Summary
Physicist and science communicator Sabine Hossenfelder examines whether quantum computing can meet the U.S. Department of Energy’s goal of deploying a fault-tolerant, scientifically relevant quantum computer by 2028, a target tied to an executive order from President Trump. She argues the timeline is unrealistic and walks through a new paper that makes the first attempt to calculate the physical resources a large error-corrected machine would need.
The analysis focuses on superconducting qubits, the most advanced approach today. A useful machine would need roughly 4 million qubits, but each cooling apparatus holds only about 10,000 at best. That puts the system at more than 800 square meters before readout and control equipment are added, roughly the footprint of a supermarket. She also notes that PsiQuantum’s photonic approach has yet to show a public demo.
The video then turns to use cases. It covers the Wellcome Leap $5 million prize for a healthcare problem where quantum beats classical computing, which went unclaimed, and a Los Alamos study finding that conventional algorithms match or beat quantum neural networks on every benchmark tested, up to more than 1,000 logical qubits. For anyone tracking compute hardware, quantum machine learning claims, and AI-adjacent investment hype, it offers a skeptical counterpoint.
📺 Source: Sabine Hossenfelder · Published November 03, 2026
🏷️ Format: Opinion Editorial
