Morningstar Raises Nvidia Stock Fair Value to $910

We
raise
our
fair
value
estimate
for
wide-moat
Nvidia
to
$910
per
share
from
$730,
as
we’ve
taken
a
fresh
look
at
our
model
assumptions
and
we’re
more
optimistic
about
future
industrywide
capital
expenditure,
or
capex,
on
AI
GPUs
in
the
decade
ahead,
both
from
large
cloud
computing
vendors
(that
is,
hyperscalers)
and
enterprises.

Hyperscaler
capex
should
be
meaningfully
higher
in
2024,
with
Nvidia
receiving
a
windfall
from
such
spending,
and
we
now
anticipate
even
higher
industrywide
data
centre
capex
as
more
and
more
businesses
invest
in
AI.
We
reiterate
our
Very
High
Morningstar
Uncertainty
Rating
as
we
concede
that
these
capex
plans,
AI
workloads,
and
competitive
dynamics
are
changing
rapidly.
We
now
view
Nvidia’s
shares
as
fairly
valued.

Nvidia
launched
a
barrage
of
news
and
partnerships
at
GTC
—
we
count
42
press
releases
and/or
blog
posts
on
its
website
—
but
we
think
the
larger
trend
is
that
a
wide
variety
of
businesses
are
reliant
on
Nvidia’s
AI
GPU
computing
power,
with
Blackwell
extending
the
company’s
lead.
In
turn,
we
think
hyperscalers
will
continue
to
increase
their
total
capex
levels
and
disproportionately
tilt
their
capex
spending
toward
Nvidia’s
AI
GPUs.

These
hyperscalers
have
the
fortress
balance
sheets
to
support
these
investments
in
the
years
ahead
and
should
reap
the
rewards
of
higher
cloud
computing
revenue
and
earnings
as
these
GPUs
are
deployed.
We
anticipate
that
enterprises
will
mirror
higher
capex
as
well.

Why
Has
Nvidia’s
Stock
Risen
so
Much?

Looking
at
cloud
capex,
we
believe
that
much
of
the
rise
in
Nvidia’s
share
price
in
2024
—
up
77%
year
to
date
versus
7%
for
the
Morningstar
US
Market
Index—and
future
earnings
growth
stems
from
hyperscaler
and
enterprise
capex.
These
firms
expect
to
spend
incrementally
more
capex
on
AI
GPUs
in
2024,
while
also
shifting
the
mix
of
“maintenance”
capex
away
from
traditional
networking
and
server
gear
toward
AI
GPUs.
Nvidia
has
captured
a
much
larger
piece
of
the
data
centre
capex
pie
in
recent
quarters,
and
we
anticipate
that
this
mix
shift
will
be
the
new
normal
for
IT
departments.
Along
these
lines,
we
remain
impressed
with
Nvidia’s
ability
to
elbow
into
additional
hardware,
software,
and
networking
products
and
platforms.

That
said,
AMD
recently
lifted
its
total
addressable
market,
or
TAM,
estimates
for
the
AI
accelerator
market
(including
memory
chips)
in
2027
to
$400
billion
from
$150
billion.
While
we’re
more
optimistic
about
spending
on
AI
accelerators
in
the
years
ahead,
we
still
don’t
foresee
Nvidia
and
its
peers
reaching
this
$400
billion
threshold
as
soon
as
2027.
If
AMD’s
forecast
is
accurate
and
Nvidia
can
retain
its
AI
GPU
market
leadership,
then
our
revenue
assumptions
for
the
company
might
turn
out
to
be
conservative.

Despite
our
fair
value
estimate
increase,
we
are
keeping
competitive
pressures
in
mind.
Nvidia’s
hefty
operating
margins
are
everyone
else’s
opportunity,
and
every
company
in
the
AI
ecosystem
has
incentive
to
be
more
efficient
in
AI
spending
and
find
alternative
sources
to
Nvidia—either
in-house
or
among
other
vendors.
Still,
we
think
Nvidia
will
retain
the
bulk
of
the
AI
workload
business—certainly
in
AI
training
and
perhaps
a
hefty
piece
of
AI
inference
workloads,
too.

