IBM has achieved a significant breakthrough in scaling quantum computing by successfully combining and cooling two modular cryogenic complexes. This engineering effort has created a single ultra-cold environment that will serve as the foundation for future fault-tolerant quantum systems.
The key feature of the design is its ability to link hundreds of quantum chips within a unified, powerful infrastructure. The cooling process to a temperature of 4 K took less than five days, after which the system stabilized at a level below 15 mK. These are critically important metrics for the stable operation of qubits.
Particular attention deserves the engineering solution with vacuum enclosures. Each of them provides up to 12 times more space for wiring compared to previous generations of systems. This directly impacts the density of connections both within modules and between them, which is the main bottleneck in scaling.
The architecture involves the use of so-called "L-couplers" for direct connection of individual chips. As early as 2026, the corporation plans to install IBM Quantum Nighthawk processors into these modules for extended testing.
The roadmap looks ambitious: by 2027, IBM intends to combine several processors via L-couplers into a system capable of supporting at least 1000 programmable qubits. And by the launch of IBM Quantum Starling in 2029, the company expects to place thousands of qubits in each module. It is Starling, announced in 2025 along with a new error correction code, that will be the ultimate goal of this engineering epic.
Recall that in July, IBM CEO Arvind Krishna voiced a forecast according to which investments in quantum computing will begin to noticeably impact the company's financial performance as early as 2028–2029.
My analysis: Connecting cryogenic modules is not just a technical achievement, but a strategic signal to the market. IBM demonstrates that the path to commercially significant quantum systems lies through modularity and engineering ingenuity, not just the raw power of individual chips. If the pace is maintained, we could indeed see the first practical quantum computations with thousands of qubits by the end of the decade.