Nvidia's latest advancements, driven by the Blackwell architecture and new AI models, are revolutionizing computing across various domains, from personal PCs to industrial robotics and autonomous vehicles, by enabling unprecedented AI capabilities and scaling.
Mind Map
Expand करने के लिए click करें
पूरा interactive mind map देखने के लिए click करें
made a new kind of
factory generator of
tokens the building blocks of
AI tokens have opened a new frontier the
first step into an extraordinary world
where endless possibilities are born [Music]
[Music]
tokens transform words into knowledge
environment tokens teach robots to move
like the Masters [Music]
Inspire new ways to celebrate our
victories a martini pleas call light
up thank you
Adam and give us peace of mind when we
need it most hi moroka hi Anna it's good
to see you again hi Emma we're going to
take your blood sample today okay don't
time they bring meaning to numbers
to help us better understand the world around
around [Music]
us predict the dangers that surround [Music]
us and find cures for the threats within us
us [Music]
[Music]
[Music]
life and restore what we've [Music]
[Music] [Applause]
lost
buddy they help us move
forward one small step at a time [Music]
and one giant leap
leap [Music]
welcome to the stage Nvidia founder and
CEO Jensen [Music]
[Music] [Applause]
[Applause] [Music]
[Music] [Applause]
[Applause]
CES are you excited to be in Las
it I thought I'd go the other way from Gary
Gary
Shapiro I'm in Las Vegas after all if
does if this doesn't work out if all of you
you
object well just get used to it I think
I really think you have to let this sink
in in another hour or so you're going to
it well uh welcome to
Nvidia in fact you're inside nvidia's digital
digital
twin and we're going to take you to
Nvidia your
AI it has been an extraordinary Journey
extraordinary year here and uh it
started in 1993 ready go with
mv1 we wanted to build computers that
can do things that normal computers
couldn't and mv1 made it possible to
have a game console in your
PC our programming architecture was called
called
UD missing the letter c until a little
while later but UDA UniFi Unified device
architecture and the first developer for
UDA and the first application that ever
worked on UDA was sega's Virtual
Fighter six years later we invented in
1999 the programmable
GPU and it
started 20 years 20 plus years of
incredible advance in this incredible
processor called the GPU it made modern
computer Graphics
possible and now 30 years later sega's
cinematic this is the new Virtual
Fighter project that's coming I just
can't wait absolutely
incredible six years after that six year
six years after
1999 we invented Cuda so that we could
explain or or expressed the
programmability of our gpus to a rich
set of algorithms that could benefit
from it Cuda
initially was difficult to explain and
it took years in fact it took
approximately six years somehow six
years later six years later or so
2012 Alex kvki ilas sus and Jeff Hinton
discovered Cuda used it to process
alexnet and the rest of it is history AI
has been advancing at an incredible Pace
since started with perception AI we now
can understand images and words and
sounds to generative AI we can generate
images and text and
sounds and now agentic ai AIS that can
perceive reason plan and act and then
the next phase some of which we'll talk
about tonight physical AI 2012 now magically
magically
2018 something happened that was pretty
incredible Google's Transformer was
released as Bert and the world of AI
really took off Transformers as you know
completely changed the land landcape for
artificial intelligence in fact it
completely changed the landscape for computing
computing
altogether we recognized properly that
AI was not just a new application with a
new business opportunity but AI more
importantly machine learning enabled by
Transformers was going to fundamentally
change how Computing works and
today Computing is revolutionized in
every single layer from hand coding
instructions that run on CPUs to create
software tools that humans use we now
have machine learning that creates and
optimizes new networks that processes on
gpus and creates artificial
intelligence every single layer of the
technology stack has been completely
changed an incredible transformation in
just 12 years
well we can Now understand information
of just about any modality surely you've
seen text and images and sounds and
things like that but not only can we
understand those we can understand amino
acids we can understand physics we
understand them we can translate them
and generate them the applications are
just completely endless in fact almost
any AI application that you you see out
there what modality is the input that it
learned from what modality of
information did it translate to and what
modality of information is it generating
if you ask these three fundamental
questions just about every single
application could be inferred and so
when you see application after
applications that are Aid driven AI
native at the core of it this
fundamental concept is there machine
learning has changed how every
application is going to be built how
computing will be done and the
possibilities Beyond well
well
gpus gForce in a lot of
ways all of this with AI is the house
that GeForce built GeForce enabled AI to
reach the masses and now ai is coming
home to
GeForce there are so many things that
you can't do without AI let me show you
some of it now
[Music] [Applause]
[Applause] [Music]
[Music] [Applause]
[Applause] [Music]
Graphics no computer Graphics researcher
no computer scientist would have told
you that it is possible for us to rate
trce every single Pixel at this point we
Ray tracing is a simulation of light the
amount of geometry that you saw was
absolutely insane it would have been
impossible without artificial
intelligence there are two fundamental
things that we did we used of course
programmable shading and Ray traced
acceleration to produce incredibly
beautiful pixels but then we have artificial
artificial
intelligence be
conditioned be controlled by that pixel
to generate a whole bunch of other
pixels not only is it able to generate
pixels spatially because it's aware of
what the colors should be it has been
trained on a supercomputer back in
Nvidia and so the neuron Network that's
running on the GPU can infer and predict
the pixels that we did not render not
only can can we do that it's called
dlss the latest generation of dlss also
generates Beyond frames it can predict
the future generating three additional
frames for every frame that we calculate
what you saw if we just said four frames
of what you saw because we're going to
render one frame and generate three if I
said four frames at full HD 4K that's 33
million pixels or so out of that 33 million
million
pixels we computed only
two it is an absolute miracle that we
can computationally comput tionally
using programmable shaders and our R
traced engine R tracing engine to
compute 2 million pixels and have ai
predict all of the other 33 and as a
result we're able to render at
incredibly high performance because AI
does a lot less computation it takes of
course an enormous amount of training to
produce that but once you train it the
generation is extremely efficient so
this is one of the incredible cap
abilities of artificial intelligence and
that's why there's so many amazing
things that are happening we used gForce
to enable artificial intelligence and
now artificial intelligence is revolutionizing
revolutionizing
GeForce everyone today we're announcing
our next
Generation the RTX Blackwell family
here it
is our brand new gForce
gForce
RTX 50 Series Blackwell architect
the GPU is just a beast 92 billion transistors
transistors
4,000 tops four pedop flops of AI three
times higher than the last generation
Ada and we need all of it to generate
those pixels that I showed you 380 Ray
tracing Tera flops so that we could for
the pixels that we have to compute
compute the most beautiful image you
possibly can and of course 125 Shader
teraflops there is actually a concurrent
Shader teraflops as well as an Inger
unit of equal performance so two dual
shaders one is for floating point one is
for integer G7 memory from Micron 1.8
terabytes Per Second Twice the
performance of our last generation and
we now have the ability to intermix AI
workloads with computer graphics
workloads and one of the amazing things
about this gener eration is the
programmable Shader is also able to now
process neuron networks so the Shader is
able to carry these neuron networks and
as a result we invented neurot texture
compression and neurom material shading
as a result of that you get these
amazingly beautiful images that are only
possible because we use AIS to learn the
texture learn a compression algorithm
and as a result get extraordinary
results okay so this is this is uh the brand
brand new
new
9
now even even the even the mechanical
design is a miracle look at this it's
got two
fans this whole graphics card is just
one giant fan you know so the question
is where's the graphics card is it
literally this
big the voltage regul to design is
state-of-the-art incredible design the
engineering team did a great job so here
you okay so those are the speeds and
fees so how does it compare
compare
well this is RTX
490 I know I know many of you have
one I I know it look it's
$1,599 it is one of the best investments
you could possibly
make you for
$10,000 PC
entertainment Command Center isn't that
right don't tell me that's not true
don't be
ashamed it's liquid
cooled fancy lights all over it
leave it's it's the modern home theater
it makes perfect sense and now for
$1,500 and99
$15.99 you get to upgrade that and
turbocharged the living Daya lights out
of it well now with the Blackwell family
RTX 570 490 performance at 549 [Applause]
[Applause]
impossible without artificial
intelligence impossible without the Four
Tops four ter Ops of AI tensor cores
impossible without the G7 memories okay
so 5070 490 performance $549 and here's
the whole family starting from 5070 all
the way up to 5090 5090 twice the
starting of course we're producing at
very large scale availability starting
January well it is incredible but we
managed to put these in in gigantic
