super fast networking to
00:03:48
connect everything, layers of AI
00:03:50
software, and they want to sell
00:03:52
companies the entire high performance AI
00:03:54
factory from top to bottom. Nvidia's
00:03:57
recent financial performance has been
00:03:58
stellar to say the least. For fiscal 25,
00:04:01
they guided towards revenue around 130.5
00:04:04
billion, which is a staggering 114%
00:04:07
increase year-over-year. The engine
00:04:09
behind this is the data center segment,
00:04:11
which grew even faster at 142%. To me,
00:04:14
this signals a fundamental widespread
00:04:16
reliance on their hardware for building
00:04:18
AI infrastructure. Profitability is
00:04:20
equally incredible. GAP gross margins
00:04:22
hovered around 75%. This level of
00:04:25
margins essentially have peaked and may
00:04:27
not be sustainable which we're going to
00:04:28
see in Q1 of 26 with a gap gross margin
00:04:31
of around 70%. This potential moderation
00:04:34
reflects the natural economic pressures
00:04:36
that come from dominant players. You
00:04:38
have increasing competition, evolving
00:04:40
product mix, and the sheer difficulty of
00:04:42
sustaining these historically high
00:04:44
margin levels indefinitely. While we
00:04:46
might see slight margin moderation as
00:04:48
competition increases, the current
00:04:50
financial picture is one of dominant
00:04:52
scale and profitability in both the
00:04:54
short and the long term. So, where does
00:04:56
the growth keep coming from? Well, more
00:04:58
data centers for sure, and some big
00:05:00
companies are scaling back data center
00:05:02
expansion until the dust settles on the
00:05:05
tariff yo-yo situation that we've been
00:05:07
dealing with. But overall, the primary
00:05:09
growth driver continues to be the data
00:05:11
center market. Fueled by generative AI
00:05:13
training and inference, cloud provider
00:05:15
buildouts, and emerging sovereign AI
00:05:18
initiatives by nations worldwide. They
00:05:20
are deepening partnerships with all
00:05:21
major cloud providers to integrate their
00:05:23
latest chips like the Blackwell
00:05:25
platform. Looking ahead, the road map
00:05:27
includes the Vera Rubin architecture
00:05:29
slated for 26, promising another
00:05:31
significant performance jump,
00:05:33
potentially 3.3 times better than
00:05:35
Blackwell Ultra for inference. And then
00:05:38
there's the subsequent Feainman
00:05:39
architecture in 2028. Beyond GPUs,
00:05:42
Nvidia offers the Grace CPU superchip
00:05:44
for specific workloads and high
00:05:46
performance networking solutions. Think
00:05:48
Infiniban, Spectrumax Ethernet, and
00:05:50
VLink Interconnects. crucial for
00:05:52
building large-scale AI clusters,
00:05:54
reinforcing its full stack approach.
00:05:57
Their roadmap of new technology could
00:05:59
literally be a video in itself. Longer
00:06:01
term, Nvidia is strategically investing
00:06:03
in potentially huge markets of
00:06:05
automotive with their drive platform for
00:06:07
autonomous vehicles and robots with
00:06:09
their Jetson platform for edge AI.
00:06:11
Gaming remains a solid profitable
00:06:13
business, but it isn't the primary
00:06:14
hyperrowth driver anymore, seeing as how
00:06:17
gaming only made up 8% of last year's
00:06:19
revenue. Its future growth may lie in
00:06:21
cars, robots, and CUDA X tied to all
00:06:24
their hardware. Nvidia is placing wide
00:06:26
bets to keep its momentum. The engine is
00:06:28
firing on all cylinders. Powerful,
00:06:30
profitable, and it's expanding its
00:06:32
reach. Before diving in, I want to share
00:06:34
some info about today's sponsor, Liberty
00:06:36
Defense, where they are revolutionizing
00:06:38
security with AI powered next-gen
00:06:40
detection solutions for concealed
00:06:42
metallic and non- metallic threats.
