Insider Brief
IBM says a new quantum error correction method called the Gross code could enable practical quantum advantage within two years.
The code reduces the number of physical qubits needed per logical qubit by a factor of ten and enables a modular design easier to manufacture and scale.
Gross code is a type of low-density parity check (LDPC) code that spreads quantum information sparsely, allowing for efficient error detection with simpler hardware layouts.
IBM reports a novel method for reducing quantum computing’s error burden could push useful quantum machines into action within two years.
The technique — called the Gross code — shrinks the number of physical qubits required to produce stable output, significantly easing the engineering burden, according to R&D World.
“The Gross code bought us two really big things,” Oliver Dial, IBM Quantum’s chief technology officer, said in an interview with R&D World. “One is a 10-fold reduction in the number of physical qubits needed per logical qubit compared to typical surface code estimates.”
Fewer physical qubits mean fewer unreliable parts to manage, from connectors to cryogenic systems.
“From an engineering perspective, that’s huge,” Dial added.
The second benefit is a streamlined design that’s far easier to manufacture and scale, according to Dial. IBM’s approach packages 12 logical qubits within a roughly 300-physical-qubit module — what Dial calls a “repeating unit cell” — that can be built, tested and integrated using current fabrication methods.
“The engineering of building this system got enormously easier,” he told the magazine.
The Gross code is IBM’s answer to a problem that has defined quantum computing from the start: quantum bits, or qubits, are noisy and prone to errors. That’s a problem for large calculations, which require reliable logical qubits — units of stable information made by combining many physical qubits and correcting for their failures. Most error-correcting schemes demand thousands of physical qubits to protect a single logical one. IBM says its new code could cut that number to just 25.
Gross, as in a Dozen Dozen
The name “Gross code” comes from the number 144 — a dozen dozen — reflected in the structure of the system. According to Dial, the architecture allows IBM to encode 12 logical qubits into a block of roughly 300 physical ones, creating a compact and repeatable structure that can scale.
That could reshape expectations for when quantum computing starts solving real-world problems. Industry analysts increasingly forecast that “quantum advantage” — the point, although admittedly a bit arbitrary, where quantum computers beat classical systems on useful tasks — will arrive in the next few years.
Error Correction AND Error Mitigation
IBM’s optimism is grounded not just in long-term error correction, but in near-term tactics like error mitigation, a strategy to extract meaningful results from today’s imperfect machines.
These techniques offer a way to recover accurate answers from computers that commit errors, Dial told R&D World. He sees this as a bridge between today’s noisy intermediate-scale quantum (NISQ) machines and tomorrow’s fully fault-tolerant quantum computers.
Competitors are also racing to prove real-world use cases. Google has published recent results in quantum error correction, while Quantinuum and JPMorgan Chase are exploring secure applications like random number generation, R&D World points out. IBM’s bet is that better codes, especially its low-density parity check (LDPC) approach refined through the Gross code, will accelerate real deployments.
How Do LDPC Codes Work?
Gross code belongs to a class of quantum LDPC codes. These codes spread information out across the qubits in ways that allow errors to be detected and corrected with fewer connections between each qubit, making physical layouts easier.
More specifically, LDPC codes — and Gross codes — work like this: Researchers create parity checks, or logical constraints, that qubits must follow. These checks act like sensors, flagging when a qubit has flipped or changed in a way it shouldn’t have. Because each qubit is part of only a few checks, and each check involves only a few qubits, the system stays sparse and manageable—even as it scales.
Unlike surface codes, which require a tight 2D grid of qubits and a high degree of connectivity (a challenge in hardware), LDPC codes like the Gross code spread information more thinly but still retain strong error-detecting power.
The design, first detailed in Nature in 2024, stands in contrast to “surface codes,” which require dense connectivity and large numbers of physical qubits to work.
IBM says its modular Gross code units reduce the scale and wiring complexity to something far more manageable. Earlier visions of fault-tolerant quantum computers imagined massive, warehouse-sized machines; the Gross code offers a smaller, more practical footprint.
That difference could be the key to IBM’s ambitious timeline, Dial suggests. The company’s public roadmap calls for significant performance milestones by the end of the decade: by 2025, faster execution and parallel computing; by 2027, circuits running 10,000 gates; and by 2029, a fully error-corrected system with 200 logical qubits capable of handling 100 million gates. IBM believes the Gross code helps keep those targets within reach.
Roadmap, Not a Death Pact
Still, the roadmap is flexible, according to Dial.
“This is a roadmap, not a death pact,” Dial said. “If circumstances change, we’ll still accomplish these goals, but we may do it differently.”
One advantage of the Gross code is its simplicity for manufacturing. Earlier approaches sometimes implied building arrays of thousands of physical qubits for each logical qubit. Those arrays could span “football-field-sized” with wiring complexity and material yields that are, as one could imagine, difficult to manage. The new unit-cell design, fitting 12 logical qubits into a relatively compact structure, is far easier to prototype and test.
Executing that plan requires commitment. Dial emphasized the stability of IBM’s quantum team, telling R&D World: “I’ve been on this team for 12 years. Almost all the original team from 12 years ago is still here because this is our life’s work—to get these machines out the door.”
While the company may have dozens of products and services, Dial said the goal of building the first useful quantum computers makes it easy to rally the company around.
Still, the promise of quantum advantage is no guarantee. Making these machines practical depends on continued progress in error management, software, and hardware development. But if IBM’s estimates hold, the field could be just two years away from crossing a critical threshold.
“Now that we’ve put it on the roadmap, that’s when the real work begins,” Dial said. “We still have to make that happen.”
