Wafer-Scale Vision and Yield Challenges

Episode overview
John Martinis, Nobel laureate and former head of Google’s quantum hardware effort, joins Sebastian Hassinger on The New Quantum Era to trace the arc of superconducting quantum circuits—from the first demonstrations of macroscopic quantum tunneling in the 1980s to today’s push for wafer-scale, manufacturable qubit processors. The episode weaves together the physics of “synthetic atoms” built from Josephson junctions, the engineering mindset needed to turn them into reliable computers, and what it will take for fabrication to unlock true large-scale quantum systems.

Guest bio
John M. Martinis is a physicist whose experiments on superconducting circuits with John Clarke and Michel Devoret at UC Berkeley established that a macroscopic electrical circuit can exhibit quantum tunneling and discrete energy levels, work recognized by the 2025 Nobel Prize in Physics “for the discovery of macroscopic quantum mechanical tunnelling and energy quantisation in an electric circuit.” He went on to lead the superconducting quantum computing effort at Google, where his team demonstrated large-scale, programmable transmon-based processors, and now heads Qolab (also referred to in the episode as CoLab), a startup focused on advanced fabrication and wafer-scale integration of superconducting qubits.

Martinis’s career sits at the intersection of precision instrumentation and systems engineering, drawing on a scientific “family tree” that runs from Cambridge through John Clarke’s group at Berkeley, with strong theoretical influence from Michel Devoret and deep exposure to ion-trap work by Dave Wineland and Chris Monroe at NIST. Today his work emphasizes solving the hardest fabrication and wiring challenges—pursuing high-yield, monolithic, wafer-scale quantum processors that can ultimately host tens of thousands of reproducible qubits on a single 300 mm wafer.

Key topics

• Macroscopic quantum tunneling on a chip: How Clarke, Devoret, and Martinis used a current-biased Josephson junction to show that a macroscopic circuit variable obeys quantum mechanics, with microwave control revealing discrete energy levels and tunneling between states—laying the groundwork for superconducting qubits. The episode connects this early work directly to the Nobel committee’s citation and to today’s use of Josephson circuits as “synthetic atoms” for quantum computing.
• From DC devices to microwave qubits: Why early Josephson devices were treated as low-frequency, DC elements, and how failed experiments pushed Martinis and collaborators to re-engineer their setups with careful microwave filtering, impedance control, and dilution refrigerators—turning noisy circuits into clean, quantized systems suitable for qubits. This shift to microwave control and readout becomes the through-line from macroscopic tunneling experiments to modern transmon qubits and multi-qubit gates.
• Synthetic atoms vs natural atoms: The contrast between macroscopic “synthetic atoms” built from capacitors, inductors, and Josephson junctions and natural atomic systems used in ion-trap and neutral-atom experiments by groups such as Wineland and Monroe at NIST, where single-atom control made the quantum nature more obvious. The conversation highlights how both approaches converged on single-particle control, but with very different technological paths and community cultures.
• Ten-year learning curve for devices: How roughly a decade of experiments on quantum noise, energy levels, and escape rates in superconducting devices built confidence that these circuits were “clean enough” to support serious qubit experiments, just as early demonstrations such as Yasunobu Nakamura’s single-Cooper-pair box showed clear two-level behavior. This foundational work set the stage for the modern era of superconducting quantum computing across academia and industry.
• Surface code and systems thinking: Why Martinis immersed himself in the surface code, co-authoring a widely cited tutorial-style paper “Surface codes: Towards practical large-scale quantum computation” (Austin G. Fowler, Matteo Mariantoni, John M. Martinis, Andrew N. Cleland, Phys. Rev. A 86, 032324, 2012; arXiv:1208.0928), to translate error-correction theory into something experimentalists could build. He describes this as a turning point that reframed his work at UC Santa Barbara and Google around full-system design rather than isolated device physics.
• Fabrication as the new frontier: Martinis argues that the physics of decent transmon-style qubits is now well understood and that the real bottleneck is industrial-grade fabrication and wiring, not inventing ever more qubit variants. His company’s roadmap targets wafer-scale integration—e.g., ~100-qubit test chips scaling toward ~20,000 qubits on a 300 mm wafer—with a focus on yield, junction reproducibility, and integrated escape wiring rather than current approaches that tile many 100-qubit dies into larger systems.
• From lab racks of cables to true integrated circuits: The episode contrasts today’s dilution-refrigerator setups—dominated by bulky wiring and discrete microwave components—with the vision of a highly integrated superconducting “IC” where most of that wiring is brought on-chip. Martinis likens the current state to pre-IC TTL logic full of hand-wired boards and sees monolithic quantum chips as the necessary analog of CMOS integration for classical computing.
• Venture timelines vs physics timelines: A candid discussion of the mismatch between typical three-to-five-year venture capital expectations and the multi-decade arc of foundational technologies like CMOS and, now, quantum computing. Martinis suggests that the most transformative work—such as radically improved junction fabrication—looks slow and uncompetitive in the short term but can yield step-change advantages once it matures.
• Physics vs systems-engineering mindsets: How Martinis’s “instrumentation family tree” and exposure to both American “build first, then understand” and French “analyze first, then build” traditions shaped his approach, and how system engineering often pushes him to challenge ideas that don’t scale. He frames this dual mindset as both a superpower and a source of tension when working in large organizations used to more incremental science-driven projects.
• Collaboration, competition, and pre-competitive science: Reflections on the early years when groups at Berkeley, Saclay, UCSB, NIST, and elsewhere shared results openly, pushing the field forward without cut-throat scooping, before activity moved into more corporate settings around 2010. Martinis emphasizes that many of the hardest scaling problems—especially in materials and fabrication—would benefit from deeper cross-organization collaboration, even as current business constraints limit what can be shared.

Papers and research discussed

• “Energy-Level Quantization in the Zero-Voltage State of a Current-Biased Josephson Junction” – John M. Martinis, Michel H. Devoret, John Clarke, Physical Review Letters 55, 1543 (1985). First clear observation of quantized energy levels and macroscopic quantum tunneling in a Josephson circuit, forming a core part of the work recognized by the 2025 Nobel Prize in Physics. Link: https://link.aps.org/doi/10.1103/PhysRevLett.55.1543
• “Quantum Mechanics of a Macroscopic Variable: The Phase Difference of a Josephson Junction” – J. Clarke et al., Science 239, 992 (1988). Further development of macroscopic quantum tunneling and wave-packet dynamics in current-biased Josephson junctions, demonstrating that a circuit-scale degree of freedom behaves as a quantum variable. Link (PDF via Cleland group):