00:06:44
Their flagship Hexwave technology offers
00:06:46
seamless and touchless screening without
00:06:48
the need to remove personal items,
00:06:50
making security faster and more
00:06:52
efficient at airports, stadiums, and
00:06:54
even schools. Liberty Defense has
00:06:56
secured an exclusive license for its
00:06:58
core technology from MIT, and has
00:07:00
invested over $50 million in R&D.
00:07:03
They've achieved significant milestones,
00:07:04
including deployments at major airports
00:07:06
like Toronto, Pearson, Manchester, and
00:07:08
Rochester, and have been awarded over 5
00:07:11
million in contracts from the TSA for
00:07:13
advanced imaging upgrades to the
00:07:15
existing body scanners. and their
00:07:17
president Brian Cunningham is also an
00:07:18
adviser and senior counsel to Palanteer
00:07:21
with a projected revenue of $20 million
00:07:23
in 2025 and 57 million in 2026 and
00:07:27
strong market drivers like the TSA
00:07:29
mandate for 100% aviation worker
00:07:31
screening. Liberty Defense is poised for
00:07:33
significant growth and they're committed
00:07:35
to creating safer environments and
00:07:37
providing peace of mind through superior
00:07:39
security technology. To learn more,
00:07:41
check out the link down in the
00:07:42
description.
00:07:43
[Music]
00:07:48
Now, Palunteer, a totally different
00:07:51
story. They started in the shadows,
00:07:53
building complex data tools for
00:07:54
governments, defense, intelligence
00:07:56
agencies, places with huge, messy,
00:07:59
secret data. Their Gotham platform
00:08:01
became legendary in those circles. For
00:08:03
years, Palanteer's financial narrative
00:08:05
was dominated by strong, albeit
00:08:07
sometimes lumpy, government revenue and
00:08:10
persistent gap losses due to high
00:08:11
operating costs and their stockbased
00:08:13
compensation. However, the past year has
00:08:16
marked a significant inflection point
00:08:18
characterized by its accelerating
00:08:20
overall growth, the achievement of
00:08:21
sustained gap profitability, and most
00:08:24
notably, its dramatic surge in its US
00:08:27
commercial business, which is what I had
00:08:29
been trying to convey in my early videos
00:08:31
on Palunteer 2 years ago. Their best
00:08:33
move was taking that government contract
00:08:36
know-how to the corporate world with
00:08:38
Foundry, helping big companies make
00:08:39
sense of their own data chaos. But the
00:08:42
gamecher, the reason we're talking about
00:08:44
them in the same breath as NVIDIA now is
00:08:47
AIP. AIP represents Palunteer's
00:08:49
strategic pivot to become the essential
00:08:51
software layer for enterprises looking
00:08:53
to deploy AI. It's designed not to build
00:08:56
the base AI models itself, but to
00:08:58
provide the secure infrastructure,
00:09:00
governance tools, and connectivity for
00:09:02
organizations to safely use various
00:09:05
large language models and other AI tools
00:09:07
on their own proprietary data integrated
00:09:10
directly into their operational
00:09:11
workflows. Boiling it down, this isn't
00:09:14
about selling the AI models like
00:09:16
ChatGpt. It's about creating the secure
00:09:18
environment for the companies to use
00:09:20
those models safely on their own within
00:09:23
their sensitive data plugged right into
00:09:25
how they actually work. Think of it as a
00:09:27
secure control room for enterprise AI.
00:09:29
And they found a clever way to sell it.
00:09:32
AIP boot camps, short intense workshops.
00:09:35
Potential customers come in, use AIP
00:09:38
with their data, and see results fast.
00:09:41
It literally slashed their sales cycle.
00:09:43
Suddenly, boom, traction. They now have
00:09:46
potential customers immediately
00:09:48
converting to a full-fledged customer.
