NVIDIA : 2024 Annual Report

2024

NVIDIA Corporation Annual Review

Notice of Annual Meeting

Proxy Statement

Form 10-K

"The sum of all that NVIDIA's doing will indeed create the next industrial revolution"

CNBC

Accelerated computing is sustainable

computing. Every data center in the world needs to be accelerated to reclaim power, achieve sustainability, and realize net-zero emissions. Accelerated data centers could save an incredible 19 terawatt-hours of electricity annually if run on GPU and DPU accelerators vs CPUs. That's about the same energy as a year's worth of trips by 2.9 million passenger cars.

The efficiency of accelerated computing paved the way for generative AI. The most critical computing platform of our generation, generative AI will reshape the world's largest industries and create an entirely new one.

NVIDIA, the pioneer of accelerated computing, is the driving force of this new era.

HGX B100	NVLINK Switch	GB200 Superchip
		Compute Node

"They basically have

1. comprehensive solution from the chip all the way to data centers at this point"

CIO

Accelerated computing starts with the most advanced processors and ends with AI factories.

From chip architecture to advanced networking to acceleration libraries, NVIDIA builds the entire computing system at data-center scale. Then, we disaggregate everything and reintegrate it into the world's computing fabric so that industries can leverage the parts and systems they need.

In the future, almost all of our experiences will be generative. Blackwell-the world's most powerful AI platform-istailor-made for the generative AI revolution.

Quantum X800 Switch	Spectrum X800 Switch
ConnectX-8 SuperNIC	BlueField-3 SuperNIC

"Continually optimized software remains NVIDIA's ace in the hole"

Forbes

Accelerated computing requires full-stack

software. NVIDIA's acceleration stacks optimize workloads on a massive scale, integrating thousands of nodes while treating network and storage as integral components.

This year, we rolled out TensorRT-LLM and NVIDIA Inference Microservices™ (NIM). TensorRT-LLM is an open-source software library that enables customers to more than double the inference performance of their GPUs. NIM are a new way to package and deliver AI software. This curated selection of microservices adds a new layer

to NVIDIA's full-stack computing platform- connecting the AI ecosystem of model developers, platform providers, and enterprises with a standardized path to run custom AI models.

Industry Standard APIs

Text, Speech, Image,

Video, 3D, Biology

Triton Inference Server

cuDF, CV-CUDA, DALI, NCCL, Post Processing Decoder

Cloud Native Stack

GPU Operator, Network Operator

Enterprise Management

GPU Health Check, Identity, Metrics, Monitoring, Secrets Management

Kubernetes

TensorRT LLM and Triton

cuBLAS , cuDNN, In-Flight Batching, Memory Optimization, FP8 Quantization

Optimized Model

Single GPU, Multi-GPU,Multi-Node

Customization Cache

P-Tuning, LORA, Model Weights

NVIDIA CUDA

100's of Millions of CUDA GPUs Installed Base

PERFORMANCE, ECOSYSTEM, REACH

		GENOMICS,			AV,
DATA	CAD,	WEATHER	6G,	ROBOTICS,	GENERATIVE
DRUG
PROCESSING	CAE, SDA	SIMULATION	QUANTUM	INDUSTRIAL	AI
DISCOVERY
				DIGITAL
					TWINS
DSL	DSL	DSL	DSL	DSL	DSL	DSL

CUDA-X LIBRARIES

SUPERCOMPUTING SYSTEMS AND SOFTWARE

APPS

GPU

CPU

NIC/DPU

SWITCH

ONE ARCHITECTURE-CUDA

DATACENTERS	CLOUD	EDGE
HGX	DGX CLOUD	AGX
MGX	OV CLOUD	IGX

DEMAND

"NVIDIA's got great chips, and more importantly, they have an incredible ecosystem"

The New York Times

NVIDIA's accelerated computing ecosystem is bringing AI to every enterprise. The NVIDIA

ecosystem spans nearly 5 million developers and 40,000 companies. More than 1,600

generative AI companies are building on

INSTALLED BASENVIDIA. CUDA®, our parallel computing model launched in 2006, offers developers more than 300 libraries, 600 AI models, numerous SDKs, and 3,500 GPU-accelerated applications. CUDA has more than 48 million downloads.

"NVIDIA's prescription for the future: transforming healthcare with AI"

Forbes

NVIDIA AI is powering the next era of drug discovery and advances in life sciences. NVIDIA

Clara™, our suite of computing platforms, software, and services for healthcare and life sciences, and NVIDIA BioNeMo™, our platform for state-of-the-art generative AI models for drug discovery, are turbocharging breakthroughs.

Genentech is tapping NVIDIA to use generative AI to discover and develop new therapeutics and deliver treatments to patients more efficiently. Recursion Pharmaceuticals is the first NVIDIA partner to offer an AI model through BioNeMo cloud APIs. And Amgen is building AI models trained to analyze one of the world's most extensive human datasets on an NVIDIA DGX SuperPOD™.