By IBMLearning Objectives
- Understand what IBM Quantum offers today and why it is the most accessible way to run real quantum circuits
- Identify IBM's current processors (Heron, Nighthawk) and its fault-tolerance roadmap (Starling, 2029)
- See how IBM uses AI — for real-time error decoding and for AI-assisted quantum coding with Qiskit
What Is IBM Quantum?
IBM Quantum is IBM's end-to-end quantum computing platform: the physical superconducting quantum processors, the cloud service that lets anyone run circuits on them, and Qiskit, the open-source software toolkit that has become the most widely used way to program a quantum computer. IBM put the first quantum computer on the cloud in 2016, and it now operates the largest fleet of cloud-accessible machines in the industry, available through a free tier all the way up to premium enterprise plans.
Unlike a research chip with no product behind it, IBM Quantum is something you can actually use today. A student can sign up, write a few lines of Qiskit, and submit a job to a real quantum processor. A bank or pharmaceutical company can buy dedicated access and install an on-premises Quantum System Two in its own data center.
💡Key Concept
Superconducting qubits: IBM builds its qubits from tiny superconducting circuits chilled to near absolute zero. This is the same broad family of qubit used by Google. The advantage is that these chips can be manufactured with semiconductor-style techniques and scaled to large qubit counts; the challenge is that each qubit is fragile and error-prone, which is why IBM now emphasizes error correction and qubit quality over raw qubit count.
Today's Processors
After its 1,121-qubit Condor chip in 2023 proved raw scale was possible, IBM deliberately pivoted away from chasing the biggest qubit count toward quality, gate fidelity, and error correction. Its current processors reflect that shift:
| Processor | Qubits | Role |
|---|---|---|
| Heron r2 | 156 | Workhorse processor; powers Quantum System Two installations worldwide |
| Nighthawk | 120 | Higher-connectivity chip (introduced November 2025) able to run up to 5,000 two-qubit gates |
| Condor | 1,121 | A 2023 scale demonstration; strategy then pivoted to quality over count |
These processors are deployed inside Quantum System Two, IBM's modular quantum computer, with installations running at sites including RIKEN in Japan and a center in Spain, in addition to IBM's own data centers in New York.
The Roadmap: Fault Tolerance by 2029
In June 2025, IBM published one of the most detailed and credible roadmaps in the field. Its goal is IBM Quantum Starling — a large-scale, fault-tolerant quantum computer with roughly 200 logical qubits — by 2029, built in a new data center in Poughkeepsie, New York.
The key technical bet is a more efficient error-correction scheme called quantum low-density parity-check codes, which IBM says can cut the overhead of error correction by roughly 90% compared with older approaches. The roadmap lays out named milestones along the way — Loon, Kookaburra, and Cockatoo — leading to Starling, and targets a demonstration of verifiable quantum advantage by the end of 2026.
📝Note
Logical versus physical qubits. Today's quantum chips have "physical" qubits that make frequent errors. Error correction combines many physical qubits into one stable "logical" qubit. A fault-tolerant machine like Starling is defined by its logical qubits — roughly 200 of them — which is why the headline number is far smaller than Condor's 1,121 physical qubits but far more powerful.
The AI Angle
IBM Quantum intersects with AI in two concrete, shipping ways:
AI for Error Correction
Correcting quantum errors in real time means decoding a flood of measurements faster than new errors appear. IBM demonstrated a real-time decoder called Relay-BP running on standard AMD FPGA hardware — an important step that uses fast, machine-learning-style decoding to make error correction practical on affordable, off-the-shelf electronics rather than exotic custom chips.
AI for Quantum Developers
Writing quantum programs is unfamiliar territory for most developers. The Qiskit Code Assistant uses IBM's watsonx Granite large language models to suggest and complete quantum code, lowering the barrier for newcomers — a clear case of using today's AI to make quantum programming more approachable.
How You Access It
| Tool | Best For |
|---|---|
| IBM Quantum Platform | Sign up and run circuits on real processors; free open plan plus premium tiers |
| Qiskit | Open-source SDK for building, optimizing, and running quantum programs |
| Qiskit Runtime | Managed execution service that runs quantum workloads efficiently at scale |
Strengths
- The most mature platform — the largest fleet of cloud-accessible quantum computers, with a free tier you can start using today
- The default software stack — Qiskit is the most widely adopted quantum programming toolkit, with a large community and learning resources
- A credible roadmap — a detailed, public path to a fault-tolerant 200-logical-qubit machine by 2029, with named interim milestones
- Real AI integration — machine-learning-style error decoding on commodity hardware and AI-assisted quantum coding via watsonx Granite
- Enterprise-ready — on-premises Quantum System Two installations for organizations with data-sovereignty or dedicated-access needs
Limitations & Considerations
- Still early — today's processors are noisy and pre-fault-tolerant; they are powerful for research and learning, not yet for production business problems
- No proven quantum advantage yet — IBM targets a verifiable advantage by the end of 2026, but it has not been demonstrated
- Roadmap risk — the 2029 Starling target is ambitious and depends on error-correction techniques that are still being scaled up
- A real learning curve — quantum programming requires new concepts even with Qiskit and the Code Assistant smoothing the path
Best Use Cases
| Scenario | Why IBM Quantum fits |
|---|---|
| Learning quantum computing hands-on | Free tier plus Qiskit and tutorials make it the easiest place to start |
| Academic and R&D research | Largest fleet of real processors with broad device access |
| Enterprise quantum exploration | Dedicated access and on-premises Quantum System Two for regulated industries |
| Tracking the path to fault tolerance | The most detailed public roadmap to a large-scale error-corrected machine |
Getting Started
- Create a free account on the IBM Quantum Platform and explore the available processors
- Install Qiskit and run your first circuit on a real quantum computer in a few lines of code
- Work through IBM's Qiskit learning materials, optionally using the Qiskit Code Assistant for AI-assisted help
- Follow the Starling roadmap milestones as a yardstick for when error-corrected quantum computing becomes practical
Key Takeaways
- IBM Quantum is the most mature, most accessible way to run real quantum circuits today — the largest cloud fleet of quantum computers plus the widely used Qiskit toolkit, from a free tier to enterprise
- Current processors Heron (156 qubits) and Nighthawk (120 qubits) power IBM's modular Quantum System Two, after IBM pivoted from raw qubit count toward quality and error correction
- IBM's public roadmap targets a fault-tolerant 200-logical-qubit machine, Starling, by 2029, using efficient error-correction codes that cut overhead by roughly 90%
- The AI connection is real and shipping: machine-learning-style error decoding on commodity FPGA hardware and the Qiskit Code Assistant built on watsonx Granite models
