Something big is happening at Broadcom

that could mean big trouble for Nvidia.

Broadcom's AI chip revenues more than

doubled year-over-year and they're

quietly locking in massive deals with

Google, Meta, OpenAI, and Anthropic.

Could Broadcom be the next king of AI?

And what does all this mean for Nvidia

stock going forward? Your time is

valuable, so let's get right into it.

First things first, I'm not here to hold

you hostage, so here's everything I'll

cover in this video. What Broadcom

actually does and how they make their

money, how their AI hardware and

software stacks up against Nvidia, what

their latest earnings call tells us

about the pace of the AI revolution, and

of course, what all this means for

Broadcom stock as a result. Broadcom is

split into two big engines. The first is

chips and the second is software. On the

chip side, Broadcom is a fabulous

semiconductor company, which means they

do the design and then send their

designs to manufacturers like TSMC to

actually build the chips. Broadcom's

chips end up in AI data centers,

networking equipment, storage systems,

smartphones, and a whole lot more. The

most important piece for investors is

their AI and data center business.

Broadcom designs custom AI accelerators

for some of the biggest players in the

game, including Google's TPU program,

Meta's in-house AI chips, and newer

deals with AI labs like OpenAI and

Anthropic. Unlike Nvidia's GPUs,

Broadcom's chips are made to order and

tuned to every customer's models and

infrastructure individually. On top of

that, Broadcom sells the high-speed

nervous system that connects AI chips in

data centers together. Specifically,

their Tomahawk and Jericho chips power

ultra-fast Ethernet switches and routers

that move data between thousands of AI

accelerators. If you've heard of million

GPU or a gigawatt-scale AI clusters, a

lot of that traffic is riding on

Broadcom's networking chips. Their

second engine is software, mostly

because of VMware. VMware's platform

lets companies run many separate virtual

machines on the same physical servers

and manage big workloads in their own

data centers and in hybrid cloud setups

in a single consistent way. Since its

acquisition, VMware shows up as

Broadcom's infrastructure software

segment, generating a lot of recurring

revenue at very high margins. This

combination of hardware and software is

why Broadcom isn't just another chip

stock. By providing the picks and

shovels for a data center's physical

layer and its control layer, Broadcom

gets paid multiple ways every time AI

infrastructure spending goes up.

Speaking of which, let's talk about

where Broadcom actually sits in the AI

stack and how they compare to Nvidia. At

a high level, Nvidia sells general

purpose GPUs and full-scale server

systems that anyone can buy. On the

other hand, Broadcom builds custom AI

accelerators and networking for a small

group of very large customers. Nvidia is

the default choice for off-the-shelf AI

compute. Broadcom is the company you

call when you want your own chip, your

own networking solutions, and more

control over cost and performance. Last

quarter, Broadcom's AI revenues hit $8.4

billion,

which is up 106%

year-over-year. About 1/3 of that came

from AI networking, while the other 2/3

came from custom compute. Nvidia's GPUs

are the gold standard for AI training,

but Broadcom is quickly scaling a

parallel set of chips that can be tuned

for individual customers and locked into

their long-term road maps. For example,

Google uses Broadcom as the long-term

co-designer for their TPUs, which train

and run models like Gemini 3. And the

latest Meta training and inference chips

were developed with Broadcom as Meta

tries to lower their own dependence on

Nvidia's GPUs. Anthropic took this even

further, signaling a $21 billion

multi-year deal for nearly a million

TPUs and full rack-scale AI systems

built by Broadcom, effectively turning

them into one of Anthropic's main custom

compute partners. On top of that, OpenAI

and Broadcom announced plans to deploy

10 gigawatts of custom accelerators,

confirming OpenAI as another flagship

customer in Broadcom's AI portfolio. So,

Broadcom isn't just supplying parts

around Nvidia systems. They're directly

displacing GPU demand by giving

hyperscalers their own custom silicon

for these AI jobs. Every chip deployed

by Broadcom means less workloads running

on Nvidia's GPUs. Today, Nvidia owns

roughly 90% of the data center GPU

market, making them the go-to solution

for general-purpose AI computing. But,

Broadcom owns roughly 70% of the custom

AI accelerator market and around 80% of

the market for data center Ethernet

switch chips. In fact, networking is

where the contrast is even sharper.

Nvidia has its own Infiniband and

Spectrum-X Ethernet stack, but Broadcom

is the dominant supplier of high-end

Ethernet switch chips through their

Tomahawk and Jericho lines. Tomahawk is

the high-bandwidth switch chip

connecting GPUs and ASICs inside and

across data center racks, while Jericho

is the router that stitches together

huge AI clusters and even entire data

centers into one massive compute system.

