IBM shut down pulse-level control on its quantum systems in February 2025. Most people missed it. The ones who didn't are now rethinking where to run their experiments.
Pulse-level control means writing directly to the quantum hardware - shaping the microwave pulses that flip qubits, tuning gate timing, optimising error correction at the physics layer. It's the difference between using a compiler's output and writing assembly by hand. For researchers trying to squeeze performance out of noisy hardware, that access is the whole point.
IBM's decision matters because they're the biggest public quantum cloud provider. Their systems are how most academic researchers and small labs access real quantum hardware. A new survey of 13 quantum vendors documents the split: IBM, Google, and other top-tier vendors are closing control-plane access. Neutral-atom providers like QuEra and mid-tier vendors are opening it. The field is bifurcating.
Why Vendors Close Access
From IBM's perspective, pulse-level access is expensive to support. Users can misconfigure gates, destabilise calibration, or run sequences that damage hardware. Debugging user-submitted pulse schedules takes engineering time. Worse, it creates reproducibility chaos - two researchers running "the same" algorithm with different pulse shapes get different results, then publish conflicting claims about quantum advantage.
Closing the control plane also simplifies the product. IBM wants enterprise customers running Qiskit workloads through a stable API, not physics PhD students hand-tuning gate sequences. That's a business decision - serve fewer users with higher-margin contracts, rather than a broad research community with unpredictable needs.
But here's the cost: you can't benchmark what you can't control. If researchers want to compare IBM's superconducting qubits against Rigetti's or IonQ's trapped ions, they need equivalent access to the control layer. Without it, cross-vendor comparisons become meaningless. You're not comparing hardware - you're comparing compilers.
The Neutral-Atom Opening
While superconducting vendors close up, neutral-atom systems are opening. QuEra, Pasqal, and Atom Computing all expose control-plane APIs. Their hardware architectures require it - neutral atoms arrange in arbitrary geometries, and the interaction graph changes per experiment. There's no way to abstract that behind a fixed gate set. Users need low-level access or the system is useless.
This creates a research advantage for neutral-atom platforms. Labs working on error correction, novel gate implementations, or quantum simulation can experiment freely. Superconducting researchers are stuck with whatever IBM's compiler team decided to expose. The gap will show up in publication counts - more novel results from platforms that let you tinker.
Mid-tier vendors like Rigetti are caught in between. They've kept pulse access open longer than IBM, but their user base is smaller. If they close it to reduce support costs, they lose their remaining research users. If they keep it open, they can't scale to enterprise customers who want stability over flexibility. It's a hard position.
What It Means for Reproducibility
The survey's biggest finding: cross-vendor reproducibility is dying. When researchers optimise algorithms at the pulse level on one platform, those results don't transfer. A circuit that works beautifully on IBM hardware fails on IonQ because the gate decomposition is different. A Rigetti-optimised sequence doesn't run at all on Google's closed system.
This fractures the field. Instead of "quantum algorithm X achieves speedup Y", papers now have to say "quantum algorithm X achieves speedup Y on platform Z, using compiler version W, with these specific pulse parameters". That's not how science is supposed to work. Results should generalise. But without control-plane access, there's no way to make them generalise.
For business owners or developers watching quantum from the outside, this matters because it delays real applications. If you can't benchmark reliably across platforms, you can't make informed hardware choices. If researchers can't reproduce results, breakthroughs slow down. And if the best-funded platforms are the least open, the most interesting research migrates elsewhere.
Where This Goes Next
The likely outcome: tiered access models. Enterprise users get high-level APIs with SLAs. Research users get control-plane access with no guarantees. Academic partnerships get dedicated hardware time with full tunability. It's already happening - IBM offers pulse access to select partners, just not publicly.
But that creates inequality. Well-funded labs with IBM contracts can run experiments smaller labs can't replicate. Open-source quantum algorithm development slows down because fewer people can test at the hardware layer. The field risks becoming less collaborative, not more.
The vendors opening their control planes - QuEra, Pasqal, Rigetti - are betting that open access creates stickiness. Researchers who learn on your hardware, publish on your hardware, and build tools for your hardware become advocates. IBM's betting the opposite: that closing access reduces support costs faster than it loses users. Both bets could be right. Or both could be wrong. But the split is real, and it's shaping who gets to do what kind of quantum research for the next five years.