performance gpus into a laptop this is a
570 laptop for
$12.99 this 570 laptop has a 4090
performance I think there's one here somewhere
somewhere
let me show you
this this is a look at this thing here
let me
here there's only so many
pockets ladies and gentlemen Janine [Applause]
[Applause]
Paul so can you imagine you get this
incredible graphics card here Blackwell
we're going to shrink it and put it in
put it in there does that make any
sense well you can't do that without
artificial intelligence and the reason
for that is because we're generating
most of the pixels using pixels using
our tensor cores so we retrace only the
pixels we need and we generate using
artificial intelligence all the other
pixels we have as a result the amount of
the Energy Efficiency is just off the
charts the future of computer Graphics
is neural rendering the fusion of
artificial intelligence and computer
amazing is oh here we go thank
you this is a surprisingly kinetic
keynote and and uh what's really amazing
is the family of gpus we're going to put
in here and so the 1590 the 1590 will
fit into a laptop a thin laptop that
last laptop was 14 14.9 mm you got a
5080 5070 TI and
5070 okay so ladies and gentlemen the
RTX Blackwell family [Applause]
[Applause]
well GeForce uh brought AI to to the
world democratized AI now ai has come
back and revolutionized GeForce let's
talk about artificial intelligence let's
Nvidia this this is literally our office
headquarters okay so let's talk about
let's talk about AI the
industry is chasing and racing to scale
artificial intelligence int artificial
intelligence and the scaling law is a
powerful model it's an empirical law
that has been observed and demonstrated
by researchers and Industry over several
Generations ations and this the the
scale the scaling law says that the more
data you have the training data that you
have the larger model that you have and
the more compute that you apply to it
therefore the more effective or the more
capable your model will become and so
the scaling law continues what's really
amazing is that now we're moving towards
of course and the internet is producing
about twice twice the amount of data
every single year as it did last year I
think the in the next couple of years we
produce uh Humanity will produce more
data than all of humanity has ever
produced uh since the beginning and so
we're still producing a gigantic amount
of data and it's becoming more
multimodal video and images and sound
all of that data could be used to train
the fundamental knowledge the
foundational knowledge of an AI but
there are in fact two other scaling laws
that has now emerged and it's somewhat
intuitive the second scaling law is post
trining scaling law posttraining scaling
law uses Technologies techniques like
reinforcement learning human feedback
basically the AI produces and generates
answers the hum based on a human query
the human then of course gives a
feedback um it's much more complicated
than that but the reinforcement learning
system uh with a fair number of very
high quality prompts causes the AI to
refine its skills it could find tune its
skills for particular domains it could
be better at solving math problems
better at reasoning so on so forth and
so it's essentially like having a mentor
or having a coach give you feedback um
after you're done going to school and so
you you get test you get feedback you
improve yourself we also have
reinforcement learning AI feedback
and we have synthetic data generation uh
these techniques are rather uh uh Ain to
if you will uh self-practice uh you know
you know the answer to a particular
problem and uh you continue to try it
until you get it right and so an AI
could be presented with a very
complicated and difficult problem that
has that is verifiable U functionally
and has a has an answer that we
understand maybe proving a theorem maybe
solving a solving a uh geometry problem
and so these problems uh would cause the
AI to produce answers and using
reinforcement learning uh it would learn
how to improve itself that's called post
training post training requires an
enormous amount of computation but the
end result produces incredible models we
now have a third scaling law and this
third scaling law has to do with uh
what's called test time scaling test
time scaling is basically when you're
being used when you're using the AI uh
the AI has the ability to now apply a
different resource allocation instead of
improving its parameters now it's
focused on deciding how much computation
to use to produce the answers uh it
wants to
produce reasoning is a way of thinking
about this uh long thinking is a way to
think about this instead of a direct
inference or One-Shot answer you might
reason about you might break down the
problem into multiple steps you might uh
generate multiple ideas and uh evaluate
you know your AI system would evaluate
which one of the ideas that you
generated was the best one maybe it
solves the problem step by step so on so
forth and so now test time scaling has
proven to be incredibly effective you're
watching this sequence of technology and
this all of these scaling laws emerge as
we see incredible achievements from chat
GPT to 01 to 03 and now Gemini Pro all
of these systems are going through this
journey step by step by step of
pre-training to posttraining to test
time scaling well the amount of
computation that we need of course is
incredible and we would like in fact we
would like in fact that Society has the
ability to scale the amount of
computation to produce more and more
novel and better intelligence
intelligence of course is the most
valuable asset that we have and it can
be applied to solve a lot of very
challenging problems and so scaling law
it's driving enormous demand for NVIDIA
Computing it's driving an enormous
demand for this incredible chip we call
Blackwell let's take a look at Blackwell
production it is incredible what it
looks like so first of all there's some
uh every every single cloud service
provider now have systems up and running
uh we have systems here from about 15 uh
15 15 U uh excuse me 15 computer makers
it's being made uh about 200 different
SKS 200 different configurations they're
liquid cooled air cooled x86 Nvidia gray
CPU versions mvlink 36 by 2 MV links 72
by1 whole bunch of different types of
systems so that we can accommodate just
about every single data center in the
world well this these systems are being
currently manufactured in some 45
factories it tells you how pervasive
artificial intelligence is and how much
the industry is jumping onto artificial
intelligence in this new Computing
model well the reason why we're driving
it so hard is because we need a lot more
computation and it's very clear it's
Janine you know
I it's hard to tell you don't ever want
to reach your hands into a dark
place hang a second is this a good
[Applause] [Music]
wait for
worthy apparently yor didn't think I was
worthy all right
this is my show and tell this is a show
and tell so uh this mvlink system this
right here this mvlink system this is
gb200 MV link 72 it is 1 and 12
tons 600,000
Parts approximately equal to 20
it has um a spine behind it that
connects all of these GPU
cable 5,000
cables this is being manufactured in 45
factories around the world we build them
we liquid cool them we test them we
disassemble them shiping parts to the
data centers because it's 1 and A2 tons
we reassemble it outside the data
centers and install them the
manufacturing is insane but the goal of
all of this is because the scaling laws
are driving Computing so hard that this
level of computation Blackwell over our
last generation improves the performance
per watt by a factor of four performance
per watt by a factor of four perform
performance per dollar by a factor of
three that's basically says that in one
generation we reduce the
cost of training these models by a
factor of three or if you want to
increase um the size of your model by a
factor of three it's about the same cost
but the important thing is this these
are generating tokens that are being
used by all of us when we use Chad GPT
or when we use Gemini use our phones in
the future just about all of these
applications are going to be consuming
these AI tokens and these AI tokens are
being generated by these
systems and every single data center is
limited by power
and so if the perf per watt of Blackwell
is four
times our last
generation then the revenue that could
be generated the amount of business that
can be generated in the data center is
increased by a factor of four and so
these AI Factory systems really are
factories today now the goal of all of
this is to so that we can create one
giant chip the amount of computation we
need is really quite incredible and this
is basically one giant chip if we would
have had to build a chip one here we go sorry
sorry
guys you see that that's
here right if we had to build this as
one chip obviously this would be the
size of the wafer but this doesn't
include the impact of yield it would
have to be probably three or four times
the size but what we basically have here
is 72 Blackwell gpus or 144 dieses this
one chip here is 1.4 exop flops the
world's largest supercomputer fastest
supercomputer only recently this entire
room supercomputer only recently
achieved an exf flop plus this is 1.4
exf flops of AI floating Point
performance it has 14 terabytes of
memory but here's the amazing thing the
memory bandwidth is 1.2 petabytes per
second that's basically basically the
entire internet traffic that's happening right
right
now the entire world's internet traffic
is being processed across these chips
okay and we have um 103 130 trillion
transistors in total
2592 CPU
cores whole bunch of networking and so
these I wish I could do this I don't
think I will so these are the black
Wells these are our
connectx networking chips these are the
mvy link and we're trying to pretend
about the Envy the the Envy Ling spine
but that's not possible okay and these
are all of the hbm memories 12 ter 14
terabytes of hbm memory this is what
we're trying to do and this is the
miracle this is the miracle of the
Blackwell system the blackwall dies
right here it is the largest single chip
the world's ever made but yet the
miracle is really in addition to that
