This tracker collects IBM and Red Hat patents that address the physical and software plumbing of quantum computers, from switches and wiring to error correction and result-predicting software. The filings point to a coordinated effort to make quantum systems easier to build, debug, and run alongside conventional computing.
26 filings · tracking since May 2026 · latest Nov 2026 · updates weekly
based on all tracked filings in this watchlist · refreshes every week
IBM's filings here are almost entirely about making quantum computers work reliably at scale, covering everything from catching and fixing errors to managing how jobs are scheduled and run.
The heaviest concentration is on error correction and noise reduction, with a secondary cluster around software tools that translate and reroute quantum programs across different hardware setups.
a dated entry each week this watchlist moves · older entries stay archived
This week's filings focus on making quantum computers more practical to use. IBM is working on sharing access between users, catching errors after calculations, and loading new tasks before old ones finish.
This week's filing focuses on sharing. IBM is exploring ways to let multiple quantum programs take turns using the same quantum machine, rather than each needing its own dedicated hardware.
This week's filing focuses on catching and correcting errors inside a quantum computer. IBM is working on ways to find mistakes and fix them automatically.
This week's filing focuses on keeping a large quantum computer running smoothly by adding a system that coordinates and manages its many moving parts at once.
All three new filings focus on making quantum computers more reliable. IBM is working on better error handling, stronger connections between parts, and cleaner program execution.
IBM’s tracked filings by month · newest first
15 or more filings that month · 5 to 14 · under 5
the problems IBM keeps filing on · each with its three newest filings · new filings join every week
Quantum computers make mistakes because tiny signals interfere with each other and hardware defects creep in. These filings cover ways to quiet that interference, find hidden defects, and catch and fix errors before they ruin a calculation.
Different quantum machines speak different languages and cannot easily share work. These filings cover middleware that bridges incompatible machines, automatic program rewriting for different hardware, and a system that reroutes jobs on the fly.
Before running a real job, it helps to know what will happen and what resources it will need. These filings cover a neural network that predicts circuit outputs, a system that predicts resource needs before jobs run, and a simulator that can handle many calculations at once.
Building quantum computers that can grow and stay reliable requires careful physical design. These filings cover chambers that snap together, a wiring layout that cuts interference, a switch that steers signals, and a way to split the machine into separate work zones.
Moving the ball around becomes the instructions, so people can build quantum programs with their hands.
Machine learning takes over the expert guesswork of tuning how a quantum computer fixes its own mistakes.
Predicting the output saves running the costly, error-prone quantum hardware every single time.
The constant errors of quantum chips become useful training data for machine learning.
every tracked filing, month by month · counts are USPTO pre-grant publications, one per publication number (an application, not a granted patent)
US 2026/0350454 A1
Calibration runs on classical test problems let engineers measure residual errors that persist after existing error-correction steps, enabling subtraction of that hidden noise from final quantum results.
US 2026/0350448 A1
Multiplexing several users' quantum jobs on a single chip cuts idle time and skips redundant error characterization between runs, letting the machine serve more people without repeating expensive calibration steps.
US 2026/0350445 A1
IBM has been working on error correction and qubit stability; this filing targets a separate efficiency gap: the dead time between jobs when control systems sit idle waiting for new instructions.
US 2026/0311017 A1
Scheduling qubits like hotel rooms prevents multiple programs from jamming the same quantum processor simultaneously, moving from first-come-first-served access to managed reservation slots.
US 2026/0300797 A1
A mathematical shortcut for decoding quantum errors in real time, potentially letting error-correction keep pace with the rate errors actually occur in hardware.
US 2026/0289375 A1
A two-tier control architecture separates high-level quantum instructions from hardware commands, letting programmers work above the machine-code level as systems grow larger and more complex.
US 2026/0268200 A1
A working quantum computer needs to keep calculations running while constantly fixing errors as they happen. This filing shows how separating storage from processing lets the machine catch and correct mistakes without interrupting the computation.
US 2026/0268197 A1
A working quantum computer needs someone to constantly tune error-correction settings by hand. This system learns which settings work for different hardware states, removing that manual step from operations.
US 2026/0268193 A1
Quantum computers need qubits to interact selectively without crosstalk corrupting neighboring bits. This coupler design isolates two-qubit gates, reducing the stray interactions that currently limit scaling to larger processor arrays.