00:09:50
But AIP is just one of four major
00:09:52
pillars for Palanteer. They began with
00:09:54
Gotham, as I'd mentioned earlier, which
00:09:56
was initially developed for the US
00:09:57
intelligence community and government
00:09:59
agencies. It excels at fusing diverse
00:10:02
data types, structured and unstructured,
00:10:04
from signals intelligence to human
00:10:05
reports, to identify hidden patterns,
00:10:08
connections, and threats. While
00:10:09
primarily a government tool, its
00:10:11
capabilities lay the groundwork for
00:10:13
Palanteer's broader ambitions. Foundry
00:10:15
is now used by all of Palanteer's
00:10:17
commercial customers and an increasing
00:10:19
number of government clients, often
00:10:21
serving as the backbone for AIP
00:10:23
deployments. And the last pillar is
00:10:25
Apollo, where it is now a commercial
00:10:27
offering that provides a single control
00:10:29
layer for continuous software delivery.
00:10:31
Think automated maintenance, security
00:10:33
updates, and configuration management.
00:10:35
Palanteer is hitting an inflection
00:10:37
point. While fiscal 2024 revenue at 2.87
00:10:41
billion, up 29% was solid. The key story
00:10:44
is the acceleration seen throughout the
00:10:46
year, especially in their high margin US
00:10:49
commercial business, which surged 64%
00:10:51
year-over-year in Q4. This clearly shows
00:10:54
that their strategy of expanding beyond
00:10:56
just government contracts and tapping
00:10:58
into the enterprise AI market via AIP is
00:11:01
gaining a lot of traction. Guidance for
00:11:03
fiscal 25 points to continued strength
00:11:06
projecting US commercial growth of at
00:11:07
least 54% achieving consistent gap
00:11:10
profitability suggesting a more
00:11:12
sustainable business model. Palanteer is
00:11:14
generating substantial cash. Q4 24 saw
00:11:18
460 million in cash from operations.
00:11:21
That's a 56% margin. Getting new
00:11:23
customers underscores the business
00:11:25
acceleration. Q4 saw a record 803
00:11:28
million in US commercial total contract
00:11:30
value closed, up 134% year-over-year.
00:11:34
This combination of strong growth and
00:11:35
high margins resulted in an impressive
00:11:38
rule of 40 score of 81% in Q4. On top of
00:11:41
all that, they also maintain a strong
00:11:43
balance sheet with over $4 billion in
00:11:46
cash and no traditional debt. The growth
00:11:48
here is a little less about sheer scale
00:11:50
like Nvidia, but it's more about rapid
00:11:53
acceleration in a key strategic market
00:11:55
segment. Now, I am going to drop a
00:11:57
speculation bomb on this video. Now, I
00:12:00
realize that I am reaching way out there
00:12:01
and I mean way out there. But I find it
00:12:04
intriguing that Palunteer recently
00:12:06
partnered with Archer Aviation and that
00:12:08
relationship is seeking to co-develop
00:12:11
nextgen software solutions for critical
00:12:13
aviation systems and EV tall route
00:12:15
planning and possible air traffic
00:12:17
control. And if you aren't aware, Archer
00:12:19
is an EV tall company looking to bring
00:12:21
an Uberlike experience to urban flights.