this is uh the grace black wall system
well the goal of all of this of course
thanks boy is there a chair I could sit
Ultra how is it possible that we're in
Stadium it's like coming to Nvidia and
you so so we need an enormous the
computation because we want to train
larger and larger models and these
inferences these inferences used to be
one inference but in the future the AI
is going to be talking to itself it's
going to be thinking it's going to be
internally reflecting processing so
today when the tokens are being
generated at you so long as it's coming
out at 20 or 30 tokens per second it's
basically as fast as anybody can read
however in the future and right now with
uh gp1 you know with the new the pre
Gemini Pro and the new GP the the 0103
models they're talking to themselves we
reflecting they thinking and so as you
can imagine the rate at which the tokens
could be ingested is incredibly high and
so we need the token rates the token
generation rates to go way up and we
also have to drive the cost way down
simultaneously so that the C the quality
of service can be extraordinary the cost
to customers can continue to be low and
uh will continue to scale and so that's
the fundamental purpose the reason why
we created MV link well one of the most
important things that's happening in the
world of Enterprise is a Genentech AI a
Genentech AI basically is a perfect
example of test time scaling it's a AI
is a system of models some of it is
understanding interacting with the
customer interacting with the user some
of it is maybe retrieving information
retrieving information from Storage a
semantic AI system like a rag uh maybe
it's going on to to the internet uh
maybe it's uh studying a PDF file and so
it might be using tools it might be
using a calculator and it might be using
a generative AI to uh generate uh charts
and such and it's iter it's taking the
the problem you gave it breaking it down
step by step and it's iterating through
all these different models well in order
to respond to a customer in the future
in order for AI to respond it used to be
ask a question answer start spewing out
in the future you ask a question a whole
bunch bu of models are going to be
working in the background and so test
time scaling the amount of computation
used for inferencing is going to go
through the roof it's going to go
through the roof because we want better
and better answers well to help the the
industry build agentic AI our our go to
market is not direct to Enterprise
customers our go to market is is we work
with software developers in the it
ecosystem to integrate our technology to
make possible new capabilities just like
we did did with Cuda libraries we now
want to do that with AI libraries and
just as the Computing model of the past
has apis that are uh doing computer
Graphics or doing linear algebra or
doing fluid dynamics in the future on
top of those acceleration libraries C
acceleration libraries will have ai
libraries we've created three things for
helping the ecosystem build agentic AI
Nvidia Nims which are essentially AI
microservices all packaged up it takes
all of this really complicated Cuda
software Cuda
DNN cutless or tensor rtlm or Triton or
all of these different really
complicated software and the model
itself we package it up we optimize it
we put it into a container and you could
take it wherever you like and so we have
models for vision for understanding
languages for speech for animation for
digital biology and we have some new new
exciting models coming for physical Ai
and these AI models run in every single
Cloud because nvidia's gpus are now
available in every single Cloud it's
available in every single OEM so you
could literally take these models
integrate it into your software packages
create AI agents that run on Cadence or
they might be S uh service now agents or
they might be sap agents and they could
deploy it to their customers and run it
wherever the customers want to run the
software the next layer is what we call
Nvidia Nemo Nemo is
essentially a digital employee
onboarding and training evaluation
system in the future these AI agents are
essentially digital Workforce that are
working alongside your employees um
working Al doing things for you on your
behalf and so the way that you would
bring these specialized agents into your
these special agents into your company
is to onboard them just like you onboard
an employee and so we have different
libraries that helps uh these AI agents
be uh trained for the type of you know
language in your company maybe the
vocabulary is unique to your company the
business process is different the way
you work is different so you would give
them examples of what the work product
should look like and they would try to
generate and you would give a feedback
and then you would evaluate them so on
so forth and so that uh and you would
guardrail them you say these are the
things that you're not allowed to do
these are things you're not allowed to
say this and and we even give them
access to certain information okay so
that entire pipeline a digital employee
pipeline is called Nemo in a lot of ways
the IT department of every company is
going to be the HR department of AI
agents in the
future today they manage and maintain a
bunch of software from uh from the IT
industry in the future they will Main
maintain you know nurture onboard and
improve a whole bunch of digital agents
and provision them to the companies to
use okay and so your H your it
department is going to become kind of
like AI agent HR and on top of that we
provide a whole bunch of blueprints that
our ecosystem could could uh take
advantage of all of this is completely
open source and so you could take take
it and uh modify the blueprints we have
blueprints for all kinds of different
different types of Agents well today
we're also announcing that we're doing
something that's really cool and I think
really clever we're announcing a whole
family of models that are based off of
llama the Nvidia llama neotron language
Foundation models llama 3.1 is a complete
complete
phenomenon the download of llama 3.1
from meta 350 650,000 times something
like that it has
been der red and turned into other
models uh about 60,000 other different
models it it is singularly the reason
why just about every single Enterprise
and every single industry has been
activated to start working on AI well
the thing that we did was we realized
that the Llama models really could be
better fine-tuned for Enterprise use and
so we fine-tune them using our expertise
and our capabilities and we turn them
into the Llama neotron Suite of open
models there are small ones that
interact in uh very very fast response
time extremely small uh they're uh sup
what we call Super llama neotron supers
they're basically your mainstream
versions of your models or your Ultra
model the ultra model could be used uh
to be a teacher model for a whole bunch
of other models it could be a reward
model evaluator uh a judge for other
models to create answers and decide
whether it's a good answer or not
give basically give feedback to other
models it could be distilled in a lot of
different ways basically a teacher model
a knowledge distillation uh uh model
very large very capable and so all of
this is now available online well these
models are incredible it's a a number
one in leaderboards for chat leaderboard
for instruction uh lead leaderboard for
retrieval um so the different types of
functionalities necessary that are used
in AI agents around the world uh these
are going to be incredible models for
you we're also working with uh the
ecosystem these Tech all of our Nvidia
AI Technologies are integrated into uh
uh the it in Industry uh we have great
partners and really great work being
done at service now at sap at Seaman uh
for industrial AI uh Cadence is during
great work synopsis doing great work I'm
really proud of the work that we do with
perplexity as you know they
revolutionize search yeah really
fantastic stuff uh codium uh every every
software engineer in the world this is
going to be the next giant AI
application next giant AI service period
is software coding 30 million software
Engineers around the world everybody is
going to have a software assistant uh
helping them code uh if if um if not
obviously you're just you're going to be
way less productive and create lesser
good code and so this is 30 million
there's a billion knowledge workers in
the world it is very very clear AI
agents is probably the next robotics
industry and likely to be a
multi-trillion dollar opportunity well
let me show you some of the uh
blueprints that we've created and some
of the work that we've done with our
agents AI agents are the new digital
Workforce working for and with
us AI agents are a system of models that
reason about a mission break it down
into tasks and retrieve data or use
tools to generate a quality
response nvidia's agentic AI building
blocks Nim pre-trained models and Nemo
framework let organizations easily
develop AI agents and deploy them
anywhere we will onboard and train our
agentic workforces on our company's
methods like we do for
employees AI agents are domain specific
task experts let me show you four
examples for the billions of knowledge
workers and students AI research
assistant agents ingest complex
documents like lectures journals
Financial results and generate
interactive podcasts for easy learning
by combining a unet regression model
with a diffusion model cordi can
downscale global weather forecasts down
from 25 km to 2
km developers like at Nvidia manage
software security AI agents that
continuously scan software for
vulnerabilities alerting developers to
what action is
needed Virtual Lab AI agents help
researchers design and Screen billions
of compounds to find promising drug
candidates faster than
ever Nvidia analytics AI agents built on
an Nvidia metr blueprint including
Nvidia Cosmos nimron Vision language
models llama neaton llms and Nemo
retriever Metropolis agents analyze
content from the billions of cameras
generating 100,000 pedes of video per
day they enable interactive search
summarization and automated
reporting and help monitor traffic flows
flagging congestion or danger
in industrial facilities they monitor
processes and generate recommendations or
or
