YouTube Transcript:
NVIDIA CEO Jensen Huang Keynote at CES 2025

Skip watching entire videos - get the full transcript, search for keywords, and copy with one click.

Video Transcript

Video Summary

Summary

Core Theme

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

Click to expand

Click to explore the full interactive mind map • Zoom, pan, and navigate

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]

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

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

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

Click on any text or timestamp to jump to that moment in the video

Most transcripts ready in under 5 seconds

One-Click Copy125+ LanguagesSearch ContentJump to Timestamps

Paste YouTube URL

Enter any YouTube video link to get the full transcript

Most transcripts ready in under 5 seconds

Get Our Chrome Extension

Get transcripts instantly without leaving YouTube. Install our Chrome extension for one-click access to any video's transcript directly on the watch page.

Add to Chrome — Free

Works with YouTube, Coursera, Udemy and more educational platforms

Get Instant Transcripts: Just Edit the Domain in Your Address Bar!

YouTube

←

→

↻

https://www.youtube.com/watch?v=UF8uR6Z6KLc

YoutubeToText

←

→

↻

https://youtubetotext.net/watch?v=UF8uR6Z6KLc

YouTube TranscriptPreparing your results…

YouTube Transcript:NVIDIA CEO Jensen Huang Keynote at CES 2025

Video Transcript

Summary

Core Theme

Paste YouTube URL

Transcript Extraction Form

Get Our Chrome Extension

Get Instant Transcripts: Just Edit the Domain in Your Address Bar!

YouTube Transcript:
NVIDIA CEO Jensen Huang Keynote at CES 2025