Vendors like Arista Networks, Juniper,

and even some Cisco switches are built

around these chips. So, Broadcom makes

money even when data centers don't buy

from them directly. So, while Nvidia

owns most of the AI infrastructure

market today, big companies are turning

to Broadcom when they want more

performance per watt for specific

applications and more control over their

AI stack in general. That's why

Broadcom's AI business is growing by

106% year-over-year. And now that you

know what Broadcom does and where they

sit in the AI stack compared to Nvidia,

let's dive into their earnings. Speaking

of earnings, a new study shows that US

workers who use AI every day earn 40%

more than those who don't. That means AI

is not optional. It's an advantage that

you either have or others have over you.

That's where Outskill comes in, the

sponsor of this video. Outskill is

running a live 2-day AI mastermind this

weekend. 16 hours of training that

brings together the knowledge of over a

hundred experts from companies like

Nvidia and Microsoft to make you

confident in using AI on your own,

staying ahead as tools rapidly evolve,

and turning those skills into

high-value, better-paid work. And

they're giving the first 1,000 people

who sign up with my link a free seat.

Whether you work in tech or sales,

management or marketing, you'll learn

how to use AI agents, create automated

workflows, and connect them to the

software and spreadsheets you already

use every day. This is a great way to

level up your AI knowledge, gain a real

competitive advantage, and understand

the science behind the stocks. Over 10

million people all over the world have

already attended, and the slots for this

one are filling up faster than ever. So,

make sure to register for your free seat

with my link below today. All right,

let's dive into Broadcom's earnings

because this latest quarter is one of

the clearest data points on where their

AI business is really headed. Broadcom

reported $19.3 billion in revenue, up

about 29% year over year, and just over

Wall Street's estimate of $19.2 billion.

And their GAAP earnings per share came

in at $1.50,

up from $1.14 a year ago. On paper, this

looks like a normal beat, but under the

surface, they're seeing massive growth

thanks to AI. Out of that $19.3 billion,

$8.4 came from AI. Like I said earlier,

that's up 106% year over year. But it's

also worth noting that AI now accounts

for roughly 44% of Broadcom's total

revenue. For context, AI is now a bigger

share of Broadcom's business than AMD's,

even though most investors still think

of AMD as Nvidia's main competitor in

AI. Broadcom's AI business is growing

faster, as well. If we break things down

by segment, semiconductor solutions did

about $12.5 billion in revenue, up 52%

year over year, and it now represents

around 65% of Broadcom's total business.

Infrastructure software did the other

6.8 billion, mostly thanks to VMware.

That's only up 1% year over year, but

gross margins for the software side came

in at 93%

with operating margins of 78%. So, the

growth is coming from AI chips, while

software provides insanely high margins,

which caused Broadcom's overall margins

to also come in very strong. They

reported GAAP gross margins of 68% or

adjusted gross margins of 77%,

15 full percentage points higher than

AMD, and almost in the same ballpark as

Nvidia. Broadcom's operating margins

came in at 66.4%,

slightly higher than a year ago. That

means they're actually expanding their

profitability while also ramping up

their AI products. And Broadcom's free

cash flows came in at $8 for the

quarter, or about 41% of their total

revenues. For context, Nvidia reported

51% free cash flow margins last quarter,

and AMD reported about 23%. But,

Broadcom's guidance is where their AI

story really starts to stand out.

Broadcom is guiding for about $22

billion in revenue next quarter, which

would be 47% year over year growth. And

they specifically called out AI

semiconductor revenue to reach about

$10.7 billion,

which would imply roughly 140%

growth year over year. Said another way,

Broadcom's AI revenues are expected to

increase by 27%

quarter over quarter and make up roughly

half of their total revenues if they can

actually hit these numbers. For

investors, Broadcom's earnings just

confirmed three big things. First, AI is

now the primary growth driver for their

business. Second, they're scaling that

AI revenue without sacrificing margins

that most hardware companies would kill

for. And third, their AI business isn't

just growing, it's accelerating, which

has big implications for the rest of the

AI revolution. For example, Broadcom CEO

Hock Tan told analysts that he has line

of sight to more than a hundred billion

dollars of AI chip revenue in 2027. And

he was very explicit that this is chips

only, not software, not services, just

semiconductors. That implies that

Broadcom's AI chip revenue will more

than double again by 2027, even from

today's much bigger base of 40 billion

dollars a year. He also said that

Broadcom already secured their supply

chain, the wafers, the advanced

packaging, and the high-bandwidth memory

to support that target. On top of that,

Broadcom's total backlog is over a

hundred and sixty billion dollars,

including a 73 billion dollar backlog

tied directly to huge orders from

hyperscalers and AI labs for custom

accelerators and AI networking products.