Improvement Metropolis agents centralize
data from hundreds of cameras and can
reroute workers or robots when incidents
occur the age of agentic AI is here for every
that was the first pitch at a baseball
that was not generated I just felt that
none of you were
impressed okay so ai ai was was created
in the cloud and for the cloud AI is
creating the cloud for the cloud and for
uh enjoying AI on on phones of course
it's perfect um very very soon we're
going to have a continuous AI that's
going to be with you and when you use
those metag glasses you could of course
uh point at something look at something
and and ask it you know whatever
information you want and so AI is is
perfect in the CL was creating the cloud
is perfect in the cloud however we would
love to be able to take that AI
everywhere I've mentioned already that
you could take Nvidia AI to any Cloud
but you could also put it inside your
company but the thing that we want to do
more than anything is put it on our PC
as well and so as you know Windows 95
revolutionized the computer industry it
made possible this new Suite of
multimedia services and it change the
way that applications was created
forever um Windows 95 this this model of
computing of course is not perfect for
AI and so the thing that we would like
to do is we would like to have in the
future your AI basically become your AI
assistant and instead of instead of just
the the 3D apis and the sound apis and
the video API you would have generative
apis generative apis for 3D and
generative apis for language and
generative AI for sound and so on so
forth and we need a system that makes
that possible while leveraging the
massive investment that's in the cloud
there's no way that we could the world
can create yet another way of
programming AI models it's just not
going to happen and so if we could
figure out a way to make Windows
PC a worldclass
aipc um it would be completely awesome
and it turns out the answer is Windows
it's Windows wsl2 Windows wsl2 Windows
wsl2 basically it's two operating
systems within one it works perfectly
it's developed for developers and it's
developed uh uh so that you can have
access to Bare Metal it's been wsl2 has been
been
optimized optimized for cloud native
applications it is optimized for and
very importantly it's been optimized for
Cuda and so wsl2 supports Cuda perfectly
out of the box as a
result everything that I showed you with
Nvidia Nims Nvidia Nemo the blueprints
that we develop that are going to be up
in ai. nvidia.com so long as the
computer fits it so long as you can fit
that model and we're going to have many
models that that fit whether it's Vision
models or language models or speech
models or these animation human digital
human models all kinds of different
different types of models are going to
be perfect for your PC and it would you
download it and it should just run and
so our focus is to turn Windows wsl2
Windows PC into a Target first class
platform that we will support and
maintain for as long as we shall live
and so this is an incredible thing for
engineers and developers everywhere let
let me show you something that we can do
with that this is one of the examples of
you generative AI synthesizes amazing
images from Simple Text prompts yet
image composition can be challenging to
control using only words with Nvidia Nim
microservices creators can use Simple 3D
objects to guide AI image generation
let's see how a concept artist can use
this technology to develop the look of a
scene they start by laying out 3D assets
created by hand or generated with AI
then use an image generation Nim such as
flux to create a visual that adheres to
the 3D
composition change camera angles to
frame the perfect
shot or reimagine the whole scene with a new
prompt assisted by generative AI and
Nvidia Nim and artists can quickly
realize their [Music]
[Music]
Vision Nvidia AI for your
PCS hundreds of millions of PCS in the
world with Windows and so we could get
them ready for AI uh oems all the PC
oems we work with just basically all of
the world's leading PC oems are going to
get their PCS ready for this stack and
good okay let's talk about physical
AI speaking of Linux let's talk about physical
physical
AI So Physical AI imagine
imagine whereas your large language
model you give it your your context your
prompt on the left and it generates
tokens one at a time to produce the
output that's basically how it works the
amazing thing is this model in the
middle is quite large has billions of
parameters the context length is
incredibly large because you might
decide to load in a PDF in my case I
might load in several PDFs before I ask
it a question those PDFs are turned into
tokens the attention the basic attention
characteristic of a transformer has
every single token find its relationship
and relevance against every other token
so you could have hundreds of thousands
of tokens and the computational load
increases quadratically and it does this
that all of the parameters all of the
input sequence process it through every
single layer of the Transformer and it
produces one token that's the reason why
we needed blackw
and then the next token is produced when
the current token is done it puts the
current token into the input sequence
and takes that whole thing and generates
the next token it does it one at a time
this is the Transformer model it's the
reason why it is so so incredibly
effective computationally demanding What
If instead of PDFs it's your surrounding
and what if instead of the prompt a
question it's a request go over there
and pick up that that you know that box
and bring it back and instead of what is
produced in tokens its text it produces action
action
tokens well that I just described is a
very sensible thing for the future of
Robotics and the technology is right
around the corner but what we need to do
is we need to create the effective
effectively the world
model of you know as opposed to GPT
which is a language model and this World
model has to understand the language of
the world it has to understand physical
Dynamics things like gravity and
friction and inertia it has to
understand geometric and spatial
relationships it has to understand cause
and effect if you drop something a fall
to the ground if you you know poke at it
it tips over it has to understand object
permanence if you roll a ball over the
kitchen counter when it goes off the
other side the ball didn't leave into
another quantum universe that that's
still there and so all of these types of
understanding is intuitive understanding
that we know that most models today have
a very hard time with and so we would
like to create a world we need a world
Foundation model today we're announcing
a very big thing we're announcing Nvidia
Cosmos a world Foundation model that is
designed that was created to understand
the physical world and the only way for
you to really understand this is to see
flip the next Frontier of AI is physical
AI model performance is directly related
to data availability but physical world
data is costly to capture curate and
label Nvidia Cosmos is a world
Foundation model development platform to
Advance Physical AI it includes Auto
regressive world found Foundation models
diffusion-based World Foundation models Advanced
Advanced
tokenizers and an Nvidia Cuda an AI
pipeline Cosmos models ingest text image
or video prompts and generate virtual
world States as
videos Cosmos Generations prioritize the
unique requirements of Av and Robotics
use cases like real world environments
lighting and object permanence
developers use Nvidia Omniverse to build physics-based
physics-based
geospatially accurate scenarios then
output Omniverse renders into Cosmos
which generates photoreal physically
based synthetic [Music]
data whether diverse
conditions like weather or time of day
or Edge case
scenarios developers use Cosmos to
generate worlds for reinforcement
learning AI feedback to improve policy
models or to test and validate model
performance even across multisensor
views Cosmos can generate tokens in real
time bringing the power of foresight and
Multiverse simulation to AI models
generating every possible future to help
the model select the right
path working with the world's developer
ecosystem Nvidia is helping Advance the
[Music]
AI Nvidia
Cosmos Nvidia
Cosmos Nvidia Cosmos the world's first
world Foundation model it is trained on
20 million hours of video the 20 million
hours of video focuses on physical
Dynamic things so n n Dynamic nature
nature themes themes uh humans uh
walking uh hands moving uh manipulating
things uh you know things that are uh
fast camera movements it's really about
teaching the AI not about generating
creative content but teaching the AI to
understand the physical world and from
this with this physical AI there are
many Downstream things that we could uh
do as a result we could do synthetic
data generation to train uh models we
could distill it and turn it into
effectively the seed the beginnings of a
robotics model you could have it
generate multiple physically based
physically plausible uh scenarios that
the future basically do a doctor strange
um you could uh because because this
model understands the physical world of
course you saw a whole bunch of images
generated this model understanding the
physical world it also uh could do of
course captioning and so it could take
videos caption it incredibly well and
that captioning and the video could be
used to train large language models
multimodality large language models and
uh so you could use this technology to
use this Foundation model to train
robotics robots as well as larger
language models and so this is the
Nvidia Cosmos the platform has an auto
regressive model for real-time
applications has diffusion model for a
very high quality image generation it's
incredible tokenizer basically learning
the vocabulary of uh real world and a
data pipeline so that if you would like
to take all of this and then train it on
your own data this data pipeline because
there's so much data involved we've
accelerated everything end to endend for
you and so this is the world's first data processing pipeline that's Cuda
data processing pipeline that's Cuda accelerated as well as AI accelerated
accelerated as well as AI accelerated all of this is part of the cosmos
