This year is the 'International Year of Quantum Science and Technology (IYQ)' designated by the United Nations (UN). It commemorates the 100th anniversary of German physicist Werner Heisenberg's presentation of 'matrix mechanics,' which mathematically expressed quantum mechanics in 1925. Google was the first to widely announce the International Year of Quantum Science and Technology. Last year, Google drew significant attention by announcing a new quantum chip named 'Willow.'
The leading corporations in the global quantum computer market are IBM and Google. Since Google began its quantum computer research in earnest in 2019, the two corporations have been fiercely competing for dominance. Both corporations have set a goal to commercialize quantum computers within the next decade. Recently, Jensen Huang, CEO of NVIDIA, remarked that it would take 20 years to commercialize quantum computers, but both Google and IBM aim to achieve this within 10 years.
While both corporations share the same commercialization goal, their strategies are significantly different. IBM is focusing on expanding qubit capabilities to maximize the computational power of quantum computers. In contrast, Google is concentrating on minimizing quantum errors that inevitably occur during quantum computing to enhance accuracy. It remains to be seen which of the two will ultimately emerge as the winner.
IBM announces a quantum computer roadmap every year, focusing on expanding qubits. According to the recently released roadmap, IBM developed the 27-qubit quantum computer chip 'Falcon' in 2019, followed by the 65-qubit 'Hummingbird' in 2020, the 127-qubit 'Eagle' in 2021, the 433-qubit 'Osprey' in 2022, and the 1,121-qubit 'Condor' in 2023. From this year, it plans to work on the 4,158-qubit 'Kookaburra.' Considering that South Korea aims to develop a 1,000-qubit-level quantum computer by 2032, IBM's technological capabilities are already well ahead of its competitors.
The products released by IBM continue to see performance improvements. The 27-qubit quantum computer chip 'Falcon' was launched in 2020, followed by the 127-qubit chip in 2022, and last year the 133-qubit 'Heron' was unveiled. This year, an announcement for the 1,092-qubit 'Flamingo' is anticipated.
Conventional computers process information based on bits divided into 0 and 1 according to the presence or absence of electrons. In contrast, the unit of a quantum computer is a qubit, which represents both 0 and 1 states in a superposition. Utilizing this characteristic, quantum computers can perform multiple operations simultaneously. The number of qubits is considered a crucial factor that determines the computational performance of quantum computers.
Han Sang-wook, a principal researcher at the Korea Institute of Science and Technology (KIST), noted, "Having many qubits means being able to process more variables simultaneously. For instance, if there are 1,000 qubits when predicting the weather, it can calculate the impact of 1,000 variables; if there are 10,000 qubits, it can consider the effects of 10,000 variables, thereby improving computational accuracy."
IBM's strategy is to achieve a technological advantage by significantly increasing the number of qubits. It is estimated that around 1,000 qubits are needed for quantum computers to surpass the performance of existing computers. However, as the number of qubits increases, the accuracy of calculations tends to decline.
Google is focusing on minimizing quantum errors rather than increasing qubit count. Quantum errors refer to computational inaccuracies caused by the instability of qubits. The scientific community estimates that the current quantum errors in existing quantum computers are between 0.1% and 1%. Assuming a quantum error rate of 1%, repeating a computation 100 times results in a sharp drop in accuracy to 36%.
The scientific community has long evaluated that correcting quantum errors, known as 'quantum error correction,' may be practically impossible. However, Google confirmed the feasibility of quantum error correction upon announcing the quantum computer chip 'Willow' on the 9th of last month. Google has previously conducted research to find ways to resolve quantum errors.
On the 2023 international journal 'Nature,' Google published a technology for correcting quantum errors using multiple qubits simultaneously. Google's strategy involves performing a single computation using a 'logical qubit' composed of multiple qubits instead of a single qubit. This approach detects quantum errors that occur during computations using several qubits and corrects them. Two months prior to the announcement of Willow, in October of last year, Google demonstrated surpassing supercomputer computing power with a 67-qubit quantum computer. Google referred to this stage as 'weak noise' and announced its success in enhancing the stability of qubits.
IBM has countered that Google's quantum error correction technology is a method that cannot be implemented without a sufficient number of qubits. Jay Gambetta, head of IBM Quantum Computing, commented, "To perform practical calculations using Google's error correction method, it would require billions of qubits."
However, the scientific community believes that for quantum computers to be commercialized, both the expansion of qubit scale and the establishment of quantum error correction must be achieved simultaneously. In fact, IBM has been actively researching quantum error correction since last year, while Google has consistently adhered to increasing its qubit scale.
Han Sang-wook noted, "While the technologies that both corporations use as marketing tools are different, ultimately, they are conducting research aimed at improving both qubit scale and quantum errors. In the commercialization phase, both corporations will likely need to secure technologies for expanding qubits and correcting quantum errors."
South Korea is also developing its own quantum error correction technology to secure the foundational technologies needed for the commercialization of quantum computers. Lee Seung-woo, head researcher at KIST's quantum technology research group, has developed a method of quantum error correction that does not accumulate errors even as qubit size increases. Unlike Google and IBM, which are developing superconducting quantum computers, the KIST research team is working on a photon-based quantum computer. The maximum optical loss threshold of the leading quantum computer company in the United States, PsiQuantum, is around 2.7%. The KIST research team has achieved a maximum threshold of up to 14%.
Lee Seung-woo stated, "Just like semiconductor chip design technology, the architecture of quantum computing is also important. Even with 1,000 qubits, if the structure for error correction isn't there, it will be difficult to perform even one unit of logical qubit operation. Although practical application of quantum computing still requires time, this research has contributed to slightly hastening that timeline."