In blog posts on its website as well as presentations given at the Q2B conference, IonQ described its strategy and goals for scaling its next generation machines. Their technology strategy includes increasing the number of qubits, improving qubit gate fidelities, miniaturizing its chips and systems, lowering the manufacturing cost, leveraging error correction to improved effective logical qubit quality, and using photonic networks to combine multiple modules into one larger system.
Although IonQ has been using Quantum Volume as a metric, they view this metric as having deficiencies because it does not work very well when you reach large numbers of qubits. As a result, they have developed their own metric they call Algorithmic Qubits which is related to Quantum Volume, but also takes into account error correction algorithms which can create a logical qubit with a higher fidelity level than the component physical qubits. They define an Algorithmic Qubit or AQ as the number of effectively perfect qubits you can deploy in a quantum program. Mathematically, AQ is related to Quantum Volume (QV) by the following formula: AQ = Log2(QV). IonQ has posted an online Algorithmic Qubit calculator that lets you enter in the number of qubits, average 2 qubit gate fidelities and other information to see what AQ score you would achieve with those parameters.
With IonQ’s latest system, which has 32 qubits and an average 2Q gate fidelity of 99.90%, they believe the AQ is about 22 (or an equivalent QV of 222 or 4,194,304). The gate fidelity is currently the limiting factor. If this fidelity increases, then their AQ would increase without adding any more physical qubits. We are hopeful that IonQ will actually run the QV test procedure as prescribed by IBM to see if the empirical results match the theory.
They have indicated that have three generations of products currently undergoing development (at different stages), and expect to regularly roll out higher performance systems on fast schedule. The company believes that they will start seeing a few users demonstrating quantum advantage for selected applications once they hit an AQ level of 40 with more users realizing quantum advantage more broadly once the AQ hits 72.
Shown below is IonQ’s roadmap for providing more powerful systems with increasingly higher AQ levels. It appears that starting in 2025, they will utilize some form of error correction to help them continue the gains in the AQ levels.
As mentioned earlier, they are also developing technology to miniaturize the systems and reduce the cost. This will make it much more feasible to network together multiple systems.
Shown below are some charts that show their goals for miniaturization for both the systems and the quantum cores. The smaller system sizes may pave the way to enabling on-premise installations for those customers who insist on these for security reasons.
December 9, 2020