all of this is part of the cosmos platform and today we're announcing that
platform and today we're announcing that Cosmos is open licensed it's open
Cosmos is open licensed it's open available on
GitHub we hope we hope that this moment and there's a there's a small medium
and there's a there's a small medium large for uh uh very fast models um you
large for uh uh very fast models um you know mainstream models and also teacher
know mainstream models and also teacher models basically not knowledge transfer
models basically not knowledge transfer models Cosmo Cosmos World Foundation
models Cosmo Cosmos World Foundation model being open we really hope will do
model being open we really hope will do for the world of Robotics and Industrial
for the world of Robotics and Industrial AI what llama 3 has done for Enterprise
AI what llama 3 has done for Enterprise AI the magic happens when you connect
AI the magic happens when you connect Cosmos to Omniverse and the reason
Cosmos to Omniverse and the reason fundamentally is this Omniverse is a
fundamentally is this Omniverse is a physics grounded not physically grounded
physics grounded not physically grounded but physics grounded it's algorithmic
but physics grounded it's algorithmic physics principled physics simulation
physics principled physics simulation grounded system it's a simulator when
grounded system it's a simulator when you connect that to
you connect that to Cosmos it provides the grounding the
Cosmos it provides the grounding the ground truth that can control and to
ground truth that can control and to condition the Osmos generation as a
condition the Osmos generation as a result what comes out of Osmos is
result what comes out of Osmos is grounded on Truth this is exactly the
grounded on Truth this is exactly the same idea as connecting a large language
same idea as connecting a large language model model to a rag to a retrieval
model model to a rag to a retrieval augmented generation system you want to
augmented generation system you want to ground the AI generation on ground truth
ground the AI generation on ground truth and so the combination of the two gives
and so the combination of the two gives you a
you a physically simulated a physically
physically simulated a physically grounded Multiverse generator and the
grounded Multiverse generator and the application the use cases are really
application the use cases are really quite exciting and of course uh for
quite exciting and of course uh for robotics uh for industrial applications
robotics uh for industrial applications uh it is very very clear this Cosmos
uh it is very very clear this Cosmos plus
plus o Omniverse plus Cosmos represents the
o Omniverse plus Cosmos represents the Third computer that's necessary for
Third computer that's necessary for building robotic systems every robotics
building robotic systems every robotics company will ultimately have to build
company will ultimately have to build three computers a robotics the robotics
three computers a robotics the robotics system could be a factory the robotics
system could be a factory the robotics system could be a car it could be a
system could be a car it could be a robot you need three fundamental
robot you need three fundamental computers one computer of course to
computers one computer of course to train the AI we call the dgx computer to
train the AI we call the dgx computer to train the AI another of course when
train the AI another of course when you're done to deploy the AI we call
you're done to deploy the AI we call that agx that's inside the car in the
that agx that's inside the car in the robot or in an AMR or you know at the uh
robot or in an AMR or you know at the uh in a in a stadium or whatever it is
in a in a stadium or whatever it is these computers are at the edge and
these computers are at the edge and they're autonomous but to connect the
they're autonomous but to connect the two you need a digital twin and this is
two you need a digital twin and this is all the simulations that you were seeing
all the simulations that you were seeing the digital twin is where the AI that
the digital twin is where the AI that has been trained goes to practice to be
has been trained goes to practice to be refined to do its synthetic data
refined to do its synthetic data generation reinforcement learning AI
generation reinforcement learning AI feedback such and such and so it's the
feedback such and such and so it's the digital twin of the AI these three
digital twin of the AI these three computers are going to be working
computers are going to be working interactively nvidia's strategy for uh
interactively nvidia's strategy for uh the industrial world and we've been
the industrial world and we've been talking about this for some time is this
talking about this for some time is this three computer
three computer system you know instead of a three three
system you know instead of a three three body problem we have a three Computer
body problem we have a three Computer Solution and so it's the Nvidia
robotics so let me give you three examples
examples all right so the first example is uh uh
all right so the first example is uh uh how we apply apply all of this to
how we apply apply all of this to Industrial digitalization there millions
Industrial digitalization there millions of factories hundreds of thousands of
of factories hundreds of thousands of warehouses that's basically it's the
warehouses that's basically it's the backbone of A50 trillion doll
backbone of A50 trillion doll manufacturing industry all of that has
manufacturing industry all of that has to become software defined all of that
to become software defined all of that has has to have Automation in the future
has has to have Automation in the future and all of it will be infused with
and all of it will be infused with robotics well we're partnering with Keon
robotics well we're partnering with Keon the world's leading Warehouse automation
the world's leading Warehouse automation Solutions provider and Accenture the
Solutions provider and Accenture the world's largest professional services
world's largest professional services provider and they have a big focus in
provider and they have a big focus in digital manufacturing and we're working
digital manufacturing and we're working together to create something that's
together to create something that's really special and I'll show you that in
really special and I'll show you that in the second but our go to market is
the second but our go to market is essentially the same as all of the other
essentially the same as all of the other software uh platforms and all the
software uh platforms and all the technology platforms that we have
technology platforms that we have through the uh developers and ecosystem
through the uh developers and ecosystem Partners uh and we have just just a
Partners uh and we have just just a growing number of ecosystem Partners
growing number of ecosystem Partners connecting to Omniverse and the reason
connecting to Omniverse and the reason for that is very clear everybody wants
for that is very clear everybody wants to digitalize the future of Industries
to digitalize the future of Industries there's so much waste so much
there's so much waste so much opportunity for Automation in that $50
opportunity for Automation in that $50 trillion doar of the world's GDP so
trillion doar of the world's GDP so let's take a look at that this one one p
let's take a look at that this one one p one example that we're doing with Keon
one example that we're doing with Keon and
Accenture Keon the supply chain solution company Accenture a global leader in
company Accenture a global leader in Professional Services and Nvidia are
Professional Services and Nvidia are bringing physical AI to the $1 trillion
bringing physical AI to the $1 trillion warehouse and Distribution Center Market
warehouse and Distribution Center Market managing high- Performance Warehouse
managing high- Performance Warehouse Logistics involves navigating a complex
Logistics involves navigating a complex web of decisions influenced by
web of decisions influenced by constantly shifting variables these
constantly shifting variables these include daily and seasonal demand
include daily and seasonal demand changes space constraints Workforce
changes space constraints Workforce availability and the integration of of
availability and the integration of of diverse robotic and automated systems
diverse robotic and automated systems and predicting operational kpis of a
and predicting operational kpis of a physical Warehouse is nearly impossible
physical Warehouse is nearly impossible today to tackle these challenges Keon is
today to tackle these challenges Keon is adopting Mega an Nvidia Omniverse
adopting Mega an Nvidia Omniverse blueprint for building industrial
blueprint for building industrial digital twins to test and optimize
digital twins to test and optimize robotic fleets first Keon's warehouse
robotic fleets first Keon's warehouse management solution assigns tasks to the
management solution assigns tasks to the industrial AI brains in the digital twin
industrial AI brains in the digital twin such as moving a load from from a buffer
such as moving a load from from a buffer location to a shuttle storage
location to a shuttle storage solution the robot's brains are in a
solution the robot's brains are in a simulation of a physical Warehouse
simulation of a physical Warehouse digitalized into Omniverse using open
digitalized into Omniverse using open USD connectors to aggregate CAD video
USD connectors to aggregate CAD video and image to 3D Light Art to point cloud
and image to 3D Light Art to point cloud and AI generated data the fleet of
and AI generated data the fleet of robots execute tasks by perceiving and
robots execute tasks by perceiving and reasoning about their Omniverse digital
reasoning about their Omniverse digital twin environment planning their next
twin environment planning their next motion and acting
motion and acting the robot brains can see the resulting
the robot brains can see the resulting State through sensor simulations and
State through sensor simulations and decide their next action the loop
decide their next action the loop continues while Mega precisely tracks