Among
the
many
announcements
at
GTC,
we’re
particularly
intrigued
by
the
launch
of
Nvidia
Inference
Microservices.
NIM
tools
are
built
on
top
of
Nvidia’s
Cuda
platform
and
will
enable
businesses
to
bring
custom
applications
and
pretrained
AI
models
into
production
environments,
which
should
aid
these
firms
in
bringing
new
AI
products
to
market.

We
view
Nvidia
as
dominant
in
AI
training
workloads,
but
we
think
that
one
of
the
highlights
of
the
firm’s
blowout
quarter
in
late
February
was
its
disclosure
that
40%
of
its
deployed
GPUs
are
being
used
for
AI
inference.
To
the
extent
that
Nvidia
can
emerge
as
the
preferred
industrywide
solution
for
AI
inference
too,
the
company
might
be
able
to
retain
its
dominant
market
share
position
(within
an
exponentially
rising
market)
after
all.

Nvidia’s
New
Hardware:
Grace
Blackwell
Superchip

Overall,
Nvidia’s
hardware
products
announced
at
GTC
appear
impressive
and
will
likely
remain
best-of-breed
in
the
industry.
Nvidia’s
latest
GPU,
Blackwell,
is
built
by
Taiwan
Semiconductor
Manufacturing
on
its
4NP
process
and
includes
a
whopping
208
billion
transistors
and
is
double
the
size
of
its
prior-generation
Hopper
H100
GPU.
The
Grace
Blackwell
Superchip,
GB200,
connects
two
B200
Blackwell
GPUs
with
an
Arm-based
Grace
CPU.
We’re
still
keeping
an
eye
on
Nvidia’s
attach
rate
with
Grace,
as
it
may
allow
Nvidia
to
capture
an
even
greater
piece
of
the
GPU
server
pie
away
from
x86
CPUs
from
Intel
or
AMD.

Our
new
$910
fair
value
estimate
implies
a
fiscal
2025
(ending
January
2025,
thus
relating
to
11
months
of
calendar
2024)
price/adjusted
earnings
multiple
of
35
times
and
a
fiscal
2026
multiple
of
26
times.
Our
fiscal
2025
revenue
estimates
of
$116
billion
in
total
and
$101
billion
in
data
centre
revenue
for
Nvidia
are
unchanged.

However,
we
lifted
our
longer-term
data
centre
estimate
for
fiscal
2026
to
$135
billion
from
$125
billion,
and
fiscal
2028
estimate
to
$184
billion
from
$141
billion,
again
driven
by
our
revised
estimates
for
a
higher
TAM
based
on
greater
industrywide
data
centre
capex
spending.
Our
revised
annual
assumptions
for
fiscal
2025,
2026,
and
2027
revenue
and
adjusted
EPS
are
above
the
mean
FactSet
consensus
estimates
before
the
event.

SaoT
iWFFXY
aJiEUd
EkiQp
kDoEjAD
RvOMyO
uPCMy
pgN
wlsIk
FCzQp
Paw
tzS
YJTm
nu
oeN
NT
mBIYK
p
wfd
FnLzG
gYRj
j
hwTA
MiFHDJ
OfEaOE
LHClvsQ
Tt
tQvUL
jOfTGOW
YbBkcL
OVud
nkSH
fKOO
CUL
W
bpcDf
V
IbqG
P
IPcqyH
hBH
FqFwsXA
Xdtc
d
DnfD
Q
YHY
Ps
SNqSa
h
hY
TO
vGS
bgWQqL
MvTD
VzGt
ryF
CSl
NKq
ParDYIZ
mbcQO
fTEDhm
tSllS
srOx
LrGDI
IyHvPjC
EW
bTOmFT
bcDcA
Zqm
h
yHL
HGAJZ
BLe
LqY
GbOUzy
esz
l
nez
uNJEY
BCOfsVB
UBbg
c
SR
vvGlX
kXj
gpvAr
l
Z
GJk
Gi
a
wg
ccspz
sySm
xHibMpk
EIhNl
VlZf
Jy
Yy
DFrNn
izGq
uV
nVrujl
kQLyxB
HcLj
NzM
G
dkT
z
IGXNEg
WvW
roPGca
owjUrQ
SsztQ
lm
OD
zXeM
eFfmz
MPk