US 2026/0236819 A1
Quantum machines must physically reroute calculations around hardware defects; training an ML model to do this routing with minimal noise degradation cuts a direct path to usable results.
US 2026/0187507 A1
After mapping the wiring and switches, IBM now automates the translation between what programmers write and what hardware can actually run, using machine learning to find efficient paths through physical constraints.
US 2026/0178950 A1
Letting multiple jobs run simultaneously on a single quantum machine cuts idle time that currently wastes expensive hardware. This confirms IBM's strategy of improving utilization rather than just building more machines.
US 2026/0170385 A1
Automatic rewriting of quantum code during compilation solves a core portability problem: programs written for one chip's instruction set now run on competing hardware without manual recoding, reducing the friction of multi-vendor quantum environments.
US 2026/0170376 A1
Quantum computers need classical computers to manage their resources in real time. This filing shows IBM working out how to forecast those resource demands upfront, so jobs won't stall or fail partway through execution.
US 2026/0172130 A1
The plumbing work continues deeper: IBM moves beyond cable shielding to redesign how control wires physically sit next to each other, preventing signal bleed before it starts rather than fighting it afterward.
US 2026/0171779 A1
Scaling cryogenic chambers by stacking modular units instead of building single large enclosures sidesteps the engineering and cost barriers of manufacturing oversized freezers.
US 2026/0170374 A1
A rolling ball maps physical rotations directly to quantum gate operations, letting programmers bypass abstract math notation and work with spatial intuition instead.
US 2026/0170373 A1
Quantum computers lose reliability as hidden defects accumulate during operation. This filing proposes using machine learning to automatically detect and map these defects so operators can compensate without shutting down the system.
US 2026/0172053 A1
After patenting the physical infrastructure for quantum systems, IBM now moves upstream to the error detection layer itself, describing a two-stage method to identify and fix qubit failures before cascading corrupts the entire calculation.
US 2026/0170380 A1
Qubit measurement noise corrupts results, so splitting the signal path into separate channels lets IBM detect the quantum state without the crosstalk that degrades conventional single-path readouts.
US 2026/0170388 A1
A failover mechanism that reassigns quantum jobs to alternative processors when the primary target goes offline, solving the practical problem of job loss from hardware unavailability without requiring manual intervention.
US 2026/0154140 A1
As IBM and Red Hat move past individual machine optimization, this filing shows them building a translation layer so different vendors' quantum systems can operate within a single workflow.
US 2026/0154595 A1
Quantum error rates usually demand filtering or correction. IBM's filing shows how to extract signal from the noise itself, feeding measurement artifacts into machine learning models instead of discarding them.
US 2026/0141282 A1
A neural network learns to mimic quantum circuit behavior from prior runs, letting developers test code paths without burning through expensive hardware cycles each time.
US 2026/0141278 A1
Quantum circuit simulators could run thousands of test variations without recalculating from scratch each time. This patent stores intermediate wave functions to let researchers skip redundant computation and run simulations in parallel.
US 2026/0135559 A1
Switching quantum signals without moving parts that could introduce heat and noise requires superconducting materials that IBM's design keeps tunable across different operating states.
The filings span hardware issues like signal routing, wiring interference, and cryogenic assembly, as well as software problems like predicting circuit outputs, managing noise, and estimating resource needs before a job runs. Together they describe an effort to make quantum computers more practical to build and operate, alongside making them more powerful.
Red Hat appears in filings focused on interoperability, including a middleware layer that connects quantum computers from different vendors and a system for rewriting quantum programs across hardware types. That suggests IBM is working with Red Hat specifically on the software layer that lets different quantum systems talk to each other.
No. A patent filing describes an idea IBM wants to protect, not a shipped product. Some filings here, like a ball-based programming controller, read more like exploratory research than something headed for a data center anytime soon.
Quantum computers are unusually sensitive to errors and noise, and several filings address that from different angles: one method treats noise as usable information rather than a flaw, another focuses on faster two-step error correction, and a third uses AI to hunt for hidden hardware defects. Reliability is a recurring concern across the watchlist.
Want this weekly breakdown for a company we don't cover? Patentlyze Pro →