In a significant leap for quantum hardware, Quantinuum today launched Helios—a 98-qubit ion trap quantum computer that addresses one of the field's most persistent challenges: error correction. Unlike classical computing, where redundant bits easily correct errors, quantum systems suffer from environmental interference that collapses fragile quantum states. Helios tackles this through an elegant ion-based architecture requiring just two physical qubits to create one error-corrected "logical qubit," dramatically improving efficiency over superconducting competitors.

Caption: Quantinuum's ion trap chip at the heart of Helios, where individual barium ions serve as qubits.

The Ion Advantage

Helios uses barium ions suspended in electromagnetic fields as qubits, a departure from the superconducting circuits favored by IBM and Google. This design enables "all-to-all connectivity"—each ion can interact directly with any other, eliminating the need for complex routing through neighboring qubits. Quantinuum VP Jennifer Strabley emphasizes this as critical: "It’s becoming increasingly apparent how important this connectivity is for high-performing systems."

When entangled (a fundamental quantum operation), Helios' qubits maintained accuracy 99.921% of the time—a record that physicist Rajibul Islam (University of Waterloo) calls unmatched: "To the best of my knowledge, no other platform is at this level." This inherent precision reduces the hardware overhead for error correction.

Error Correction Showdown

Quantum error correction typically demands massive physical qubit counts to safeguard logical qubits. Recent efforts illustrate the gap:
- Google: 105 physical qubits per logical qubit (2024)
- IBM: 12 physical qubits per logical qubit (2025)
- AWS: 9 physical qubits per logical qubit (2025)

Helios achieves the same with just two, thanks to ion mobility and parallel error detection via NVIDIA GPUs. David Hayes, Quantinuum’s Director of Computational Theory, notes this enables real-time "on-the-fly" correction—a first for the company.

Scaling the Roadmap

While still years away from commercial applications like materials science or financial modeling, Quantinuum’s trajectory is aggressive:
- 2027: 192-qubit "Sol" system
- 2029: "Apollo" with thousands of qubits and full fault tolerance

Islam cautions that the quantum race remains open: superconducting qubits are easier to manufacture, while neutral atoms (like QuEra’s approach) simplify trapping. Yet Helios proves ions offer a compelling path—fewer qubits, higher fidelity.

Already, Helios simulates electron behavior in high-temperature superconductors, tackling problems the Department of Energy prioritizes. As error correction efficiency improves, these machines edge closer to solving problems beyond classical reach.

Source: Technology Review