What Can You Realistically Learn Without Expensive Quantum Hardware?

Two separate things get conflated here: learning quantum computing, and running on real quantum hardware. They're not the same requirement. And the second one, when you actually need it, is cheaper than most people assume.

Real hardware, for free

IBM's Open Plan gives anyone with a free IBM account real access to its quantum hardware fleet (100+ qubit processors, capable of circuits with up to 5,000 gates) at no cost, with up to 10 minutes of quantum runtime per month.^[1] That's enough to run real circuits and see real hardware noise and results firsthand, which matters for building honest intuition about where today's quantum computers still fall short. Educators can additionally set up Classroom Accounts so students don't need to provide a personal credit card at all.^[1]

Amazon Braket's free tier works differently but is equally real: one free hour per month of managed cloud simulation across its SV1, DM1, and TN1 simulators, enough to run substantial simulated circuits before any cost applies. Braket's real-hardware runtime is priced from the start rather than offered free, which is simply a different tradeoff, not a lesser one, from IBM's free-hardware/limited-minutes model.^[2]

Microsoft's Azure Quantum takes a third approach, worth knowing before you assume all three major clouds work the same way: its simulators are free across every provider (Rigetti's Quantum Virtual Machine is explicitly free for all users under all plans, and IonQ's own simulator is free of charge through Azure too), but none of Azure Quantum's real hardware is.^[3] Pricing there is built for research teams and enterprises rather than an individual learner, ranging from per-execution charges up to five- and six-figure monthly subscriptions depending on the provider.^[3] That's not a worse deal, it's simply not the free-hardware-minutes model IBM offers; know which model you're looking at before picking a platform.

Simulators: free, no account needed at all

Before any of that, the tools most people actually learn on are open-source and run entirely on a laptop: Qiskit Aer, Cirq, and PennyLane are all free, install locally, and simulate circuits with no cloud account, credit card, or monthly quota at all. For learning (building intuition about superposition, gates, and measurement), a simulator behaves identically to real hardware except for the noise real qubits introduce, which is itself a useful thing to learn to reason about separately.

Structured courses, also free

Free tools still leave "where do I start" unanswered, and that's covered too. IBM Quantum Learning, IBM's own course library, is free to the entire community with no paywall on the material itself.^[4] The standout is "Understanding Quantum Information and Computation," a full university-level course taught by John Watrous, IBM Quantum's technical director for education and author of a standard graduate textbook in the field, covering the actual mathematical foundations rather than a pop-science version.^[4] It sits alongside more applied courses like Quantum Computing in Practice, aimed at people building real circuits rather than studying the theory in the abstract.

None of this requires picking one platform and committing. A realistic path looks like: learn the concepts and build circuit intuition on a free local simulator first, since there's no account friction slowing you down, take a structured course once you hit ideas that don't click from documentation alone, then move to IBM's free hardware minutes once you specifically want to see real noise, not simulated noise, on a circuit you already understand. Cost never forces that order; curiosity does.