Lesson Plan

The Cold Computer

1962 to 1983, Cambridge and Yorktown Heights • Q-CTRL / IBM, 6 May 2026: 3,000-fold speedup on a Fermi-Hubbard simulation using a 120-qubit IBM superconducting quantum processor (Josephson-junction transmon qubits at 15 millikelvin). First practical quantum advantage on an industrial problem. • May 9, 2026

Before using this lesson plan: Read the full story or listen to the 5-minute narration (works as a bell-ringer).

Quick Use (jump to the section you need)

Discussion questions Extension activities

Lesson Plan: The Cold Computer

Subject area

STEM: physics (superconductivity, quantum mechanics), computer architecture, history of technology, materials science.

Recommended audience

High-school physics (Years 11-13 / Grades 10-12)

First-year undergraduate physics, electrical engineering, or computer science

General STEM enrichment for ages 14+

Lesson length

60-75 minutes (one period plus optional follow-up).

Learning objectives

By the end of the lesson, students will be able to:

Describe what a Josephson junction is and why it switches in picoseconds at four kelvin.

Outline the timeline and goals of IBM's 1967-1983 Josephson Computer Technology project.

Identify the technical reason the project was cancelled and explain why silicon won the classical-computing race.

Describe what a transmon qubit is and explain why it is built from a Josephson junction.

Distinguish between a 4 K cryostat and a 15 mK dilution refrigerator and explain why quantum computers run colder than classical superconducting computers ever did.

Explain what the Q-CTRL / IBM May 2026 Fermi-Hubbard result demonstrates about practical quantum advantage.

Key terms

Superconductor, Cooper pair, Josephson effect, Josephson junction

Niobium, aluminium oxide, niobium-aluminium oxide-niobium tri-layer

Cryostat, dewar, liquid helium, dilution refrigerator

Kelvin, millikelvin

Logic gate, memory cell, magnetic flux trapping

Qubit, transmon, superposition, decoherence

Quantum advantage, Fermi-Hubbard model

Error suppression, error correction

Pre-class preparation

A periodic table on the wall.

A photograph of an IBM dilution refrigerator (the gold-plated "wedding cake" cone is iconic and easily found online).

A printed schematic of a Josephson junction cross-section (two superconductors with a thin insulating barrier between them).

Lesson outline

1. Hook (5 min): "the chip switches faster than silicon"

Display a photograph of a 1980-vintage IBM Josephson cryostat. Pose: in 1980, IBM was operating chips that switched in picoseconds (10⁻¹² seconds), thousands of times faster than the best silicon transistors of the time. Three years later, IBM cancelled the entire project. Why?

2. Superconductivity in 30 seconds (5 min)

Below a critical temperature, certain metals lose all electrical resistance.

Discovered 1911 by Heike Kamerlingh Onnes in mercury at 4 kelvin.

The mechanism (the BCS theory of Bardeen, Cooper, Schrieffer) was worked out in 1957: paired electrons (Cooper pairs) move coherently through the lattice without scattering.

For typical low-temperature superconductors, you need liquid helium (boils at 4.2 K) to keep them cold.

High-temperature superconductors (above 30 K) were discovered in 1986; they use cuprate ceramics. Not relevant to this story; the technologies discussed here all use low-temperature superconductors.

3. The Josephson effect (10 min)

1962: Brian Josephson, age 22, a graduate student at the Cavendish Laboratory in Cambridge, predicts that a thin insulator (a few nanometres thick) sandwiched between two superconductors will allow Cooper pairs to tunnel through it without resistance.

The current that flows depends on the phase relationship of the two superconductors. The junction switches between two states in picoseconds.

Confirmed experimentally in 1963.

Brian Josephson awarded the Nobel Prize in Physics in 1973.

Discussion: a switch in picoseconds, dissipating almost no heat, looks like the perfect computing element. What is the catch? (It has to live at 4 K.)

4. IBM's bet, 1967-1983 (15 min)

Juri Matisoo at IBM demonstrates sub-nanosecond Josephson logic in 1966.

Wilhelm Anacker takes over the engineering project in 1969 (formally launched 1967).

Goal: build a complete general-purpose computer running on superconducting Josephson logic. Mainframe-class machine.

By 1980, the project employs over 100 researchers across Yorktown Heights, Zurich, and Tokyo. Total investment by 1983: about $300M in then-money.

Logic works. Memory does not. The Josephson memory cell is a tiny superconducting loop; in dense arrays, the loops trap stray magnetic flux and lose their state.

Meanwhile silicon advances rapidly. The 1971 4004 microprocessor has 2,300 transistors. The 1979 Motorola 68000 has 68,000. Silicon scaling is on the cusp of Moore's Law.

September 1983: IBM cancels the project. Science magazine reports: "Problems with a high-speed memory chip would delay a Josephson junction computer long enough for semiconductors to catch up."

Discussion: was the cancellation the right decision in 1983? Yes. The architecture was not converging on commercial readiness; silicon was. Time and money are finite. Was the architecture doomed forever? No.

