IBM Plans to Develop a Functional Quantum Supercomputer by 2029

In less than five years, you’ll have access to a Quantum SuperComputer without errors, according to IBM. The company has unveiled a roadmap for a machine named Starling, set to be available for academic and industrial researchers by 2029.

“These are scientific dreams that have been transformed into engineering achievements,” says Jay Gambetta at IBM. He mentions that he and his team have developed all the required components to make Starling a reality, giving them confidence in their ambitious timeline. The new systems will be based in a New York data center and are expected to aid in manufacturing novel chemicals and materials.

IBM has already constructed a fleet of quantum computers, yet the path to truly user-friendly devices remains challenging, with little competition in the field. Errors continue to thwart many efforts to utilize quantum effects for solving problems that typical supercomputers struggle with.

This underscores the necessity for a fault-tolerant quantum computer that can autonomously correct its mistakes. Such capabilities lead to larger, more powerful devices. There is no universal agreement on the optimal strategy to tackle these challenges, prompting the research team to explore various approaches.

All quantum computers depend on qubits, yet different groups create these essential units from light particles, extremely cold atoms, and in Starling’s case, superconducting qubits. IBM is banking on two innovations to enhance its robustness against significant errors.

First, Starling establishes new connections among its qubits, including those that are quite distant from one another. Each qubit is embedded within a chip, and researchers have innovated new hardware to link these components within a single chip and connect multiple chips together. This advancement enables Starling to be larger than its forerunners while allowing it to execute more complex programs.

According to Gambetta, Starling will employ tens of thousands of qubits, permitting 100 million quantum manipulations. Currently, the largest quantum computers house around 1,000 physical qubits, grouped into roughly 200 “logical qubits.” Within each logical qubit, several qubits function together as a single computational unit resilient to errors. The current record for logical qubits belongs to the Quantum Computing Company Quantinuum with a count of 50.

IBM is implementing a novel method for merging physical qubits into logical qubits via LDPC codes. This marks a significant shift from previous methods employed in other superconducting quantum computers. Gambetta notes that utilizing LDPC codes was once seen as a “pipe dream,” but his team has now realized crucial details to make it feasible.

The benefit of this somewhat unconventional technique is that each logical qubit created with an LDPC approach requires fewer physical qubits compared to competing strategies. Consequently, they are smaller and faster error correction becomes achievable.

“IBM has consistently set ambitious goals and accomplished significant milestones over the years,” states Stephen Bartlett from the University of Sydney. “They have achieved notable innovations and improvements in the last five years, and this represents a genuine breakthrough.” He points out that both the distant qubits and the new hardware for connecting the logical qubit codes deviate from the well-performing devices IBM previously developed, necessitating extensive testing. “It looks promising, but it also requires a leap of faith,” Bartlett adds.

Matthew Otten from the University of Wisconsin-Madison mentions that LDPC codes have only been seriously explored in recent years, and IBM’s roadmap clarifies how it functions. He emphasizes its importance as it helps researchers pinpoint potential bottlenecks and trade-offs. For example, he notes that Starling may operate slower than current superconducting quantum computers.

At its intended scale, the device could address challenges relevant to sectors such as pharmaceuticals. Here, simulations of small molecules or proteins on quantum computers like Starling could replace costly and cumbersome experimental steps in drug development, Otten explains.

IBM isn’t the only contender in the quantum computing sector planning significant advancements. For instance, Quantinuum and Psiquantum have also announced their intentions to develop fault-tolerant utility-scale machines by 2029 and 2027, respectively.