Americas Technology: Hardware: Expert Network Series: Former Director of AI Transformation at Microsoft on AI infrastructure demand

研报英文原文证据摘录

Americas Technology: Hardware: Expert Network Series: Former Director of AI Transformation at Microsoft on AI infrastructure demand

Equity Research

10 July 2026 | 2:30PM EDT

Americas Technology: Hardware: Expert Network Series: Former

Director of AI Transformation at Microsoft on AI infrastructure demand

We hosted an investor webinar with William Fong, former Director of AI Katherine Murphy

+1(212)902-1151 |

Transformation at Microsoft on July 10th, 2026 as part of our Expert Network katherine.a.murphy@gs.com

Goldman Sachs & Co. LLC

Series to discuss AI infrastructure demand trends.

Michael Ng, CFA

+1(212)902-8618 | michael.ng@gs.com

Bottom line: (1) Agentic AI is driving organizations to invest in on-premise and edge Goldman Sachs & Co. LLC

infrastructure because of better token cost, latency, and governance/data

sovereignty relative to the cloud; this should be meaningful tailwind for enterprise

compute vendors like DELL and HPE; (2) Mr. Fong does not believe that AI

infrastructure is at risk of being overbuilt, largely due to the massive demand from

agentic AI as well as ongoing demand for net-new training infrastructure; and (3)

Organizations are finding budget for hardware investments through infrastructure

performance efficiencies (e.g., power cost savings from new generation CPU servers)

and reduced headcount.

Key takeaways

Agentic AI is driving workloads on-prem because of better cost, latency, and

governance. To optimize across these constraints, Mr. Fong has observed

organizations deploying a hybrid compute strategy where they are increasingly

running individual AI agents locally on-premise to achieve millisecond-level latency

and maintain strict data privacy, while utilizing the cloud for the complex multi-agent

orchestration layers. Mr.