Every Major AI Stock, Explained From the Inside

Coin Crunch
8 followers
AI Generated·May contain errors
Video Summary
AI GeneratedThis video breaks down the AI industry by categorizing major stocks based on their specific role in the \"AI stack.\" Rather than grouping them all as \"AI stocks,\" it distinguishes between those providing the physical hardware, the manufacturing equipment, the cloud infrastructure, and the end-user software applications.
Almost none of them actually do the same thing. One designs chips, one manufactures them,
one just sells cloud space by the hour. Here is every major AI stock and exactly where it
actually sits in the machine. NVIDIA. NVIDIA does not build artificial intelligence,
it builds the chips that make artificial intelligence physically possible to train
and run. Its graphics processing units, originally designed for video games,
turned out to be extraordinarily good at the kind of parallel math AI models require.
NVIDIA controls an estimated 90% of the market for these specialized AI chips.
That dominance is not just hardware. NVIDIA built CUDA, the software layer developers use
to actually program its chips. Switching away from NVIDIA doesn't just mean buying different
hardware, it means rewriting software built around an ecosystem the company has spent nearly
two decades entrenching. This is why competitors keep gaining chip market share slowly,
while NVIDIA's software advantage remains almost impossible to dislodge quickly.
Microsoft. Microsoft's AI story runs through two separate channels at once.
Azure, its cloud computing platform, rents out the computing power companies need to train
and run AI models, competing directly with Amazon and Google for that business.
Separately, Microsoft made an enormous investment into OpenAI, the company behind ChatGPT,
embedding that technology directly into Office, Windows, and its own search products.
This gives Microsoft exposure on both sides. It profits when other companies buy cloud computing
to build AI, and it profits again when its own products get more valuable because of AI
baked directly into them. Few companies on this list are positioned in two different layers of
AI stack simultaneously, the way Microsoft is. That dual exposure is exactly why the company's
fortunes are watched so closely as a signal for the broader AI economy. If you want to actually
understand what you own instead of just recognizing the name, hit subscribe now. Costs nothing.
Alphabet. Google's parent company has arguably the deepest AI research history of any company
on this list, largely through its DeepMind division. Alphabet designs its own AI chips
internally, reducing its dependence on NVIDIA the way most competitors cannot. It also runs
Google Cloud, competing directly for the same enterprise AI computing business as Microsoft
and Amazon. The risk sits somewhere unusual for this list. Alphabet's core business, search advertising,
is the thing AI chatbots are most directly positioned to disrupt. Alphabet is simultaneously
one of the biggest beneficiaries of the AI boom and one of the companies most exposed to it changing
how people search for information at all. Few companies illustrate the double-edged nature of
this technology as clearly as the one that built much of the research behind it. Amazon. Amazon Web
Services is the largest cloud computing provider in the world, and AI compute has become one of its
fastest growing revenue lines. Beyond simply renting NVIDIA's chips to customers, Amazon has
also developed its own custom AI chips, Trainium and Inferentia, aiming to reduce costs and
dependence on outside suppliers. Amazon has also invested heavily in Anthropic, the AI company
behind Claude, deepening its exposure to the technology beyond just infrastructure. For Amazon,
AI is largely an infrastructure story rather than a product story. The company doesn't need to build
the most famous chatbot. It needs to be the place where everyone else's chatbot actually runs. That
distinction matters. Being the landlord of an industry is a different bet than being its most
visible tenant. Meta. Meta does not sell cloud computing or AI chips to anyone. It builds AI
purely for its own platforms, Facebook, Instagram, and WhatsApp. The company has committed to spending
well over $100 billion in a single year on AI infrastructure, an almost unprecedented scale
of internal investment. Meta released its Llama models as open source, a notably different strategy
from competitors who keep their most advanced models closed and proprietary. The bet is
straightforward. Better AI means better content recommendations, better advertising targeting,
and new products like AI-powered assistants embedded directly into apps billions of people
already use daily. Meta's AI spending is entirely a bet on itself, not a bet on selling AI infrastructure
to others. No other company on this list is spending this aggressively purely to improve
products it already owns, rather than to sell something new to somebody else. AMD. AMD is
Nvidia's most direct competitor in AI chips, though it remains a distant second in market share. The
company's strategy centers on pricing its chips more aggressively while offering an open alternative
to Nvidia's proprietary CUDA software ecosystem. Major cloud providers have begun adopting AMD's
chips, specifically to avoid being entirely dependent on a single supplier for their AI
infrastructure. AMD also supplies the processors inside many of the servers running AI workloads,
not just the specialized AI chips themselves. This makes AMD a bet on competition itself.
