IT领导者的AI：部署面向未来的IT基础设施

Future-Proof ITInfrastructure On the RightSide of Disruption We’re witnessing the evolution of artificial intelligence (AI)right before our eyes. Getting artificial intelligence (AI) right is a race to the finish line, and enterprises arefeeling pressured to harness AI to build unique value ahead of their competitors. We are only beginning to understand the myriad ways AI will disrupt business, butone thing is clear: to be on the right side of disruption, enterprises must embracechange and leverage it to their advantage. Business outcomes should be at the center of any enterprise’s AI strategy, and IT leadersplay a critical role by building a strong foundation that can handle continuous disruption.Hybrid IT enables enterprises to be agile and scalable, which allows them to grow. IT leaders have long known the value of data, and AI use cases highlight itsimportance – not just for the insights it can extract, but how a strategic approach toits management is directly tied to its ability to create value. This includes its location,proximity to data and users, and the ecosystem in which it’s operating. AI deployments' success depends on the physical infrastructure that provides thehigh-performance cooling, layout, and connectivity AI requires to operate effectively.That’s why companies need purpose-built infrastructure and the right partners to fuelinnovation today and well into the future. We hope the following whitepaper helps as you develop or iterate on your IT infrastructurestrategy to enable AI and other high-performance computer (HPC) workflows. Let’s innovate together. Chris SharpChief Technology Officer, Digital Realty Contents 01 Realize thepromise of thedata economy The key to business differentiation:extracting value from enterprise data. range between $2.6 to $4.4 trillion inproductivity.2Boards worldwide arechallenging enterprise leaders to definean AI strategy that creates value andcompetitive differentiation. We’re in the middle of a data-driven digital transformationthat the World Economic Forumpredicts will boost the globaleconomy by $100 trillion by 2025.1 But what does it take to deploy AI in asustainable, scalable, and future-proof waythat also achieves long-term businessvalue? This is the challenge for IT leaders asthey partner with the business to develop,enable, and iterate towards mature AIstrategies. The data economy is a global digitalecosystem where data is collected,organized, and exchanged by a networkof companies, individuals, and institutionsto create economic value. Enterprises are working with data at scale,and artificial intelligence (AI) and high-performance compute (HPC) applicationsare both a tool to extract insights andcreate value, as well as an acceleratordriving the growth of data overall. This whitepaper examines the ITinfrastructure needs for long-term AIenablement, the factors enterprisesneed to consider to process AI-relateddatasets, and the evaluation criteria forpicking the right data center partners fortheir future-proof AI journey. We’ll cover the distinctions between thesedifferent technologies and the evolution ofAI in more detail shortly, but it’s clear thatthe buzz around generative AI (GenAI), inparticular, gained new steam in 2023. While the use of AI has been growing foryears, driven by early adopters, we are atan acceleration point, especially when itcomes to enterprise investment: globalspending on AI will reach $300 billion by2026, according to research firm IDC.3 According to McKinsey & Company,GenAI could add trillions of dollarsto the global economy — estimates GenAI and the ability to deliver contentmight dominate headlines, but it’s only asubset of a larger field. These strategies and applications canlead to cost savings, improvements inproductivity, and a leaner, more agileenterprise. Strategic goals for enterprise AI 2. Business transformation Why are enterprises investing in AI? Thereare two primary goals, although they arenot mutually exclusive. Disruptors and enterprise leaderslooking to transform industries willleverage AI as a unique businessoffering, monetizing AI-related servicesas service providers and creating newAI-powered products and capabilities. 1. Operational efficiency AI can streamline and enhancebusiness-as-usual operations by takingon repetitive or labor-intensive tasks atscale. Operational efficiency is often bestachieved by assisting and augmentingemployees, rather than replacing them. These strategies and applications ofAI can lead to transformative growth,especially if the infrastructure backsthem up to support that growth. AI-as-a-Service (AIaaS) solutions forenterprises are now widely available,ranging from off-the-shelf applicationsto highly customizable or configurablesolutions. Enterprises may also invest inbespoke internally-developed AI tools toimprove operational efficiency, customerservice, and advanced analytics. Regardless of the goal, enterprisesshould adopt a mindful approach f