5. Forty-three years in the wilderness (5 min)

1983-2000: Josephson computing largely abandoned in the US.

Japan (the MITI superconducting electronics programme) carried the work through the 1990s.

A small cryogenic-electronics community persisted at NIST, in Sweden, and at a handful of universities.

The expertise survived. The infrastructure (fabrication facilities, cryogenic engineering, niobium tri-layer process) survived.

6. Quantum computing arrives (10 min)

1981: Richard Feynman lecture at MIT proposes a quantum computer for simulating quantum systems.

1985: David Deutsch proposes a universal quantum computer.

1994: Peter Shor's algorithm shows quantum computers could factor large integers exponentially faster than any classical algorithm. RSA cryptography panics.

2000s: multiple physical implementations explored (trapped ions, neutral atoms, photons, NMR, superconducting circuits).

Yale, Delft, Saclay groups in the late 1990s and early 2000s show that Josephson junctions in superconducting circuits can be operated as qubits. Different junction geometry, but the same fundamental quantum element IBM had abandoned.

2007: the transmon qubit (Robert Schoelkopf, Michel Devoret, Steven Girvin and colleagues at Yale) makes superconducting qubits robust to charge noise. The transmon is a Josephson junction shunted by a large capacitor.

2014-2019: small superconducting quantum processors built (UCSB → Google, IBM, Rigetti).

2019: Google Sycamore demonstrates quantum supremacy on a contrived random-circuit-sampling task.

7. The May 2026 result (10 min)

6 May 2026: Q-CTRL Inc (Sydney) announces a 3,000-fold speedup on a Fermi-Hubbard model simulation using a 120-qubit IBM superconducting quantum processor.

The Fermi-Hubbard model: a foundational model of interacting electrons on a lattice. Used to study correlated-electron materials, high-temperature superconductors, transition-metal oxides, lithium-ion battery cathode materials. Easy to write down, exponentially hard for classical algorithms in non-trivial parameter regimes.

Performance: 2 minutes on the quantum machine; over 100 hours on the best available classical methods running on a contemporary supercomputer.

The quantum algorithm used 120 qubits and over 10,000 two-qubit gates.

Q-CTRL's contribution: error-suppression software running on top of IBM's hardware. Errors in current quantum hardware accumulate as you run more gates; Q-CTRL's software shapes the control pulses to suppress errors.

Q-CTRL describes this as the first practical quantum advantage on a problem of real industrial value.

Note the platform: the chip is a Josephson-junction-based superconducting processor at 15 millikelvin, fabricated at IBM's Watson Research Center in Yorktown Heights. The same lab IBM had used for the cancelled 1969-1983 project.

8. Wrap and assignment (5 min)

Pose: "Identify a technology being abandoned today that might come back in 40 years for a different application. What property of the abandoned technology would have to be valued differently in the future?"

Optional homework: write a 500-word memo from the perspective of a 1985 IBM researcher who left the cancelled Josephson programme. The memo is dated 2026 and reflects on what the cryostats turned out to be good for.

Discussion questions

Why does a Josephson junction switch in picoseconds with almost no heat dissipation? Use Cooper pairs and tunnelling in your answer.

The IBM Josephson memory cell trapped magnetic flux when packed densely. Why is this fatal for a memory? What strategies might engineers have used to mitigate it (and why were they ultimately insufficient)?

The Yale transmon qubit (2007) uses the same Josephson junction IBM abandoned in 1983. What makes a transmon a quantum two-level system rather than a classical logic gate?

Modern dilution refrigerators run at 15 millikelvin, two thousand times colder than IBM's 1980 cryostats. Why must quantum computers run so much colder than classical superconducting computers? Consider thermal noise and decoherence.

The Q-CTRL/IBM May 2026 result is a 3,000× speedup on a problem of known industrial value. What would it take to convince you that this is "practical quantum advantage" rather than a contrived demonstration?

IBM's 1983 cancellation decision was, in business terms, correct. The same architecture is now central to one of the most important technologies of the 21st century. What does this say about how to evaluate research investments that might pay off only over multi-decade timescales?

Extension activities

Lab demonstration (where facilities allow): demonstrate the Meissner effect (a superconductor expelling a magnetic field) using YBCO and a small magnet immersed in liquid nitrogen. Connect to the broader theme of cryogenic physics.

Reading exercise: read the abstract of Anacker's 1980 IBM Journal review of Josephson Computer Technology and the abstract of the Q-CTRL May 2026 announcement. Identify what each claims.

Calculation: at 4 kelvin, the thermal energy is approximately k_B T = 5.5 × 10⁻²³ J. At 15 millikelvin, it is 2.1 × 10⁻²⁵ J. Compare to the Josephson energy of a typical transmon (~10 GHz, or 6.6 × 10⁻²⁴ J in energy terms) and explain why the transmon must operate well below 1 K.

Writing assignment: write a 1000-word essay on what factors a research-funding agency should consider when deciding whether to abandon a technology programme. Use the IBM Josephson programme as the case study.

Time Rhymes