Megaquop

What Is a Megaquop?

A megaquop is a quantum computer that can run about one million quantum operations before an uncorrected error is likely to spoil the answer. The term comes from physicist John Preskill, who used it to name the stage between today's noisy machines and the large fault-tolerant computers that the most demanding algorithms will need.

The unit counts operations, not qubits. A "quop" is one quantum operation, such as a gate or a measurement: a megaquop is \(10^6\) of them, a gigaquop \(10^9\), a teraquop \(10^{12}\). A megaquop machine is described by how long a computation it can complete reliably, not by how many qubits it holds.

Where the Term Comes From: Beyond NISQ

In 2018, Preskill introduced NISQ (noisy intermediate-scale quantum) in "Quantum Computing in the NISQ Era and Beyond" (Quantum, 2018). NISQ devices have enough qubits to be hard to simulate classically but no error correction, so noise limits them to circuits of roughly a thousand operations.

In 2025 Preskill returned to the question in "Beyond NISQ: The Megaquop Machine" (ACM Transactions on Quantum Computing, 2025). He proposed the megaquop as a concrete milestone: a machine with errors near \(10^{-6}\) per operation, reached through quantum error correction rather than by improving physical devices alone. It gives the work beyond NISQ a target to measure progress against.

The Arithmetic Behind a Million Operations

Suppose each operation fails independently with probability \(p\). The chance that \(N\) operations all succeed is approximately

\[ P_{\text{success}} \approx (1-p)^N \approx e^{-pN} \]

For \(N = 10^6\), the product \(pN\) stays near or below one only if \(p\) is about \(10^{-6}\) or lower. Physical two-qubit gates today have error rates of roughly \(10^{-3}\) to \(10^{-2}\), so hardware alone falls short by several orders of magnitude.

Error correction closes the gap. Many physical qubits encode one logical qubit, and when physical errors are below the code's threshold, the logical error rate falls as the code grows. A megaquop machine is one where that suppression reaches about \(10^{-6}\) across many logical qubits and a full set of gates.

One subtlety: operations do not all cost the same. Non-Clifford gates are the expensive ones, typically needing magic state distillation or related methods, so "a million operations" depends on what is counted.

Megaquop, Gigaquop and What Useful Work Needs

ScaleOperationsStatus
NISQabout \(10^3\)Available today; noise limits depth
Megaquop\(10^6\)Target for early error-corrected machines
Gigaquop\(10^9\)Needed for larger algorithms
Teraquop\(10^{12}\)Further out

Reaching a megaquop does not put every useful problem in range. Resource estimates for factoring a 2048-bit RSA integer run to billions of Toffoli gates (Gidney and Ekerå, 2021), well beyond a megaquop. Whether a megaquop system delivers quantum utility depends on the specific problem and is settled by experiment, not by the label.

The megaquop is better read as a stepping stone toward fault-tolerant computing at larger scale.

How QuEra's Roadmap Uses the Term

QuEra states its fault-tolerant roadmap in operations. As described in Why Fault-Tolerant Quantum Computers Need a New Approach to Software, Libra is planned as a megaquop system with 256 logical qubits on more than 10,000 physical qubits, in 2028 on Amazon Braket, with a gigaquop system planned for 2028 to 2029. These are plans, not delivered systems.

The path runs through neutral atoms. Aquila, an analog system, has been on Braket since 2022 without error correction. Gemini, on premises since 2025, uses a zoned, reconfigurable array and serves as a testbed for the codes Libra is meant to run. The post From Megaquop to Gigaquop argues that operations, not qubit count, set the ceiling for an algorithm. QuEra is also building simulators ahead of the hardware, as in PPVM, so teams can prepare early.

FAQ

Who coined the term megaquop?

John Preskill, the Caltech physicist who also coined NISQ in 2018. He laid out the megaquop machine in his 2025 paper "Beyond NISQ: The Megaquop Machine". It names a target for the period between NISQ devices and fully scaled fault-tolerant computers.

Is a megaquop the same as a fault-tolerant quantum computer?

Not exactly. Fault tolerance is a design principle that lets errors be corrected faster than they accumulate, while a megaquop is a performance level, about a million reliable operations. A megaquop machine would use error correction, but a fault-tolerant design could in principle scale well beyond it.

Can a megaquop machine break RSA encryption?

No. Published resource estimates for factoring 2048-bit RSA (Gidney and Ekerå, 2021) involve billions of Toffoli gates, far beyond a million operations. A megaquop is an intermediate milestone, not a cryptographic threat.

Key Takeaways

  • A megaquop is about one million reliable quantum operations, a measure of computation length rather than qubit count.
  • Preskill proposed it in 2025 as the next milestone after NISQ, requiring logical error rates near \(10^{-6}\) per operation.
  • Reaching it requires error correction; physical error rates of roughly \(10^{-3}\) to \(10^{-2}\) fall several orders of magnitude short.
  • Many useful workloads need gigaquop scale or more, and QuEra's Libra and next-generation systems are planned targets, not delivered products.
No items found.

Megaquop

What Is a Megaquop?

