Why quantum has been "10 years away" for 30 years | E2316

Quantum computing has been described as five to ten years away for roughly 30 years, but the field may now be approaching a genuine inflection point. The core misconception is that quantum computers are faster general-purpose machines. They are not. They are suited only to specific problem categories such as RSA encryption and combinatorics, and will function as accelerators in hybrid systems alongside GPUs and CPUs rather than replacing classical computers.

The qubit numbers tell a convergence story. Ten years ago working systems had one to ten qubits. Five years ago researchers believed one million qubits were needed to solve meaningful problems. More recent estimates have reduced that requirement to around 10,000 qubits, roughly one percent of earlier projections, while a US university has already demonstrated a system with 6,000 physical qubits. The required qubit count and the available physical qubit count are predicted to converge around 2030. Current qubits still produce errors frequently, sometimes every hundred or every thousand operations, and quantum error correction is the primary approach being pursued to make systems robust. AI is already being applied to this problem, with transformer models used to identify and estimate errors during computation.

The quantum hardware landscape is in an early competitive phase with multiple physical approaches being tested simultaneously. Google and IBM use superconducting qubits that require chips cooled in large refrigerators at extremely low temperatures, with a single superconducting qubit costing close to one million dollars to produce. Other approaches use trapped ions, photons, or semiconductors. Yaquma, a Kyoto-based startup, uses neutral atoms arranged in arrays inside a vacuum chamber, with one atom serving as one qubit. A company called Yakumo is working with new materials it claims could reduce qubit costs by 100 to 1000 times compared to superconducting qubits. Toshiba's SQBM quantum system costs approximately 90 dollars per hour to access, while IBM's quantum computing rental runs around 3,000 dollars per hour.

Yaquma is a spin-off from Kyoto University and the Institute for Molecular Science in Aichi Prefecture. Professor Yoshiro Takashi of Kyoto University has studied neutral atoms, specifically ytterbium, for more than 30 years and is credited with discovering 60 to 70 percent of the fundamental scientific knowledge about that atomic species. Professor Kenji Omori, head of IMS, co-founded the company alongside Takashi after the two chose to merge their efforts rather than compete. The company was launched approximately 15 months before the recording, has close to 70 employees, and was hiring at roughly one to two people per week. Approximately 20 to 30 percent of the technical staff are non-Japanese, and two of the three division heads are non-Japanese, recruited from Sydney and Turkey respectively. Yaquma has raised over 10 million dollars including grants and venture capital, with 18 months of runway following a seed extension round. Investors include Kyoto ICAP, All Night Ventures marking its first investment in Japan, and Connation, a French quantum hardware venture firm also making its first investment in Japan.

Yaquma's strategic model is to assemble components from suppliers and sell a complete quantum computer, analogous to how Toyota integrates parts to produce a finished vehicle. The most critical component for a neutral atom system is a high-fidelity, low-noise laser. Key lasers are sourced from Hamamatsu Photonics in Japan and its subsidiary NKT Photonics in Copenhagen. The three parties have signed a memorandum of understanding to develop specialized devices for neutral atom quantum computers. Transferring IP from Kyoto University to the startup was described as a tough negotiation, reflecting a broader pattern in Japan where university-to-startup IP transfer lacks an established methodology compared to overseas institutions.

On cryptographic risk, recent literature suggests quantum computation could threaten RSA encryption within five to ten years, though it will not happen within the next two years. Some nation-states are already collecting RSA-encrypted data now with the intent to decrypt it later once capable quantum systems exist, a strategy referred to as harvest now decrypt later. The US government has stated a goal of having a useful quantum computer by 2028, and China, the United States, and Japan are described as the most aggressive countries in quantum development. Most major payment networks including Visa and Mastercard use AES-256 rather than RSA, which changes the threat profile. Expanding AES key sizes from 256 bits to 2048 or 4096 bits could buy meaningful time against future quantum attacks. At around 40 to 50 qubits, quantum computers already consume less energy than classical computers for particular tasks.

Cryptography is framed as only one application, and quantum chemistry is described as a larger opportunity. Specific possibilities raised include virtual cell models that could fundamentally transform biology and medicine. Yakumo plans to sell full-stack quantum computers within five years, initially targeting academic research, and believes that after 2030 it can produce fault-tolerant quantum computers suitable for data centers, eventually offered as a service in a model similar to SaaS.