for AI Infrastructure Key Considerations Why AI? A Digitized World Requires Data – and the Infrastructure to Handle It Artificial intelligence (AI) is transforming how consumers and enterprises work, learn, and play. Winning organizations will be those that use AI – as well as machine learning (ML) and deep learning (DL) – to: Data is core for AI/ML/DL use cases, and unstructured data is driving data growth in AI initiatives: Handling ever-growing data sets is critical to accelerate AI initiatives. What’s needed is an AI infrastructure solution that balances compute and storage, memory, and networking capabilities. Cost is the main reason for moving accelerated workloads from the cloud to on-premise, followed by data security and lack of performance in the cloud across all accelerator technologies. Traditional datacenter architecture can’t run AI workloads efficiently. Enter acceleration technologies for AI. An IDC survey on customer perceptions of acceleration technologies indicates: Many companies are jumping in with both feet: But few are in full production: are currently exploring AI initiatives Deliver meaningful insights and value-added predictions of organizations had AI models in production were in the proof-of-concept stage are experimenting with AI Uncover new business opportunities and revenue sources Improve processes, products, or services across industries and use cases Enable digital transformation Source: IDC, Q3 2019 Market View: Infrastructure for AI Source: IDC’s Artificial Intelligence Global Adoption Trends and Strategies survey Source: IDC's Worldwide File and Object-based Storage Forecast 2018 Source: How End Users in 10 Industries Are Perceiving Importance and Satisfaction with Server Acceleration: An IDC survey of 200 respondents on Customer Perceptions regarding Acceleration Technologies 2018 – Part 1, n=200 49% 18% 15% 16% By 2022, 881 exabytes of storage capacity will be shipped in support of unstructured data Do not use GPUs Deploy in cloud Of users deploy GPUs on-premises across all industries On-premise GPU deployment is expected to increase On-premises vs. Cloud from to 50.8% today 54.2% in 12 months Industries and Criteria for Infrastructure Choice Business needs to obtain AI insights at speed, especially in certain industries. Acceleration for AI/DL occurs most often in: An IDC survey of technology buyers considering compute and storage infrastructure found: The top three industry criteria in choosing an AI infrastructure: 1. 1. 2. 2. 3. 3. Finance Life Sciences Broadcast/Telecom and Healthcare Ships with full stack of frameworks, SDKs, tool kits Provides AI/HPC-related support/consulting Offers specific server form factor Source: How End Users in 10 Industries Are Perceiving Importance and Satisfaction with Server Acceleration: An IDC survey of 200 respondents on Customer Perceptions regarding Acceleration Technologies 2018 – Part 1, n=200 66% 82% 60% Plan to use or currently use compute and storage in a corporate datacenter or co-location Would purchase new infrastructure including servers and storage for thier AI initiatives Have purchased SSD-based all flash arrays for AI training and inferencing Source: IDC's Enterprise Infrastructure Market Pulse 2019 IDC Guidance The ideal AI infrastructure is an optimal combination of compute and storage. Other key considerations for companies planning to jump into AI include: Ease of use in terms of deployment, management, and scale Partnerships that include ecosystem of software and frameworks Experience as technology supplier Support from supplier or integrator for compute, storage, and networking Flexibility to support various requirements across different AI projects Portfolio benefits of supporting integrated products for AI from a single technology supplier To learn more about the importance of choosing AI infrastructure, download the IDC Technology Spotlight, “Unleashing the Power of AI Initiatives with the Right Infrastructure,” sponsored by Dell EMC All IDC research is © 2019 by IDC. All rights reserved. All IDC materials are licensed with IDC’s permission and in no way does the use or publication of IDC research indicate IDC’s endorsement of Dell’s products or strategies. Document #US45578019 An IDC Infographic, sponsored by Dell EMC 53% 34% 13%