What a roadmap target can and cannot tell you
The tool everybody wants here picks a roadmap year and returns a verdict. This one does not, because a year and a yes read as a forecast however they are captioned. It answers the narrower question instead: given what a vendor has actually stated, what can be compared?
Every figure below is a target
A vendor-stated target, not a prediction, a guarantee, or evidence that any system exists. KetQat does not forecast whether or when a target will be met.
IBM IBM Quantum Starling
Stated: 200 logical qubits; 100 million quantum gates. Source, published 2025-06-10.
Against reference workloads, on logical qubits alone
| Reference workload | Needs | Target states | Comparison |
|---|---|---|---|
| Simulating a 12-qubit circuit: measured classical baseline | 12 | 200 | Within the stated figure |
| Factoring 2048-bit RSA: resource counts from Gidney and Ekera (2019) | 6,189 | 200 | Exceeds the stated figure |
Not compared: A workload KetQat refuses to assess — KetQat refuses to assess it, so it has no requirement to compare. Its logical-qubit count exists, but a number KetQat will not stand behind is not a requirement.
What this target does not state
- Physical error rate
- Surface-code cycle time
- Magic-state throughput
- Physical qubit capacity
A resource estimate depends on every one of those. Without them, the comparison above is one requirement against one number and nothing more — it cannot become a statement that a workload would run, in 2029 or ever.
IBM IBM Quantum Blue Jay
Stated: 2,000 logical qubits; one billion gates. Source, published 2025-06-10.
Against reference workloads, on logical qubits alone
| Reference workload | Needs | Target states | Comparison |
|---|---|---|---|
| Simulating a 12-qubit circuit: measured classical baseline | 12 | 2,000 | Within the stated figure |
| Factoring 2048-bit RSA: resource counts from Gidney and Ekera (2019) | 6,189 | 2,000 | Exceeds the stated figure |
Not compared: A workload KetQat refuses to assess — KetQat refuses to assess it, so it has no requirement to compare. Its logical-qubit count exists, but a number KetQat will not stand behind is not a requirement.
What this target does not state
- Physical error rate
- Surface-code cycle time
- Magic-state throughput
- Physical qubit capacity
A resource estimate depends on every one of those. Without them, the comparison above is one requirement against one number and nothing more — it cannot become a statement that a workload would run, in 2033 or ever.
Why there is no year-by-year projection here
Interpolating between announced milestones would produce a curve, and a curve invites reading off a date. The announcements do not describe a rate of progress; they describe two points a vendor intends to reach. Fitting a line through them would be this product's own invention presented as somebody else's plan.
If you want to know what a machine would have to achieve, that question has an answer and KetQat computes it: the threshold conditions on every assessment state the cycle time, error rate and capacity that would change the comparison. A condition is checkable against a datasheet. A date is not.
See what has been measured, how the estimates are produced, or what KetQat will not tell you.