continues while Mega precisely tracks the state of everything in the digital
the state of everything in the digital twin now Keon can simulate infinite
twin now Keon can simulate infinite scenarios at scale while measuring
scenarios at scale while measuring operational kpis such as throughput
operational kpis such as throughput efficiency and utilization all before
efficiency and utilization all before deploying changes to the physical
deploying changes to the physical Warehouse together with Nvidia
Warehouse together with Nvidia Keon and Accenture are Reinventing
Keon and Accenture are Reinventing industrial
industrial autonomy in the future is that that's
autonomy in the future is that that's incredible everything is in
incredible everything is in simulation in the future in the future
simulation in the future in the future every Factory will have a digital twin
every Factory will have a digital twin and that digital twin operates exactly
and that digital twin operates exactly like the real factory and in fact you
like the real factory and in fact you could use Omniverse with Cosmos to
could use Omniverse with Cosmos to generate a whole bunch of future
generate a whole bunch of future scenarios and you pick then an AI
scenarios and you pick then an AI decides which which one of the scenarios
decides which which one of the scenarios are the most optimal for whatever kpis
are the most optimal for whatever kpis and that becomes the programming
and that becomes the programming constraints the program if you will the
constraints the program if you will the AI that will be uh deployed into the
AI that will be uh deployed into the real factories the next example
real factories the next example autonomous vehicles the AV revolution
autonomous vehicles the AV revolution has arrived after so many years with weo
has arrived after so many years with weo success and Tesla's success it is very
success and Tesla's success it is very very clear autonomous vehicles has
very clear autonomous vehicles has finally arrived well our offering to
finally arrived well our offering to this industry is the three computers the
this industry is the three computers the training systems the training the AIS
training systems the training the AIS the simulation systemss and and the and
the simulation systemss and and the and the synthetic data generation systems
the synthetic data generation systems Omniverse and now Cosmos and also the
Omniverse and now Cosmos and also the computer that's inside the car each car
computer that's inside the car each car company might might work with us in a
company might might work with us in a different way use one or two or three of
different way use one or two or three of the computers we're working with just
the computers we're working with just about every major car company around the
about every major car company around the world whmo and zuk and Tesla of course
world whmo and zuk and Tesla of course in their data center byd the largest uh
in their data center byd the largest uh EV company in the world jlr has got a
EV company in the world jlr has got a really cool car coming Mercedes because
really cool car coming Mercedes because a fleet of cars coming with Nvidia
a fleet of cars coming with Nvidia starting with this starting this year
starting with this starting this year going to production and I'm super super
going to production and I'm super super pleased to announce that today Toyota
pleased to announce that today Toyota and Nvidia are going to partner together
and Nvidia are going to partner together to create their next Generation
AVS just so many so many cool companies uh lucid and rivan and Shi and of course
uh lucid and rivan and Shi and of course uh Volvo just so many different
uh Volvo just so many different companies Wabi is uh building uh
companies Wabi is uh building uh self-driving trucks Aurora we announced
self-driving trucks Aurora we announced this week also that Aurora is going to
this week also that Aurora is going to use Nvidia to build self-driving trucks
use Nvidia to build self-driving trucks autonomous 100 million cars build each
autonomous 100 million cars build each year a billion cars vehicles on a road
year a billion cars vehicles on a road all over the world a trillion miles that
all over the world a trillion miles that are driven around the world each year
are driven around the world each year that's all going to be either highly
that's all going to be either highly autonomous or you know fully autonomous
autonomous or you know fully autonomous coming up and so this is going to be a
coming up and so this is going to be a very L very large industry I predict
very L very large industry I predict that this will likely be the first
that this will likely be the first multi-trillion dollar
multi-trillion dollar robotics industry this IND this business
robotics industry this IND this business for us um notice in just just a few of
for us um notice in just just a few of these cars that are starting to ramp
these cars that are starting to ramp into the world uh our business is
into the world uh our business is already $4 billion and this year
already $4 billion and this year probably on a run rate of about $5
probably on a run rate of about $5 billion so really significant business
billion so really significant business already this is going to be very large
already this is going to be very large well today we're announcing that our
well today we're announcing that our next generation processor for the car
next generation processor for the car our next generation computer for the car
our next generation computer for the car is called Thor I have one right here
is called Thor I have one right here hang on a second
okay this is Thor this is
Thor this is Thor this is this is a robotics
Thor this is this is a robotics computer this is a robotics computer
computer this is a robotics computer takes sensors and just a Madness amount
takes sensors and just a Madness amount of sensor information process it you
of sensor information process it you know een teed cameras high resolution
know een teed cameras high resolution Radars Liars they're all coming into
Radars Liars they're all coming into this chip and this chip has to process
this chip and this chip has to process all that sensor turn them into tokens
all that sensor turn them into tokens put them into a Transformer and predict
put them into a Transformer and predict the next PATH and this AV computer is
the next PATH and this AV computer is now in full production Thor is 20 times
now in full production Thor is 20 times the processing capability of our last
the processing capability of our last generation Orin which is really the
generation Orin which is really the standard of autonomous vehicles today
standard of autonomous vehicles today and so this is just really quite quite
and so this is just really quite quite incredible Thor is in full production
incredible Thor is in full production this robotics processor by the way also
this robotics processor by the way also goes into a full robot and so it could
goes into a full robot and so it could be an AMR it could be a human or robot
be an AMR it could be a human or robot could be the brain it could be the
could be the brain it could be the manipulator this Rob this processor
manipulator this Rob this processor basically is a universal robotics
basically is a universal robotics computer the second part of our drive
computer the second part of our drive system that I'm incredibly proud of is
system that I'm incredibly proud of is the dedication to safety Drive OS I'm
the dedication to safety Drive OS I'm pleased to announce is now the first
pleased to announce is now the first softwar defined programmable AI computer
softwar defined programmable AI computer that has been certified up to asold D
that has been certified up to asold D which is the highest standard of
which is the highest standard of functional safety for automobiles the
functional safety for automobiles the only and the highest and so I'm really
only and the highest and so I'm really really proud of this asold ISO
really proud of this asold ISO 26262 it is um the work of some 15,000
26262 it is um the work of some 15,000 engineering years this is just
engineering years this is just extraordinary work and as a result of
extraordinary work and as a result of that Cuda is now a functional safe
that Cuda is now a functional safe computer and so if you're building a
computer and so if you're building a robot Nvidia Cuda y
okay so so now I wanted to I told you I was going to show you what would we use
was going to show you what would we use Omniverse and Cosmos to do in the
Omniverse and Cosmos to do in the context of self-driving cars and you
context of self-driving cars and you know today instead of showing you a
know today instead of showing you a whole bunch of uh uh videos of of cars
whole bunch of uh uh videos of of cars driving on the road I'll show you some
driving on the road I'll show you some of that too um but I want to show you
of that too um but I want to show you how we use the car to reconstruct
how we use the car to reconstruct digital twins automatically using Ai and
digital twins automatically using Ai and use that capability to train future am
use that capability to train future am models okay let's play
it the autonomous vehicle Revolution is here building autonomous vehicles like
here building autonomous vehicles like all robots requires three computers
all robots requires three computers Nvidia dgx to train AI models Omniverse
Nvidia dgx to train AI models Omniverse to test drive and generate synthetic
to test drive and generate synthetic data and drive agx a supercomputer in
data and drive agx a supercomputer in the car
the car building safe autonomous vehicles means
building safe autonomous vehicles means addressing Edge scenarios but real world
addressing Edge scenarios but real world data is limited so synthetic data is
data is limited so synthetic data is essential for
essential for training the autonomous vehicle data
training the autonomous vehicle data Factory powered by Nvidia Omniverse AI
Factory powered by Nvidia Omniverse AI models and Cosmos generates synthetic