A megaquop is a quantum computer that can run about one million quantum operations before an uncorrected error is likely to spoil the answer. The term comes from physicist John Preskill, who used it to name the stage between today's noisy machines and the large fault-tolerant computers that the most demanding algorithms will need.

The unit counts operations, not qubits. A "quop" is one quantum operation, such as a gate or a measurement: a megaquop is \(10^6\) of them, a gigaquop \(10^9\), a teraquop \(10^{12}\). A megaquop machine is described by how long a computation it can complete reliably, not by how many qubits it holds.

Where the Term Comes From: Beyond NISQ

In 2018, Preskill introduced NISQ (noisy intermediate-scale quantum) in "Quantum Computing in the NISQ Era and Beyond" (Quantum, 2018). NISQ devices have enough qubits to be hard to simulate classically but no error correction, so noise limits them to circuits of roughly a thousand operations.

In 2025 Preskill returned to the question in "Beyond NISQ: The Megaquop Machine" (ACM Transactions on Quantum Computing, 2025). He proposed the megaquop as a concrete milestone: a machine with errors near \(10^{-6}\) per operation, reached through quantum error correction rather than by improving physical devices alone. It gives the work beyond NISQ a target to measure progress against.

The Arithmetic Behind a Million Operations

Suppose each operation fails independently with probability \(p\). The chance that \(N\) operations all succeed is approximately

\[ P_{\text{success}} \approx (1-p)^N \approx e^{-pN} \]

For \(N = 10^6\), the product \(pN\) stays near or below one only if \(p\) is about \(10^{-6}\) or lower. Physical two-qubit gates today have error rates of roughly \(10^{-3}\) to \(10^{-2}\), so hardware alone falls short by several orders of magnitude.

Error correction closes the gap. Many physical qubits encode one logical qubit, and when physical errors are below the code's threshold, the logical error rate falls as the code grows. A megaquop machine is one where that suppression reaches about \(10^{-6}\) across many logical qubits and a full set of gates.

One subtlety: operations do not all cost the same. Non-Clifford gates are the expensive ones, typically needing magic state distillation or related methods, so "a million operations" depends on what is counted.

Megaquop, Gigaquop and What Useful Work Needs

ScaleOperationsStatus
NISQabout \(10^3\)Available today; noise limits depth
Megaquop\(10^6\)Target for early error-corrected machines
Gigaquop\(10^9\)Needed for larger algorithms
Teraquop\(10^{12}\)Further out

Reaching a megaquop does not put every useful problem in range. Resource estimates for factoring a 2048-bit RSA integer run to billions of Toffoli gates (Gidney and Ekerå, 2021), well beyond a megaquop. Whether a megaquop system delivers quantum utility depends on the specific problem and is settled by experiment, not by the label.

The megaquop is better read as a stepping stone toward fault-tolerant computing at larger scale.

How QuEra's Roadmap Uses the Term

QuEra states its fault-tolerant roadmap in operations. As described in Why Fault-Tolerant Quantum Computers Need a New Approach to Software, Libra is planned as a megaquop system with 256 logical qubits on more than 10,000 physical qubits, in 2028 on Amazon Braket, with a gigaquop system planned for 2028 to 2029. These are plans, not delivered systems.

The path runs through neutral atoms. Aquila, an analog system, has been on Braket since 2022 without error correction. Gemini, on premises since 2025, uses a zoned, reconfigurable array and serves as a testbed for the codes Libra is meant to run. The post From Megaquop to Gigaquop argues that operations, not qubit count, set the ceiling for an algorithm. QuEra is also building simulators ahead of the hardware, as in PPVM, so teams can prepare early.

FAQ

Who coined the term megaquop?

John Preskill, the Caltech physicist who also coined NISQ in 2018. He laid out the megaquop machine in his 2025 paper "Beyond NISQ: The Megaquop Machine". It names a target for the period between NISQ devices and fully scaled fault-tolerant computers.

Is a megaquop the same as a fault-tolerant quantum computer?

Not exactly. Fault tolerance is a design principle that lets errors be corrected faster than they accumulate, while a megaquop is a performance level, about a million reliable operations. A megaquop machine would use error correction, but a fault-tolerant design could in principle scale well beyond it.

Can a megaquop machine break RSA encryption?

No. Published resource estimates for factoring 2048-bit RSA (Gidney and Ekerå, 2021) involve billions of Toffoli gates, far beyond a million operations. A megaquop is an intermediate milestone, not a cryptographic threat.

Key Takeaways

  • A megaquop is about one million reliable quantum operations, a measure of computation length rather than qubit count.
  • Preskill proposed it in 2025 as the next milestone after NISQ, requiring logical error rates near \(10^{-6}\) per operation.
  • Reaching it requires error correction; physical error rates of roughly \(10^{-3}\) to \(10^{-2}\) fall several orders of magnitude short.
  • Many useful workloads need gigaquop scale or more, and QuEra's Libra and next-generation systems are planned targets, not delivered products.
Abstract background with white center and soft gradient corners in purple and orange with dotted patterns.