►
Description
Welcome + Opening Remarks, Program Commitee Member - Steve Wong, VMware
A
You
saw
the
official
theme
when
you
registered
some
impressive
numbers
here:
Edge
compute
four
times
larger
than
Cloud
generating
75
percent
of
data
worldwide
in
marketing
forecasts
like
this
are
are
called
a
pitch
deck.
If
this
got
your
boss
to
okay,
you
coming
to
this
event,
that's
great,
but
let's
be
Skeptics
and
question
the
likely
growth
in
Edge
as
a
thought
exercise.
A
Why
will
Edge
be
generating
and
handling
75
percent
of
all
data?
Why
will
people
decide
to
increase
investments
in
Edge
compute?
Let
me
outline
a
few
technology
shifts
taking
place
and
lay
out
an
idea
of
what
this
could
lead
to.
Then
you
can
decide
for
yourself
whether
that
growth
forecast
is
real.
There
are
five
significant
changes
going
on
right
now.
A
The
world
is
using
approaches
like
get
Ops
and
device
twins
to
manage
Edge
at
scale,
and,
of
course,
the
elephant
in
the
room
is
AI.
These
changes
are
the
building
blocks
that
are
going
to
account
for
dramatic
changes
in
growth
at
Edge.
But
let
me
walk
you
through
an
example
that
puts
all
of
these
things
together.
A
A
A
The
savings
aren't
just
in
cloud
storage.
If
you
process
workloads
such
as
high
def
video,
the
downside
goes
beyond
latency
and
added
points
of
failure.
It
can
bring
wasteful
energy
burn
at
each
step
of
copying
and
caching.
What
if
you
could
summarize
that
edge?
What
if
you
could
use
something
to
analyze
and
act
on
data
at
Edge,
avoiding
round
trips
to
the
cloud
chat?
A
Gtp
gpta
recently
woke
up
the
world
as
to
the
potential
of
AI,
but
chat
is
not
the
only
application
for
AI
and
a
Central
Public
cloud
is
not
the
only
place
where
this
could
be
useful.
Data
is
being
generated
in
manufacturing,
retail
utilities
and
other
Edge
use
cases,
and
we
have
affordable,
AI
Technologies
to
act
on
this
data.
The
Google
USB,
TPU
and
I've
got
one
here
to
show
you
it's
real
and
the
mg24
soc
are
just
a
few
examples
of
technology
for
far
Edge
that
it
that
exists
to
run
inference.
A
Engines
at
low
cost
and
high
performance
technology
also
exists
for
mid-tier
locations
that
can
use
higher
power
accelerators
as
a
shared
resource.
Let's
look
at
a
specific
example
of
how
these
machine
learning
and
AI
accelerators
could
be
used
when
I
worked
in
Industrial
Automation
I'd
run
into
plants
where
they'd
have
a
veteran
who
could
come
in
in
the
morning
and
tell
just
by
ear
if
a
turbine
was
off.
This
slide
from
Silicon
Labs
is
based
on
training
and
ml
inference.
A
A
Maybe,
but
what's
the
cost
of
an
error
if
it
occurs,
a
false
positive
might
mean
that
you
change
the
bearing
a
little
early,
but
if
this
also
cap
catches
and
prevents
occasional,
unplanned
outages,
what
would
that
be
worth
bottom
line?
Is
this
doesn't
have
to
be
perfect?
It
has
to
be
work
most
of
the
time
and
be
better
than
the
status
quo,
and
it
becomes
valuable.
A
Don't
get
me
wrong
here.
This
is
new
stuff
and
it's
changing
rapidly,
but
the
opportunity
is
huge.
These
devices
aren't
multi-million
dollar
quantum
computers.
Maybe
you
want
to
consider
doing
a
proof
of
concept
to
learn
about
this
chat.
Gpt
was
trained
on
the
resource
of
accumulated
web
pages
from
the
internet,
but
it
seems
that
training.
This
training
was
maybe
burning
through
the
world's
accumulated
information
like
burning
up
fossil
fuel
as
a
resource.
A
You
see
here
a
quote
that
just
came
out
yesterday
from
the
CEO
of
open
AI,
the
company
behind
chat,
GPT,
well,
Edge
data,
if
you'd
view
mining
the
existing
internet
text
as
a
fossil
fuel
burn.
Edge
data
is
more
like
a
continuous
source
of
renewable
energy
in
the
form
of
data
screens
that
just
don't
ever
run
out
now
with
chat
gpt4,
you
could
upload
limited
amounts
of
additional
information
and
use
it
as
training
data,
but
not
all
Edge
data
is
text
and
doing
this
might
be
giving
away
Trade,
Secrets
or
customer
information
good.
A
So
AI
has
massive
uncertainties
and
risks,
but
and
for
those
in
competitive
businesses
that
get
caught
unprepared,
every
business
is
going
to
change,
whether
we
like
it
or
not,
and
and
on
top
of
that
AI
is
likely
to
alter
your
career,
maybe
sooner
than
you
think.
Ai
is
not
going
to
replace
Engineers,
but
Engineers
who
know
how
to
deploy
AI
are
going
to
replace
those
who
don't.
A
If
you
wanted
to
use
AI
at
Edge,
there's
more
to
it
than
just
buying
some
chips,
you're
going
to
need
tools
for
deployments,
updates
security,
connectivity
and
more
this
stuff
is
available,
and
we're
going
to
talk
about
it
today.
So
here
are
the
sessions
that
we've
got
laid
out
this
afternoon.
Please,
please,
take
a
look
at
this
list.
A
Ai
Tech
can
be
an
intellectual
Force
multiplier,
just
like
the
steam
engine
and
electricity,
where
physical
Force
multipliers
that
change
the
world
and
got
12
year
old,
kids
out
of
coal
mines.
Suppose
we
could
use
AI
to
improve
the
efficiency
of
a
manufacturing
or
Transportation
operation
by
just
two
percent.
These
kinds
of
advances,
save
resources
and
improve
quality
of
environment
while
elevating
the
standard
of
living
and
quality
of
life.
For
all.
This
is
a
critical
time
to
come
together
to
listen,
learn,
share
and
be
curious
about
what
comes
next,
let's
explore
together.
A
The
way
to
get
the
most
out
of
this
event
is
to
go
beyond
passive.
Listening
make
an
attempt
to
meet
the
people
here
at
the
conference
and
hold
discussions
out
in
the
hallway
during
breaks
or
after
hours,
introduce
yourself
to
the
people
around
you
who
don't
work
for
your
own
organization.
This
event
is
a
all
about
community.
A
Finally,
this
event
is
subsidized
by
the
event
sponsors.
Yes,
you
paid
a
registration,
but
I
these
sponsors
help
cover
the
budget,
so
I
want
to
quickly
recognize
them.
So
that's
the
end
of
my
talk.
Next
up,
we've
got
Danielle
coming
up
to
talk
about
kubernetes
for
low
resource
Edge,
so
give
us
a
few
minutes
to
Cable
up
a
different
laptop.