Descriptions:
In this Bloomberg Technology interview timed to Nvidia’s Q1 2026 earnings, Baillie Gifford portfolio manager Paulina offers a detailed long-term investment perspective on the AI chip supply chain. On China, her base case for Nvidia models zero revenue — not because of U.S. export controls, but because the Chinese government is blocking domestic companies from purchasing H100s that have already been approved for export. She argues that even if China eventually builds a domestic chip industry, two factors protect Nvidia’s lead for at least five years: existing Chinese high-performance chips are still a fraction of Nvidia’s performance, and building a scaled semiconductor supply chain is structurally harder than industries like EVs or batteries because it requires globally coordinated, highly specialized manufacturing.
The conversation turns to memory as the current AI infrastructure bottleneck. Paulina contrasts SK Hynix and TSMC as investment propositions. SK Hynix benefits from the rise of HBM and custom HBM, which creates customer lock-in that commoditized memory historically lacked, and from long-term purchase agreements that reduce the boom-bust cyclicality of the memory market. TSMC, Baillie Gifford’s largest position, she views as categorically different — a company with demonstrated dominance over an increasingly complex supply chain where competitors fall further behind with each new process node.
Zooming out, Paulina frames the broader AI chip story as ‘designed in America, manufactured internationally,’ pointing to TSMC (Taiwan), SK Hynix (Korea), and ASML (Netherlands) as the key non-U.S. nodes that investors outside the U.S. market are beginning to focus on more seriously.
📺 Source: Bloomberg Technology · Published May 20, 2026
🏷️ Format: Interview