00:12:24
What if, and I realize it's a big if,
00:12:27
Palanteer uses this as a test bed for
00:12:30
something much bigger. What if this is
00:12:32
just an opportunity to upgrade the US
00:12:34
air traffic control systems with their
00:12:36
foundry capabilities to integrate
00:12:39
real-time data, think weather data,
00:12:41
flight plans, and airport operations
00:12:43
data, and use AIP to build an AI
00:12:46
assisted tool for controllers. Like I
00:12:48
said, I think this is just pure
00:12:50
speculation, but think about it for just
00:12:52
a moment. There is no other software
00:12:54
company with these capabilities that are
00:12:57
already this tight with the US
00:12:58
government contracts. And strangely
00:13:00
enough, I have a background in air
00:13:02
traffic control where I used to train
00:13:04
air crews in electronic warfare and
00:13:05
bombing runs. Now, let's move on to the
00:13:08
downside of Palanteer. Their sales model
00:13:10
and ability to scale are really holding
00:13:13
them back. I've mentioned this in a few
00:13:15
other videos that the boot camps are
00:13:17
probably having great conversion of
00:13:19
creating the sale, but it requires
00:13:21
companies to commit a resource to create
00:13:23
a use case for testing AIP and then they
00:13:26
have to go through a near week-long boot
00:13:28
camp just to test it out. It requires a
00:13:31
lot of commitment on the front end that
00:13:33
a lot of companies just aren't going to
00:13:34
take the time to do. And there's other
00:13:36
competitive companies that have a
00:13:38
traditional approach of just selling
00:13:40
capabilities without requiring a boot
00:13:42
camp. These competitors may have a lower
00:13:44
conversion rate, but they can scale
00:13:46
faster and they can get in front of a
00:13:48
lot more customers in a lot less time.
00:13:50
And in my opinion, this is going to hurt
00:13:52
Palanteer's ability to scale. Another
00:13:54
challenge for Palanteer is to justify
00:13:56
its stock valuation. Its forward price
00:13:59
to sales ratio is between 75 to 100x and
00:14:03
price to earnings have been well over
00:14:05
100x. And yes, the future potential is
00:14:08
there. But these high ratios, it demands
00:14:11
perfection from Palanteer, meaning they
00:14:14
can't have any slip ups in their revenue
00:14:16
or their margin guidance. It's a lot of
00:14:18
pressure for them to keep exceeding
00:14:20
analyst forecasts. And the last
00:14:22
challenge that I'll cover for Palanteer
00:14:24
is its reliance on government contracts.
00:14:26
Last year, the government segment
00:14:28
accounted for 55% of their revenue. And
00:14:30
as consistent and nice as government
00:14:33
contracts are, they are subject to
00:14:35
lumpiness, meaning that the contracts
00:14:37
are awarded infrequently and it doesn't
00:14:39
create a constant revenue stream. And
00:14:42
that's why I call it lumpy. So far, I've
00:14:44
covered both companies extensively to
00:14:46
provide a lot of context about what they
00:14:48
do best, where they have some exposed
00:14:50
weaknesses, and where they plan to go.
00:14:52
Let's break down a complete comparison.
00:14:54
When it comes to growth trajectory,
00:14:56
Nvidia is already at the top of their
00:14:58
game. At the 3-year, I see them as high
00:15:00
growth. The 5-year is more like medium
00:15:03
to high. And the 10-year, I only see it
00:15:05
as medium based on what we know today
00:15:07
and the potential slowdown in data
00:15:09
center growth. Now, on to Palunteer's
00:15:11
growth at 3 years, very high, continued
00:15:14
high at 5 years and then maybe slowing
00:15:17
to medium high at 10 years. Palunteer
00:15:19
will be accelerating for a very long
00:15:21
time and they are early in their growth
00:15:23
cycle. Next, let's look at the margin
00:15:25
profiles. Nvidia is very high today and
00:15:28
they're peaking in margin and have the
00:15:30
most risk of margin compression over
00:15:32
time. As for Palanteer, it is high and
00:15:35
improving. For Palanteer, they have a
00:15:37
lot of strong expansion and their main
00:15:39
pressure comes with the cost associated
00:15:41
with that expansion. So, I expect their
00:15:43
margins to continue improving over the
00:15:45
next 5 years. Next is a competitive moat
00:15:48
where Nvidia is very strong. They have
00:15:50
over 80% market share in their data
00:15:52
center hardware and their CUDA ecosystem
00:15:55
locks in their clients. Palanteer is
00:15:57
strong with its niche offering of
00:15:58
integrated platforms and ontology but
00:16:01
they face broad competition that may be
00:16:03
able to scale a little bit quicker than
00:16:05
them. The next section is key risks and
00:16:07
for Nvidia theirs happens to be
00:16:09
competition. Everyone is creating their
00:16:11
own GPUs these days including their
00:16:14
clients like Amazon, Microsoft and
00:16:16
Google. Plus they are peaking in their
00:16:18
margins. There's going to have to be
00:16:20
some compression and margins will come
00:16:22
down just to be competitive. And as for
00:16:24
Palanteer, they need perfect execution
00:16:27
because of their high valuation. They
00:16:28
can't afford to have any mistakes. And
00:16:30
then of course, their commercial
00:16:31
business model can't scale like their
00:16:34
competitors, and their government
00:16:35
dependency creates a little bit of
00:16:37
lumpiness for their revenue and their
00:16:38
margin. Now, on to my speculative
00:16:40
upside. Nvidia owns the physical AI
00:16:43
space with high upside across the board.
