加速实现影响：借助 NVIDIA on AWS 在云端构建 AI 工厂（由 NVIDIA 赞助）.pdf

《加速实现影响：借助 NVIDIA on AWS 在云端构建 AI 工厂（由 NVIDIA 赞助）.pdf》由会员分享，可在线阅读，更多相关《加速实现影响：借助 NVIDIA on AWS 在云端构建 AI 工厂（由 NVIDIA 赞助）.pdf（14页珍藏版）》请在三个皮匠报告上搜索。

1、 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.A I M 1 1 6-SCharlie HuangSathwik Madhusudhan Jamie CruiseSpeed to Impact:AI Factory in the Cloud

2、with NVIDIA on AWSSr.Product Marketing ManagerCloud AINVIDIAPrincipal Scientist and Research LeadAprielServiceNowBusiness Line DirectorData&AI PlatformSLB 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.AI Market TrendThe NVIDIA-AWS PartnershipNVIDIA DGX Cloud on AWS OverviewPanel

3、 Discussion with ServiceNow and SLBQ&AAgenda 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.AI Adoption is EverywhereA new industrial revolution powered by AI factoriesAI FACTORIESTransportationNatural ResourceHealthcareManufacturingRetailTelcoInternet and ITFinanceSupercomputing

4、RoboticsSmart CitiesEducationTokens 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.AI Scaling Laws Drive Exponential Demand for ComputeTrillions of Tokens/Model100s of Thousands of Post-Trained Models100s of Thousands of Tokens/Query100s of Millions of Queries/Month20T Tokens/Mod

5、el100s of Pre-Trained ModelsPRE-TRAINING SCALING“TEACHING A MODEL THE KNOWLEDGE OF THE INTERNET”POST-TRAINING SCALING“WHERE MODELS GO TO SCHOOL-TEACHING THEM TO THINK”TEST-TIME SCALING“REASONING OR LONG THINKING”2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.NVIDIA DGX CloudNVIDI

6、A HopperAmazon SageMaker AINVIDIA NIM,TRT-LLM,TritonAmazon EMRRAPIDSAmazon EKSNVIDIA NeMo,NIM,TRT-LLM,DynamoNVIDIA AI EnterpriseAWS IoT GreengrassNVIDIA Jetson,IGXNVIDIA Blackwell,Project CeibaAWS HealthOmicsNVIDIA BioNeMoNVIDIA Ada LovelaceAWS BatchNVIDIA Earth-2,Isaac Lab,NeMoAmazon BraketNVIDIA C