Control-Plane Openness in Near-Term Quantum Computing: A Survey of Vendor Stacks and Field Implications The author maintains an open-source pulse-control software project that uses one of the vendor stacks surveyed here (IQM) as its primary integration target. No claim in this paper depends on or references that project, and the IQM row is graded only on public documentation.

Rylan N. Malarchick · 2026

Read Control-Plane Openness in Near-Term Quantum Computing: A Survey of Vendor Stacks and Field Implications The author maintains an open-source pulse-control software project that uses one of the vendor stacks surveyed here (IQM) as its primary integration target. No claim in this paper depends on or references that project, and the IQM row is graded only on public documentation. by Rylan N. Malarchick (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-15233.txt.

Introduction

Rylan Malarchick offers a six-axis rubric gauging control-plane openness in near-term quantum computing, the layer between gate-level circuits and physical electronics. The rubric is defined so same evidence yields same grades. It is validated by blinded re-grading, an edge-case method, and a published protocol. Tied to IBM's February 2025 removal of pulse-level access, it tracks change over time. Applied to thirteen vendors, it shows harms including lost redo access to pre-2025 pulse tests.

Catalog record

CREATORRylan N. Malarchick
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE54450
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.15233

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Control-Plane Openness in Near-Term Quantum Computing: A Survey of Vendor Stacks and Field Implications The author maintains an open-source pulse-control software project that uses one of the vendor stacks surveyed here (IQM) as its primary integration target. No claim in this paper depends on or references that project, and the IQM row is graded only on public documentation. Rylan N. Malarchick Abstract Public access to pulse-level and control-electronics interfaces in commercial quantum computing has bifurcated. The largest superconducting cloud platforms have closed access at this layer, with IBM removing pulse-level control from all production QPUs in February 2025; mid-tier superconducting vendors and the more open neutral-atom platforms have moved in the opposite direction. We survey thirteen commercial vendors across superconducting, trapped-ion, neutral-atom, and photonic modalities, grading each on six axes of openness at what we call the control plane: the layer between gate-level circuit specification and physical control electronics. The catalog ships as a separate machine-readable artifact under CC-BY-4.0. The bifurcation is documented row by row, with implications for reproducibility, hardware-aware research, and cross-vendor benchmarking. We do not propose an architecture or a reference implementation; we describe what the field has lost as the access landscape has shifted, and what minimally open access at this layer would have to look like. I Introduction The control plane is the layer of the quantum computing stack between gate-level circuit specification and the physical control electronics. It includes pulse synthesis, calibration loading and drift compensation, pulse and measurement scheduling, measurement readout, and feedback dispatch. The control plane is what sits between “I want a CNOT” and “the FPGA emits these specific microwave envelopes.” It is the layer at which most of what makes a quantum experiment reproducible or hardware-aware actually lives. The term is borrowed from networking and cloud infrastructure and is not standard quantum-computing terminology; we use it to refer specifically to this layer for the duration of the paper. We grade the control plane on two surfaces, and the catalog covers both. The first surface is what happens inside the control plane: pulse synthesis algorithms, scheduling logic, feedback dispatch protocols, and the structure of the calibration routine. The second surface is what the control plane publishes outward to users: calibration data, provenance metadata, reproducibility commitments, and the schemas through which any of these are exposed. A vendor can be open on the inward surface and closed on the outward, or open on the outward and closed on the inward. The catalog distinguishes the surfaces in its column structure. Fig. 1 shows the layers we have in mind. This is the only architectural figure in the paper. We deliberately keep it as a flat horizontal stack rather than as an architectural decomposition: the layer labels are descriptive of what every quantum computing platform has, not prescriptive of any particular project’s internal structure. Compiler frontend (gate-level circuits, OpenQASM 3, etc.) Control plane pulse synthesis, calibration loading, scheduling, measurement readout, feedback dispatch Control electronics (FPGA gateware, RF chain, AWGs) Physical hardware (qubits, lasers, traps, photonics) Fig. 1: The layers of a quantum computing platform as referenced in this paper. The control plane (shaded) is the layer the catalog grades. The compiler frontend, control electronics, and physical hardware are out of scope for the catalog itself but are referenced in the per-column rubric and in the steel-man section to delineate what the control plane is and is not. The figure is descriptive of what every commercial platform has, not prescriptive of any particular project’s internal architecture. Public access to the control plane has bifurcated. The largest superconducting cloud platforms have closed access at this layer over the past three years, with IBM’s removal of pulse-level control from all production QPUs on February 3, 2025 standing as the most consequential single event in the period under review [ 10 , 30 , 31 ] . Mid-tier superconducting vendors (IQM, Rigetti, OQC) and the more open neutral-atom platforms (Pasqal, QuEra) have moved in the opposite direction, with mature pulse-level interfaces published and maintained [ 14 , 38 , 26 , 28 , 35 ] . Trapped-ion vendors have held steady at native-gate-level access [ 11 , 32 ] . The photonic landscape is split three ways [ 42 , 24 , 25 ] . The bifurcation has implications for what can be reproduced, what can be researched, and what can be compared across platforms. I-A Three Harms Three specific harms follow from the bifurcation. We define each narrowly so that the catalog can demonstrate them concretely. The first harm is reproducibility . A previously-published result cannot be re-run on the platform it was performed on because the access required is no longer available. The canonical example is the body of pre-2025 IBM Quantum methods literature that used Qiskit Pulse: as of February 3, 2025, IBM removed pulse-level control from all QPUs [ 10 ] , which means a non-trivial fraction of that literature cannot be re-executed on IBM hardware as of the May 1, 2026 cutoff used here. This is not a hypothetical harm; it is dated. The second harm is hardware-aware research . A class of experiments that depends on access to a specific layer cannot be performed on a given platform because the access is not exposed. Examples include closed-loop measurement-conditioned feedback, which depends on real-time pulse dispatch at the control electronics; novel pulse decompositions for specific gate sets, which depend on raw waveform synthesis; dynamical decoupling sequences, which depend on precise inter-pulse spacing; and calibration drift studies, which depend on published calibration data over time [ 41 , 40 ] . The catalog documents, per platform, which of these are publicly possible. The third harm is cross-vendor benchmarking . The same benchmark protocol cannot be executed identically on two platforms because the calibration metadata is published in incompatible formats or not published at all, or because the underlying pulse-level interface differs in ways that make “the same circuit” mean different things across hardware. Comparing a closed-loop benchmark on Rigetti, which exposes Quil-T pulse-level access, to one on IonQ, which exposes native gates only, is structurally impossible at the pulse layer; the comparison can only happen at a layer of abstraction high enough to lose the signal the benchmark is trying to measure [ 38 , 11 , 17 , 21 ] . I-B Paper Structure Section II names what is in and out of scope. Section III positions this work relative to recent perspectives and standardization efforts. Section IV presents the catalog: thirteen vendors, six columns, with per-row commentary and a per-column rubric. Section V reads what the catalog shows. Section VI takes the strongest counterarguments seriously, eight of them, with a response to each. Section VII looks at three historical analogues from open computing infrastructure: Linux, LLVM, and the SkyWater open process design kit. Section VIII sketches what minimally open access at the control plane would have to look like, framed as needs the field has, not as a product proposal. Section IX concludes. The catalog itself is also released as a separate machine-readable artifact with source URLs and accessed-on dates per cell. II Scope and Methodology This paper covers gate-model commercial quantum computing platforms across four hardware modalities: superconducting transmon, trapped ion,

Read the complete text