The company benefits anytime a customer wants meaningful leverage against Nvidia's pricing power.
Betting on AMD is less a bet that it wins outright, and more a bet that nobody wants a single company
controlling the entire chip supply. Broadcom. Broadcom doesn't sell AI chips to the general
market. It designs custom AI chips specifically for individual large customers, tailored to their
exact workloads. Companies like Google use Broadcom to help design chips optimized precisely for their
own AI systems, rather than using off-the-shelf hardware. Broadcom also supplies the networking
equipment that connects thousands of AI chips together inside a single data center, an often
overlooked but essential part of the process. This makes Broadcom's business model fundamentally
different from Nvidia's. It is not selling one product to everyone. It is selling customized
solutions to a handful of the largest technology companies on earth. TSMC. TSMC does not design any
chips itself. It manufactures the chips that Nvidia, AMD, and nearly every other major chip
company design, but cannot physically produce themselves. This single company manufactures an
overwhelming majority of the world's most advanced semiconductors, giving it a position that is
arguably more critical than any chip designer on this list. Every AI chip from Nvidia and AMD
is manufactured in TSMC's facilities, primarily concentrated in Taiwan. This creates a genuine
geopolitical risk layered directly into an investment thesis. The entire global AI industry
depends on manufacturing capacity concentrated in a single company, in a single region,
surrounded by significant political tension. ASML. ASML makes something almost nobody outside
the industry has heard of. Extreme ultraviolet lithography machines. The equipment required to
manufacture the most advanced chips in the world. Without ASML's machines, TSMC cannot manufacture
Nvidia's chips at all. There is currently no other company on earth capable of producing this
specific equipment. Each individual machine costs well over 100 million dollars and takes months to
assemble and calibrate. This makes ASML arguably the single most irreplaceable company in the entire
AI supply chain, sitting one full layer beneath even TSMC. Nobody buys an AI chatbot subscription
thinking about ASML. Nothing in this entire list would exist without it. Palantir. Palantir is the
clearest application layer bet on this list, building AI software rather than the infrastructure
underneath it. The company's platform helps governments and large corporations analyze
enormous amounts of data, historically centered on defense and intelligence work, before expanding
into commercial industries. Unlike the chip and cloud companies above, Palantir's revenue depends
entirely on convincing customers its specific software delivers results, not on selling raw
computing power. This makes Palantir considerably more speculative than the infrastructure names on
this list. Its valuation depends heavily on continued rapid growth expectations, rather than
the more predictable contract-based revenue models of cloud providers. Arista Networks. Arista makes
the networking switches that connect thousands of AI chips together inside massive data centers,
a layer of the AI stack that rarely gets discussed. Training a large AI model requires enormous
numbers of chips working in coordination simultaneously, and that coordination depends
entirely on extremely fast networking equipment moving data between them without bottlenecks.
Arista has become a preferred supplier for major cloud providers building out these AI data centers
at scale. This is a business built on being essential but invisible. Nobody thinks about the
networking switches connecting an AI data center together, right up until they fail and the entire
system slows down. Oracle. Oracle spent decades known primarily for enterprise database software,
a business that seemed far removed from the AI boom entirely. That changed as Oracle Cloud
Infrastructure secured massive contracts to provide the computing power behind major AI companies,
including large deals tied directly to OpenAI's infrastructure needs. This represented a genuine
repositioning, transforming a company most people associated with corporate software
into a legitimate AI infrastructure provider almost overnight. Oracle's AI story depends
heavily on a small number of enormous contracts, rather than broad diversified demand. That
concentration is exactly what makes Oracle both one of the more dramatic AI turnaround stories on
this list, and one of the more contract-dependent ones. Every AI stock. 12 completely different
businesses. One identical question hiding underneath every single one. What layer of the
stack does this company actually sit in? NVIDIA and AMD design the chips. TSMC and ASML make
manufacturing those chips physically possible. Microsoft, Amazon, and Alphabet rent out the
computing power. Meta spends purely on itself. Broadcom builds custom solutions for a handful
of giants. Palantir, Arista, and Oracle each occupy their own distinct corner. Software,
networking, and cloud contracts, respectively. Calling all 12 of these AI stocks is technically
true, and almost completely unhelpful. The next time you hear that phrase, you will not just nod
along. You will ask the only question that actually matters. What does this company sell, and to whom?
