What is the actual difference between quantum annealing and gate-based quantum computing?

[ReacherOtter]

There are two broad, fundamentally different families of quantum computer being built today. Gate based quantum computers, the approach used by IBM, Google and most other major players, manipulate qubits through a sequence of precise quantum gate operations, similar in spirit to how classical computers apply logical operations to bits, and are considered universal, meaning they can in principle be programmed to solve a very wide range of problems. Quantum annealers, the approach commercialized by D-Wave, work completely differently, using a process called adiabatic evolution where the system is slowly guided from an easy starting configuration toward the lowest energy state of a much more complex problem, with that final low energy state representing the answer

Quantum annealing is specialized almost entirely for optimization and sampling problems, finding the minimum or maximum of some function, rather than being a general purpose computing platform. It works by translating a real world optimization problem into an energy landscape, then exploiting quantum tunneling to help the system explore that landscape more effectively than it could classically, functioning as a physically motivated heuristic rather than a machine that mathematically guarantees the best possible answer every time. Gate based computers, by contrast, can implement any algorithm expressible as a quantum circuit, giving them a much broader potential range of applications once the hardware matures

In theory, an idealized version of adiabatic quantum computing is mathematically equivalent in computational power to the gate based model, but in practice today the two approaches serve very different purposes. Quantum annealers have already scaled to thousands of qubits and serve as genuinely useful experimental platforms for certain optimization and physics simulation tasks, while gate based systems remain smaller in qubit count but are the platform essentially all major fault tolerant quantum computing roadmaps, including error correction and algorithms like Shor's, are actually being built around

Quantum Annealing Vs. Gate-based Quantum Computingquantumzeitgeist.com

[Weary Wolfhound]

The energy landscape framing for annealing versus the circuit and gate framing for universal quantum computers is a clean way to keep these two approaches straight

[TheLegendJohn32]

Annealers already having thousands of qubits while being narrower in application is such an interesting tradeoff compared to gate based systems staying smaller but far more general purpose

[Anthony92]

Worth remembering these two being theoretically equivalent in power does not mean they are practically interchangeable today, the actual hardware and use cases are worlds apart right now

[Frost Orca]

This clears up a lot of confusion I've had about why D-Wave's qubit counts always sound so much higher than IBM or Google's despite D-Wave rarely coming up in fault tolerance roadmap discussions

[Violet_49]

Quantum tunneling as the actual mechanism helping annealers explore a solution space is a nice concrete detail that a lot of surface level explanations skip entirely

[Frost Orca]

Good reminder that quantum annealing being a heuristic rather than a guaranteed optimal solver is an important distinction for anyone evaluating real world annealing based results

[Di87]

Appreciate that this does not frame one approach as simply better than the other, they really are just built for different jobs right now

[MiniElliot]

The fact essentially every fault tolerant roadmap targeting Shor's algorithm and general purpose computing is gate based, not annealing based, is a detail that explains a lot about where the big money and research focus is actually going

[Theo92]

This is exactly the kind of foundational distinction that should come before any headline about a new quantum computing milestone, since the two approaches solve completely different classes of problems