00:16:45
They also have a focus on automotive
00:16:47
robotics and their software ecosystem
00:16:49
where Palunteer has AIP ubiquity broad
00:16:52
OS enterprise adoption. And then there's
00:16:54
the speculation of future potential like
00:16:57
perhaps air traffic control. Seriously,
00:16:59
if it does happen, you better remember
00:17:00
it was me. And if not, well, you can
00:17:02
just call me crazy. So, how does all of
00:17:04
this boil down with regards to estimated
00:17:07
stock growth? My disclaimer is that I'm
00:17:09
using very broad brush strokes based on
00:17:11
the information that's available today,
00:17:13
but please feel free to give your point
00:17:15
of view down in the comments. Let's
00:17:16
start with the estimated stock growth at
00:17:18
3 years. I have Nvidia being more tight
00:17:21
with its potential of being positive 60
00:17:23
to 90%. And if the restrictions on
00:17:26
selling to China are lifted, that could
00:17:28
easily become so much higher. And for
00:17:31
Palanteer, well, it has a little bit
00:17:33
more variability in the short term. So,
00:17:35
I have them anywhere between 40 to 100%
00:17:38
in 3 years because of the high- risk,
00:17:40
highreward and the lumpiness in the
00:17:41
government contracts. And at the 5-year
00:17:44
mark, Nvidia is still a bit tighter
00:17:46
between 100 to 185%. And Palanteer is
00:17:50
still a little loose, but more upside
00:17:51
potential between 60 to
00:17:53
250%. Because it hangs solely on their
00:17:56
commercial business. And at the 10-year
00:17:58
mark, I have Nvidia at between 200 and
00:18:00
300% growth. And it could easily be a
00:18:03
lot more than that, but they have the
00:18:05
most to lose with competition since they
00:18:07
are so high on a pedestal. And as for
00:18:09
Palanteer, I see them with between 200
00:18:11
and 500% growth based on how they've
00:18:14
been performing and accelerating. But of
00:18:16
course, if they make strides in scaling
00:18:17
up their commercial business and land
00:18:19
some lofty government contracts on the
00:18:21
side, they can blow this estimate way
00:18:23
out of the water. In the short term, I
00:18:25
see Nvidia as a saver play with all the
00:18:27
hits that they've already taken on the
00:18:29
stock price. Now, in the medium-term,
00:18:31
they are both great options and nearly
00:18:34
tied, but I'd give a little bit more of
00:18:36
a nod to Nvidia as more of a sure thing.
00:18:39
And on the long term, I see Palanteer as
00:18:42
the best option between the two. But
00:18:44
don't ignore the fact that both are
00:18:47
great companies to invest in, and
00:18:49
they'll continue to do very well for the
00:18:51
next decade. That was my best attempt at
00:18:53
breaking down these two companies in a
00:18:54
deep dive, and I certainly hope that you
00:18:56
learned a little something new. As
00:18:58
always, thanks for watching.