models and Cosmos generates synthetic driving scenarios that enhance training
driving scenarios that enhance training data by orders of
data by orders of magnitude first omnimap fuses map and
magnitude first omnimap fuses map and geospatial data to construct drivable 3D
environments driving scenario variations can be generated from replay Drive logs
can be generated from replay Drive logs or AI traffic
or AI traffic generators next a neural reconstruction
generators next a neural reconstruction engine uses autonomous vehicle sensor
engine uses autonomous vehicle sensor logs to create High Fidelity 4D
logs to create High Fidelity 4D simulation
simulation environments it replays previous drives
environments it replays previous drives in 3D and generates scenario Vari ations
in 3D and generates scenario Vari ations to amplify training
to amplify training data finally edify 3DS automatically
data finally edify 3DS automatically searches through existing asset
searches through existing asset libraries or generates new assets to
libraries or generates new assets to create Sim ready
scenes the Omniverse scenarios are used to condition Cosmos to generate massive
to condition Cosmos to generate massive amounts of photo realistic data reducing
amounts of photo realistic data reducing the Sim toore
the Sim toore Gap and with text prompts generate near
Gap and with text prompts generate near infinite variations of the driving
infinite variations of the driving scenario with Cosmos neotron video
scenario with Cosmos neotron video search the massively scaled synthetic
search the massively scaled synthetic data set combined with recorded drives
data set combined with recorded drives can be curated to train
models nvidia's AI data Factory scales hundreds of drives into billions of
hundreds of drives into billions of effective miles setting the standard for
effective miles setting the standard for safe and advanced autonomous driving
safe and advanced autonomous driving [Music]
we take take thousands of drives and turn them into billions of miles we are
turn them into billions of miles we are going to have mountains of training data
going to have mountains of training data for autonomous vehicles of course we
for autonomous vehicles of course we still need actual cars on the road of
still need actual cars on the road of course we will continuously collect data
course we will continuously collect data for as long as we shall live however
for as long as we shall live however synthetic data generation using this
synthetic data generation using this Multiverse physically based physically
Multiverse physically based physically grounded capability so that we generate
grounded capability so that we generate data for training AIS that are
data for training AIS that are physically grounded and accurate and or
physically grounded and accurate and or plausible so that we could have an
plausible so that we could have an enormous amount of data to train with
enormous amount of data to train with the AV industry is here uh this is an
the AV industry is here uh this is an incredibly exciting time super super
incredibly exciting time super super super uh uh excited about the next
super uh uh excited about the next several years I think you're going to
several years I think you're going to see just as computer Graphics was
see just as computer Graphics was revolutionized such incredible pace
revolutionized such incredible pace you're going to see the pace of Av
you're going to see the pace of Av development increasing tremendously over
development increasing tremendously over the next several
years I I think I think um I I think the next part is is
um I I think the next part is is robotics so um
[Applause] friends the chat GPT moment for General
friends the chat GPT moment for General robotics is just around the corner and
robotics is just around the corner and in fact all of the enabling technologies
in fact all of the enabling technologies that I've been talking about is going to
that I've been talking about is going to make it possible for us in the next
make it possible for us in the next several years to see very rapid break
several years to see very rapid break breakthroughs surprising breakthroughs
breakthroughs surprising breakthroughs in in general robotics now the reason
in in general robotics now the reason why General robotics is so important is
why General robotics is so important is whereas robots with tracks and wheels
whereas robots with tracks and wheels require special environments to
require special environments to accommodate them there are three
accommodate them there are three robots three robots in the world that we
robots three robots in the world that we can make that require no green
can make that require no green fields Brown field adaptation is perfect
fields Brown field adaptation is perfect if we if we could possibly build these
if we if we could possibly build these amazing robots we could deploy them in
amazing robots we could deploy them in exactly the world that we've built for
exactly the world that we've built for ourselves these three robots are one
ourselves these three robots are one agentic robots agentic AI because you
agentic robots agentic AI because you know they're information workers so long
know they're information workers so long as they could accommodate uh the
as they could accommodate uh the computers that we have in our offices is
computers that we have in our offices is going to be great number two
going to be great number two self-driving cars and the reason for
self-driving cars and the reason for that is we spent 100 plus years building
that is we spent 100 plus years building roads and cities and then number three
roads and cities and then number three human or robots if we have the
human or robots if we have the technology to solve these three this
technology to solve these three this will be the largest technology industry
will be the largest technology industry IND the world's ever seen and so we
IND the world's ever seen and so we think that robotics era is just around
think that robotics era is just around the corner the critical capability is
the corner the critical capability is how to train these robots in the case of
how to train these robots in the case of human or
human or robots the imitation information is
robots the imitation information is rather hard to collect and the reason
rather hard to collect and the reason for that is uh in the case of car you
for that is uh in the case of car you just drive it we're driving cars all the
just drive it we're driving cars all the time in the case of these human robots
time in the case of these human robots the imitation information the the human
the imitation information the the human demonstration is rather laborious is to
demonstration is rather laborious is to do and so we need to come up with a
do and so we need to come up with a clever way to take hundreds of
clever way to take hundreds of demonstrations thousands of human
demonstrations thousands of human demonstrations and somehow use
demonstrations and somehow use artificial intelligence and
artificial intelligence and Omniverse to synthetically
Omniverse to synthetically generate
generate millions
millions of
of synthetically generated motions and from
synthetically generated motions and from those motions the AI can learn uh how to
those motions the AI can learn uh how to perform a task let me show you how
perform a task let me show you how that's
done developers around the world are building the next wave of physical AI
building the next wave of physical AI embodied robots
embodied robots humanoids developing general purpose
humanoids developing general purpose robot models requires massive amounts of
robot models requires massive amounts of real world data which is costly to
real world data which is costly to capture and
capture and curate Nvidia Isaac Groot helps tackle
curate Nvidia Isaac Groot helps tackle these challenges providing humanoid
these challenges providing humanoid robot developers with four things robot
robot developers with four things robot Foundation
Foundation models data
models data pipelines simulation
pipelines simulation Frameworks and a Thor robotics
Frameworks and a Thor robotics computer the Nvidia Isaac Groot
computer the Nvidia Isaac Groot blueprint for synthetic motion
blueprint for synthetic motion generation is a simulation workflow for
generation is a simulation workflow for imitation learning enabling developers
imitation learning enabling developers to generate exponentially large data
to generate exponentially large data sets from a small number of
sets from a small number of demonstrations first Groot teleop
demonstrations first Groot teleop enables skilled human workers to portal
enables skilled human workers to portal into a digital twin of their robot using
into a digital twin of their robot using the Apple Vision
the Apple Vision Pro this means operators can capture
Pro this means operators can capture data even without a physical robot and
data even without a physical robot and they can operate the robot in a
they can operate the robot in a risk-free environment eliminating the
risk-free environment eliminating the chance of physical damage or wear and
chance of physical damage or wear and tear to teach a robot a single task
tear to teach a robot a single task operators capture motion trajectories
operators capture motion trajectories through a handful of teleoperated
through a handful of teleoperated demonstrations then use Groot mimic to
demonstrations then use Groot mimic to multiply these trajectories into a much
multiply these trajectories into a much larger data
larger data set next they use Gro gen built on
set next they use Gro gen built on Omniverse and Cosmos for domain
Omniverse and Cosmos for domain randomization and 3D to real
randomization and 3D to real upscaling generating an exponentially
upscaling generating an exponentially larger data
larger data set the Omniverse and Cosmos Multiverse
set the Omniverse and Cosmos Multiverse simulation engine provides a massively
simulation engine provides a massively scaled data set to train the robot
scaled data set to train the robot policy once the policy is trained
policy once the policy is trained developers can perform software in the
