#1NVIDIA® Networks for Al Gilad Shainer - SVP, Networking#2FORWARD-LOOKING STATEMENTS Except for the historical information contained herein, certain matters in this presentation including, but not limited to, statements as to: our markets and market opportunity; our growth and growth drivers; and the benefits, impact, and performance of our products and technologies are forward-looking statements. These forward-looking statements and any other forward-looking statements that go beyond historical facts that are made in this presentation are subject to risks and uncertainties that may cause actual results to differ materially. Important factors that could cause actual results to differ materially include: global economic conditions; our reliance on third parties to manufacture, assemble, package and test our products; the impact of technological development and competition; development of new products and technologies or enhancements to our existing product and technologies; market acceptance of our products or our partners' products; design, manufacturing or software defects; changes in consumer preferences and demands; changes in industry standards and interfaces; unexpected loss of performance of our products or technologies when integrated into systems and other factors. NVIDIA has based these forward-looking statements largely on its current expectations and projections about future events and trends that it believes may affect its financial condition, results of operations, business strategy, short-term and long-term business operations and objectives, and financial needs. These forward-looking statements are subject to a number of risks and uncertainties, and you should not rely upon the forward-looking statements as predictions of future events. The future events and trends discussed in this presentation may not occur and actual results could differ materially and adversely from those anticipated or implied in the forward-looking statements. Although NVIDIA believes that the expectations reflected in the forward-looking statements are reasonable, the company cannot guarantee that future results, levels of activity, performance, achievements or events and circumstances reflected in the forward-looking statements will occur. Except as required by law, NVIDIA disclaims any obligation to update these forward-looking statements to reflect future events or circumstances. For a complete discussion of factors that could materially affect our financial results and operations, please refer to the reports we file from time to time with the SEC, including our most recent Annual Report on Form 10-K, Quarterly Reports on Form 10-Q, and Current Reports on Form 8-K. Copies of reports we file with the SEC are posted on our website and are available from NVIDIA without charge. nVIDIA |#3APPLICATION FRAMEWORKS PLATFORM ACCELERATION LIBRARIES SYSTEM SOFTWARE HARDWARE MODULUS ΜΟΝΑΙ RTX CUDA-X CUDA Magnum 10 RTX MAXINE GPU DGX 100 CPU NEMO DOCA HGX NVIDIA Al DPU AVATAR EGX NIC DRIVE Base Command OVX ISAAC NVIDIA OMNIVERSE AGX SWITCH ER METROPOLIS Forge MLNX SOC HOLOSCAN nVIDIA#4AI CLOUD 10 CLOUD AI FACTORY 100 1k The Data Center is The Computer The network defines the data center NVIDIA Spectrum-X Al Ethernet Fabric Traditional Ethernet NVLink+ InfiniBand 10k # of GPU in Cluster 100k InfiniBand 1M+ Cloud Generative Al Cloud Multi-tenant Variety of workloads including larger scale Generative Al Traditional ethernet network for North-South traffic NVIDIA Spectrum-X ethernet for Al fabric (East-West) ● ● Multi-tenant Variety of small-scale workloads Traditional ethernet network can suffice Al Factories ● Single or few users Extremely large Al models NVIDIA NVLink and InfiniBand gold standard for Al fabric nVIDIA |#5Software Acceleration Hardware Acceleration The NVIDIA AI Network Advantage Hardware & software accelerated in-network computing NCCL - NVIDIA Collective Communication Library The SDK library for Al communications - connects the GPUs and the network for the Al network operations. SHARP - NVIDIA Scalable Hierarchical Aggregation and Reduction Protocol Technology SHARP is part of the InfiniBand and NVLink switch ASICs. It enables the network to perform data reduction operations, an important element of Al workloads. This decreases the amount of data traversing the network and dramatically reduces collective operations time. 8 NCCL Performance With vs Without SHARP (In-network Computing) 16 32 64 1.7X HIGHER InfiniBand (with SHARP) Best Theoretical Ethernet performance 128 256 512 1024 2048 4096 8192 16384 Message Size (MiB) nVIDIA |#6In the Cloud - The Two Worlds of Ethernet East-West for distributed and disaggregated processing | North-South for user-to-cloud communications Control / User Access Network (North-South) Traditional Ethernet Loosely Coupled Applications TCP (Low Bandwidth Flows and Utilization) High Jitter Tolerance Heterogeneous Traffic Average Multi-Pathing Al Fabric (East-West) Spectrum-X Distributed Tightly-Coupled Processing ROCE (High Bandwidth Flows and Utilization) Low Jitter Tolerance (Long Tail Kills Performance) Bursty Network Capacity Predictable Performance North - South XX XX XX 9-63 XX ← XX ← XX Data Center East West XX XX NVIDIA#7NVIDIA Spectrum-X Ethernet Al Platform The ethernet fabric for Al Spectrum-4 Switch 100 billion transistors, TSMC 4N 51.2T bandwidth, 100G SerDes 64 X 800G Ports, 128 X 400G ports 8K GPUs in 2-level, 500K in 3 level ● BlueField-3 DPU Architectural Advantages over Traditional Ethernet... End-to-end optimized (DPU to Switch) Adaptive routing over lossless ethernet Congestion control for multi-tenant traffic isolation Spectrum-4 Switch BlueField-3 DPU SONIC Cumulus NetQ SAI/SPSDK DOCA New class of ethernet, built from the ground-up for Generative Al cloud workloads DOCA Services Air Magnum 10/ NCCL ...Deliver 1.6X Better Al Fabric Performance vs Traditional Ethernet 95% effective bandwidth vs 60% for traditional ethernet Maintains NCCL performance in noisy environment Secured virtualized network nVIDIA#8Quantum-2 Switch 57 billion transistors, TSMC 7N 64 X 400G ports, 100G SerDes 64 In-Network Computing Al Engines 65K GPUs in 3-level, 2M in 4-level NVIDIA Quantum InfiniBand Platform The gold standard for large scale Al BlueField-3 DPU InfiniBand's Inherent Architectural Differentiators... Architected for large scale, end-to-end optimized • In-network computing with SHARP Pure software-defined network Quantum-2 Switch BlueField-3 DPU NVOS DOCA UFM DOCA Services Hardened over 20+ years for the most extreme-scale workloads in the world's largest supercomputers Air Magnum 10 / NCCL ...Deliver >2X Al Fabric Performance over Traditional Ethernet Limitless GPU scaling (3-level 65K GPUs, 4-level 2M GPUs) 1.7X higher NCCL AllReduce bandwidth performance Nearly 100% effective bandwidth at scale Extremely low latency, 10x better at scale and under load NVIDIA |#9The Network "Pays for Itself"! Performance gains far outweigh the entire cost of the network NDR 400G InfiniBand (64-port switches) 3 switch level network topology (up to 65K nodes) 4 switch level network topology (up to 2M nodes) InfiniBand offers the most scalability... Relative Al Fabric Performance (NCCL Allreduce) Traditional Ethernet 1.6X NVIDIA Spectrum-X Ethernet >2X NVIDIA InfiniBand and the fastest performance nVIDIA |#10Strong Growth and Large Opportunity Data Center Networking Growth Turbocharged by Al Revenue more than doubled since Mellanox acquisition NVIDIA InfiniBand revenue more than tripled Spectrum-X to boost cloud Al ethernet market Q1 CY20 Q1 CY23* InfiniBand ■ Ethernet & Other NVIDIA Networking Revenue *Q1 CY23 refers to Q1 of NVIDIA'S FY24, ending April 30, 2023. The Mellanox acquisition closed on April 27, 2020. Data Processing Units - In every server Data Center Switching - InfiniBand & ethernet High speed optical modules & cables $60 BILLION Networking Market Opportunity nVIDIA |#11NVIDIA®