NVIDIA and AMD reshape AI infrastructure

Nvidia and AMD used back-to-back announcements this week to show how the AI build-out is moving beyond a race for faster chips and into a contest over manufacturing capacity, packaging, regional footprints and full-system deployments. Nvidia on May 20 reported another revenue beat and described the datacenter boom as the rise of “AI factories,” while AMD said on May 21 it would invest more than $10 billion across Taiwan’s AI sector. The disclosures pointed to the same pressure point: the companies selling AI compute are also tying themselves more tightly to the foundries, assembly networks and cloud customers needed to turn chips into deployed infrastructure. Nvidia also said its Vera CPU systems are already reaching early customers including Anthropic, OpenAI and Oracle Cloud Infrastructure. ### Why are Nvidia and AMD both talking about infrastructure, not just chips? Nvidia on May 20 framed the current spending wave as a “global explosion of datacenters” and described those facilities as “AI factories,” according to reporting on the company’s results. That language put the emphasis on the full build-out of computing sites rather than on standalone accelerators. (cnbc.com) AMD on May 21 made a similar point from the supply side. The company said it would invest more than $10 billion in Taiwan’s AI sector to deepen partnerships and expand its ability to build and assemble advanced AI chips, according to Reuters reporting carried by The Standard. ### Why does Taiwan sit at the center of this week’s announcements? (cnbc.com) AMD said its next-generation “Venice” chips have entered production at TSMC’s 2-nanometer process in Taiwan, linking its next server platform directly to the island’s most advanced manufacturing line. The company tied that production milestone to a broader investment push across Taiwan’s AI ecosystem. (thestandard.com.hk) TSMC’s role matters because leading-edge process technology and advanced packaging remain concentrated in a small number of facilities and partners. AMD’s announcement made Taiwan not just a manufacturing base for one product generation, but the center of a larger partnership and capacity plan. (thestandard.com.hk) ### What is Nvidia adding beyond GPUs? Nvidia has started shipping Vera CPU systems to early customers including Anthropic, OpenAI and Oracle Cloud Infrastructure, according to Yahoo Finance and related market coverage. The rollout extends Nvidia’s position in AI systems beyond GPUs into CPUs designed for large-scale AI training and inference workloads. (thestandard.com.hk) Nvidia introduced Vera in March as a processor “purpose-built” for agentic AI and reinforcement learning, and said manufacturing partners included Dell Technologies, HPE, Lenovo, Supermicro and several large Asian hardware makers. That partner list showed how the company is trying to anchor a broader hardware stack around its own architecture. ### Why is Nvidia opening a research hub in Singapore now? (finance.yahoo.com) Nvidia said on May 20 it will launch a new research center in Singapore, its first such hub in the city-state and its second in Asia-Pacific. CNBC reported the lab will support research tied to “physical AI,” alongside broader measures announced by Singapore to expand its AI plans. Singapore gives Nvidia another regional foothold as the company broadens its presence outside the United States and Taiwan-centered manufacturing networks. (finance.yahoo.com) The announcement came as Nvidia Chief Executive Jensen Huang also said the company had “largely conceded” China’s advanced AI chip market to Huawei, underscoring the regional pressures around market access and research expansion. (cnbc.com) ### What should readers watch next? Oracle Cloud Infrastructure plans hundreds of thousands of Vera deployments starting in 2026, according to market coverage of Nvidia’s rollout. That gives investors and customers a concrete next checkpoint for whether Nvidia’s push beyond GPUs gains volume in commercial systems. AMD’s next markers will be the scale and timing of its Taiwan investment and the ramp of Venice production at TSMC’s 2-nanometer node. (cnbc.com) Nvidia’s next visible test will come through additional customer shipments, regional research expansion and future earnings updates tied to its “AI factory” build-out. (thestandard.com.hk) (finance.yahoo.com)