developers can perform software in the loop testing and validation in Isaac Sim
loop testing and validation in Isaac Sim before deploying to the real
before deploying to the real robot the age of General robotics is
robot the age of General robotics is arriving powered by Nvidia Isaac
Groot we're going to have mountains of data to train robots with
Nvidia Isaac group Nvidia Isaac group this is our platform to provide
this is our platform to provide technology platform technology elements
technology platform technology elements to the robotics industry to accelerate
to the robotics industry to accelerate the development of General
the development of General Robotics and um well I have one more
Robotics and um well I have one more thing that I want to show you none of
thing that I want to show you none of none of this none of this would be
none of this none of this would be possible if not for uh this incredible
possible if not for uh this incredible project that we started uh about a
project that we started uh about a decade ago inside the company what
decade ago inside the company what called project project digits deep
called project project digits deep learning GPU intelligence training
learning GPU intelligence training system
system digits well before we launched it uh I
digits well before we launched it uh I shrunk it to
shrunk it to dgx and to harmonize it with
dgx and to harmonize it with RTX agx ovx and all of the other X's
RTX agx ovx and all of the other X's that we have in the company and and um I
that we have in the company and and um I and and it really revolutionized uh djx1
and and it really revolutionized uh djx1 really
really revolutionized where where's djx1
revolutionized where where's djx1 dgx-1 revolutionized artificial
dgx-1 revolutionized artificial intelligence the reason why we built it
intelligence the reason why we built it was because we wanted to uh make it
was because we wanted to uh make it possible for researchers and startups to
possible for researchers and startups to have an out-of-the-box AI supercomputer
have an out-of-the-box AI supercomputer imagine the way supercomputers were
imagine the way supercomputers were built in the past you really have to uh
built in the past you really have to uh build your own facility and you have to
build your own facility and you have to go build your own infrastructure and
go build your own infrastructure and really engineer it into existence and so
really engineer it into existence and so we created a supercomputer for AI for AI
we created a supercomputer for AI for AI development for researchers and and
development for researchers and and startups that comes literally one out of
startups that comes literally one out of the box I delivered the first one to a
the box I delivered the first one to a startup company in 2016 called open Ai
startup company in 2016 called open Ai and Elon was there and and Ilia sus was
and Elon was there and and Ilia sus was there and many of Nvidia Engineers were
there and many of Nvidia Engineers were there and and um uh we we celebrated the
there and and um uh we we celebrated the arrival of djx1 and obviously uh it
arrival of djx1 and obviously uh it revolutionized uh artificial
revolutionized uh artificial intelligence and Computing um but now
intelligence and Computing um but now artificial intelligence is everywhere
artificial intelligence is everywhere it's not just in researchers and and and
it's not just in researchers and and and startup Labs you know we want artificial
startup Labs you know we want artificial intelligence as I mentioned in the
intelligence as I mentioned in the beginning of our
beginning of our this is now the new way of doing
this is now the new way of doing Computing this is the new way of doing
Computing this is the new way of doing software every software engineer every
software every software engineer every engineer every creative artist everybody
engineer every creative artist everybody who uses computers today as a tool will
who uses computers today as a tool will need a AI
need a AI supercomputer and so I just wished I
supercomputer and so I just wished I just wish that djx1 was smaller and
um you know so so um you know imagine ladies and gentlemen
supercomputer and and it's finally called project digits right now and if
called project digits right now and if you have a good name for it uh reach out
you have a good name for it uh reach out to us um uh this here's the amazing
to us um uh this here's the amazing thing this is an AI supercomputer it
thing this is an AI supercomputer it runs the entire Nvidia AI
runs the entire Nvidia AI stack all of nvidia's software runs on
stack all of nvidia's software runs on this dgx Cloud runs on
this dgx Cloud runs on this this
this this sits well somewhere and it's wireless or
sits well somewhere and it's wireless or you know connect it to your computer
you know connect it to your computer it's even a workstation if you like it
it's even a workstation if you like it to be and you could access it you could
to be and you could access it you could you could reach it like a like a cloud
you could reach it like a like a cloud supercomputer and nvidia's AI works on
supercomputer and nvidia's AI works on it and um it's based on a a super secret
it and um it's based on a a super secret chip that we've been working on called
chip that we've been working on called GB 110 the smallest Grace Blackwell that
GB 110 the smallest Grace Blackwell that we make and I have well you know what
we make and I have well you know what let's show let's show everybody insight
isn't it just isn't just it's just so cute and this is the chip that's
cute and this is the chip that's inside it is in it is in
inside it is in it is in production this top secret chip uh we
production this top secret chip uh we did in collaboration the CPU the gray
did in collaboration the CPU the gray CPU was a uh is built for NVIDIA in
CPU was a uh is built for NVIDIA in collaboration with mediatech
collaboration with mediatech uh they're the world's leading s so
uh they're the world's leading s so company and they worked with us to build
company and they worked with us to build this CPU this CPU s so and connect it
this CPU this CPU s so and connect it with chipto chip mvy link to the
with chipto chip mvy link to the Blackwell GPU and uh this little this
Blackwell GPU and uh this little this little thing here is in full production
little thing here is in full production uh we're expecting this computer to uh
uh we're expecting this computer to uh be available uh around May time frame
be available uh around May time frame and so it's coming at you uh it's just
and so it's coming at you uh it's just incredible what we could do and it's
incredible what we could do and it's just I think it's you
really I was trying to figure out do I need more hands or more
need more hands or more pockets all right so so uh imagine this
pockets all right so so uh imagine this is what it looks
is what it looks like you know who doesn't want one of
like you know who doesn't want one of those and if you if you use
those and if you if you use PC Mac you know anything because because
PC Mac you know anything because because uh you know it's it's a cloud platform
uh you know it's it's a cloud platform it's a cloud computing platform that
it's a cloud computing platform that sits on your desk you could also use it
sits on your desk you could also use it as a l Linux workstation if you like uh
as a l Linux workstation if you like uh if you would like to have double
if you would like to have double digits this is what it looks like you
digits this is what it looks like you know and you you connect it you connect
know and you you connect it you connect it together uh uh with connectx and it
it together uh uh with connectx and it has
has nickel GPU direct all of that out of the
nickel GPU direct all of that out of the box it's like a supercomputer our entire
box it's like a supercomputer our entire supercomputing stack uh is available and
supercomputing stack uh is available and so Nvidia Project digits
so Nvidia Project digits [Applause]
[Applause] okay well let me let me let me tell you
okay well let me let me let me tell you what I told you I told you that we are
what I told you I told you that we are in production with three new Blackwells
in production with three new Blackwells not only is the grace Blackwell
not only is the grace Blackwell supercomputers mvlink 72s in production
supercomputers mvlink 72s in production all over the world we now have three new
all over the world we now have three new Blackwell systems in production one
Blackwell systems in production one amazing AI foundational M World
amazing AI foundational M World Foundation model the world's first
Foundation model the world's first physical AI Foundation model is open
physical AI Foundation model is open available to activate the world's
available to activate the world's industries of Robotics and such and
industries of Robotics and such and three and three robotics three robots
three and three robotics three robots working on uh agentic AI uh human or
working on uh agentic AI uh human or robots and self-driving
robots and self-driving cars uh it's been an incredible year I
cars uh it's been an incredible year I want to thank all of you for your
want to thank all of you for your partnership uh thank all of you for
partnership uh thank all of you for coming I made you a short video to
coming I made you a short video to reflect on last year and look forward to
reflect on last year and look forward to the next year play please w
[Music] [Applause]
[Applause] [Music]
have a great C us everybody happy New
everybody happy New Year thank you
Video के उस moment पर जाने के लिए कोई भी text या timestamp click करें
Share करें:
ज्यादातर transcripts 5 सेकंड से कम में तैयार
एक Click में Copy125+ भाषाएंContent Search करेंTimestamps पर जाएं
YouTube URL Paste करें
कोई भी YouTube video link डालें और पूरा transcript पाएं
Transcript निकालें
ज्यादातर transcripts 5 सेकंड से कम में तैयार
हमारा Chrome Extension लें
YouTube छोड़े बिना transcript तुरंत पाएं। हमारा Chrome extension install करें और watch page पर ही किसी भी video का transcript one-click में access करें।
