Summary
In this Bloomberg Technology segment, fund manager Daniel Pilling argues that the AI chip market will remain deeply supply-constrained for years, citing three compounding forces: adoption curves comparable to Zoom during COVID, extremely low current penetration (estimated at just 2-3% of the world’s roughly one billion office workers), and semiconductor industry structure characterized by long manufacturing lead times and oligopolistic pricing power — all converging on a market ill-equipped to meet demand quickly.
Pilling outlines a full-stack investment view that extends well beyond NVIDIA. He holds memory names SK Hynix, Samsung, and Micron (the video’s headline subject); semiconductor capital equipment companies ASML and Lam Research; electricity infrastructure play Bloom Energy; and ARM as a CPU beneficiary of agentic AI workloads. On NVIDIA specifically, he models roughly 60% combined market share across training and inference, applying that to Jensen Huang’s projection of $3-4 trillion in industry-wide CapEx by 2030 to frame the free cash flow opportunity.
On Huawei’s newly announced “logic folding” technique — a transistor signal-routing method intended to sidestep Moore’s Law miniaturization limits — Pilling is measured but skeptical. He notes that chip-stacking concepts have been in development at TSMC for roughly a decade and that heat dissipation at scale remains an unsolved problem. His broader view on China’s semiconductor ambitions: without domestic access to ASML’s extreme ultraviolet lithography, China is likely a decade away from competing at the leading edge (sub-5nm), regardless of architectural breakthroughs.
📺 Source: Bloomberg Technology · Published May 26, 2026
🏷️ Format: Interview
