QuEra laid out a roadmap from "megaquop" to "gigaquop" fault-tolerant quantum computing. The industry has spent years fielding one question: when will this actually be useful? QuEra's answer is specific. A 2028 delivery date. Named commercial partners. Peer-reviewed research behind every claim.

This post draws on a conversation between Yuval Boger, Chief Commercial Officer at QuEra Computing, and Phil Alsop, Editor at Digitalisation World Magazine. Catch the full interview here.

Operations, not qubits, set the ceiling

A megaquop is roughly a million error-free operations. A gigaquop is roughly a billion. The distinction matters, Boger said, because an algorithm's complexity decides how many operations it needs. What limits a quantum computer is the number of operations it can run, not its raw qubit count.

Individual qubits are fragile. They pick up errors from their environment. So useful computation depends on logical qubits: groups of physical qubits arranged so errors can be caught and corrected as they happen. Getting to gigaquop scale takes two things at once, lowering the error rate of each qubit and using more efficient error-correction codes.

Why neutral atoms

QuEra thinks about the problem in terms of space and time. Space is how many qubits a computation needs. Time is how long it takes to reach an answer. Neutral-atom hardware helps on both, Boger argued. Atoms are identical by nature, which sidesteps the per-unit calibration and manufacturing variability other qubit types fight with. They also carry no electrical charge, so they skip the dilution refrigeration that superconducting systems depend on. Boger put the contrast in physical terms. A superconducting setup can fill a stadium-scale footprint and draw tens of megawatts. A neutral-atom equivalent fits on roughly two dining tables and runs on tens of kilowatts.

Moving qubits means fewer chances for error

The bigger advantage is structural. Neutral-atom qubits can be physically moved. Two atoms interact only when you bring them close, so a calculation becomes a kind of choreography: atoms move into position, interact, then move apart. Nothing is locked to a fixed chip layout.

Boger reached for a game of telephone to explain it. On a fixed-connectivity chip, a message handed qubit to qubit degrades like a whisper passed down a line of fifteen children. On a neutral-atom system, the first child just walks the message to the fifteenth. Fewer hops, fewer chances to introduce an error. That same flexibility is why QuEra's architecture can support several error-correction code families, rather than being stuck with whatever scheme was etched into a fixed chip.

The ratio that reshapes the roadmap

The single most important input to the roadmap is the ratio of physical to logical qubits. Old assumptions pegged it near 1,000 to 1. By that math, 500 logical qubits, enough to beat classical simulation, would demand roughly 500,000 physical qubits. High-rate quantum low-density parity check (QLDPC) codes are rewriting that number.

Libra is QuEra's fault-tolerant system, planned for Amazon Braket in 2028. It is designed for 256 error-corrected logical qubits from just over 10,000 physical qubits, at a logical error rate of 10⁻⁶. A gigaquop-class system follows, targeted for 2028/2029, designed for 1,000 or more logical qubits from roughly 20,000 physical qubits, at a 10⁻⁹ error rate. Every building block behind those targets has already shown up in peer-reviewed work, QuEra says. Turning a lab result into a shipped product still takes time, and the company is clear about that.

Quantum and classical, working together

Boger was blunt on one point. Quantum computers will augment classical infrastructure, not replace it. Take decoding, the job of reading error signals off a logical qubit and correcting them. QuEra's published work with NVIDIA uses machine learning to make decoding faster, which cuts time to solution directly. The relationship runs both ways. AI speeds up quantum error correction and system design. Quantum systems, in turn, can generate training data, in fields like chemistry, that classical AI cannot easily produce on its own.

Why waiting until 2028 is the real risk

The most practical takeaway was about timing. Do not wait until 2028 to start, Boger warned. He described a staged co-design process, run through programs like the FTQC Founder's Circle, that begins by scoping a real business problem and moves through joint algorithm exploration before any code touches hardware. Wait too long, and you read about a competitor's quantum-enabled breakthrough in the press instead of delivering your own. Chemistry and materials science sit closest to measurable value, since their behavior is governed by quantum mechanics to begin with. Optimization and finance are in active exploration too.

Access through the cloud

Cloud access is what makes early engagement possible. QuEra's Aquila system has been on Amazon Braket since 2022, and Libra is set to follow the same route in 2028. That gives an organization of any size room to experiment long before it needs to own hardware. Strip away the specific date, and QuEra's pitch comes down to a published trail of evidence linking today's research to tomorrow's delivery. The organizations best positioned in 2028 will be the ones that started early.

‍