So, a large part of their projected AI

revenue growth is already under

contract. Hock Tan also mentioned that

demand from Google, Meta, Anthropic, and

OpenAI, and other large customers is

actually accelerating, and that he

expects AI to be the main driver for the

growth of their semiconductor business

for many years to come. So, if you're an

investor trying to figure out whether

we're close to the top of this AI

spending cycle or we're still in the

early innings of the AI revolution,

Broadcom's backlog is saying we're still

very early. It also reduces near-term

execution risk because a lot of AI

demand is already under a contract and

their supply chain is locked in. But

there are other kinds of risks that

investors need to know about. And if

you're finding this video valuable,

consider hitting the like button and

subscribing to the channel. That really

helps me out and it lets me know to make

more content like this. Thanks. Now

let's talk about the things that could

actually break Broadcom's story if they

go wrong. The first big risk is customer

concentration. Broadcom's AI revenue is

heavily tied to a very small group of

hyperscalers and AI labs like Google,

Meta, OpenAI, and Anthropic. Analysts

estimate that Broadcom already controls

over 60% of the custom AI chip market

and their three largest customers could

drive over $60 billion in annual AI chip

revenue by 2027 if the current road map

holds. That sounds great, but even if

one of their big customers slows down

their spending, delays a deployment, or

shifts more workloads back to Nvidia

once they ship Vera Rubin at scale,

Broadcom's growth could start to fall

and their stock price would fall with

it. The second risk is margin pressure

from the hyperscalers themselves. Custom

chips have lower margins because they're

designed for a single customer over many

years with very high upfront research

and development costs that usually get

amortized over one customer instead of a

much broader install base. At the same

time, hyperscalers have huge bargaining

power which puts a cap on the kinds of

margins that Broadcom can make before

customers simply switch to Nvidia.

Broadcom and AMD effectively sell cost

savings versus Nvidia, not a unique

platform of their own like CUDA. And

Broadcom's full rack-scale systems could

lower margins even further. When

Broadcom sells a complete AI rack, they

bundle their chips and networking

solutions with a lot of third-party

components like memory, other kinds of

processors, and sometimes even GPUs, all

of which Broadcom passes through to

their customers almost at cost. That

means there's a big chunk of low-margin

system revenue that sits on top of their

higher-margin chip revenues. Nvidia

doesn't have these problems because

their GPUs are standard products that

get sold to many customers, which means

their R&D costs get spread over a much

wider base. They also have a lot of

pricing power thanks to the CUDA

ecosystem, which customers can't really

get anywhere else. And when Nvidia sells

a full system, most of the value is in

their own hardware and software, with

far fewer third-party components getting

passed through to customers at low

margins. At the end of the day,

competing with Nvidia is going to be any

AI company's biggest risk. But

Broadcom's management sounded pretty

confident on the latest earnings call,

and they guided to 77% adjusted gross

margins. But the more their product mix

moves towards custom AI, the more I'm

watching for a drop in their gross

margins over time. All right, let's put

everything together and see if Broadcom

stock deserves a spot next to Nvidia in

long-term AI portfolios. Like I've been

saying for years, Broadcom is the only

real competitor to Nvidia because

they're going after different parts of

the stack, custom AI processors for a

few key tech giants, and the Ethernet

switch chips connecting thousands of

accelerators together. Nvidia owns

roughly 90% of the data center GPU

market, but Broadcom controls around 70%

of the custom AI accelerator market and

close to 80% of all Ethernet switch

chips. That's what real competition in

data centers looks like. But Broadcom

doesn't have to be Nvidia to win big.

They just have to win the companies that

don't want to fully depend on them. If

hyperscalers keep spending hundreds of

billions of dollars a year on AI

infrastructure, Broadcom's $160 billion

backlog will keep growing right

alongside that spend. AI already

accounts for 44% of Broadcom's total

revenues today, and they expect it to

more than double again in 2027. So, is

Broadcom stock better than Nvidia? I

don't think that's the right question.

Broadcom is more diversified and gives

my portfolio exposure to a different

side of AI spending altogether. ASICs

versus GPUs, Ethernet versus Infiniband,

and custom infrastructure versus

off-the- shelf platforms. That way I win

no matter which way the AI market goes.

But if you believe we're still in the

early innings of this multi-year AI

build-out, then Broadcom is a great

stock to hold alongside Nvidia, not

instead of it, making it a great way to

get rich without getting lucky. And if

you want to see what else I'm buying to

get rich without getting lucky, check

out this video next. Either way, thanks

for watching, and until next time, this

is ticker symbol U. My name is Alex,

reminding you that the best investment

you can make is in you.