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

Episode Summary

Executive Summary: This episode features Kazu Nakashogi of Yakumo, a Kyoto-based quantum hardware startup, discussing the state of quantum computing, why it remains hard, and how neutral-atom QPUs may accelerate commercialization. The conversation covers hardware tradeoffs, university spinouts, fundraising, supply chains, AI-assisted error correction, and the dual-use/security implications of quantum breakthroughs.

Main Topics: Quantum computing's commercialization timeline: Kazu argues quantum computing is moving closer to practical use, with useful systems potentially arriving around 2030 and earlier academic installations already deployed in Japan. Hardware approach: neutral atoms vs other QPU architectures: The discussion compares superconducting qubits, ions, semiconductors, and Yakumo's neutral-atom approach, emphasizing cost, scalability, and industrialization. Error correction and fragility of quantum systems: Kazu explains that qubits are error-prone and that quantum error correction (QEC), supported by classical computing and AI, is essential for scalable systems. University spinout and Japan's deep-tech startup model: Yakumo is presented as a university-originated deep-tech startup, with Kazu describing IP transfer, negotiation, and the relatively nascent Japanese spinout ecosystem. Fundraising, runway, and international investors: The company has raised over $10M through grants and venture backing, including first-in-Japan investments from U.S. and French investors, with ~18 months of runway. AI and quantum synergy: The episode explores 'AI for quantum' (error estimation/correction) and 'quantum for AI' (future use of quantum outputs to improve models), though the latter is still mostly aspirational. Security and dual-use risks: The conversation ends on the risk that future quantum machines could threaten RSA and enable 'harvest now, decrypt later' strategies, prompting government scrutiny and careful disclosure.

Key Arguments: Quantum computing is not a faster replacement for CPUs/GPUs; it is useful for specific classes of problems such as combinatorics and cryptography. Practical quantum use is closer than many expect because software and algorithms have improved, reducing required qubit counts from ~1 million to around 10,000 for some tasks. Neutral-atom hardware may be more economical than superconducting qubits, potentially making large-scale systems more viable. Quantum error correction is the central technical challenge; without it, qubit errors make computation unreliable. AI already helps quantum computing by estimating and correcting errors, while quantum’s impact on AI will likely come later. Yakumo aims to be an integrated 'Toyota-like' quantum computer company, assembling the full system rather than only one component. University IP commercialization in Japan is still developing, and spinouts often require tough negotiations with universities over equity or options. Quantum computing is a dual-use technology, so founders must engage carefully with governments and security concerns. Commercial business models may evolve in phases: academic systems first, then data-center sales, then quantum-as-a-service/entropy-as-a-service. Quantum may also have energy advantages over classical computation for certain workloads, potentially lowering CO2 and power consumption.

Data Points: Company age: 15 months - Yakumo was launched about 15 months before the interview. Team size: Close to 70 employees - Kazu says the startup has grown rapidly, mostly on the technical side. Technical hiring ratio: 1 to 2 hires per week - Describes ongoing rapid expansion and hiring pressure. Non-Japanese staff: 20% to 30% - Share of employees who are non-Japanese. Non-Japanese division heads: 2 of 3 - Two of the three heads under the CTO organization are non-Japanese. Runway: 18 months - Expected runway after completing the seed extension round. Total capital raised: Over $10 million - Kazu mentions the company has raised more than $10M through grants and venture funding. Seed extension investor: All Night Ventures - U.S. firm that made its first investment in Japan. Institutional support: Kyoto ICAP - University-linked entity that helped connect Yakumo to foreign investors. First investment in Japan: All Night Ventures and Corn Nation - Both investors are described as making their first investment into Japan. University demonstration: 6,000 physical qubits - Cited as a recent demonstration by a U.S. university. Earlier expected qubit requirement: 1 million qubits - Five years ago, people thought some problems required this scale. Updated qubit requirement: 10,000 qubits - Kazu says software and algorithm advances have sharply reduced the estimate. Historical qubit scale: 1 to 10 qubits - What was available roughly 10 years ago. Energy crossover point: 40 to 50 qubits - Referenced paper suggests quantum may use less energy than classical methods at this scale for some problems. Quantum computer cost: ~$1 million per physical qubit - Cited from a colleague working on superconducting qubits. Alternative cost reduction: 100x to 1,000x cheaper - Yakumo believes neutral atoms could dramatically reduce qubit costs. Commercial timeline: Around 2030 - Kazu expects useful fault-tolerant quantum computers to emerge around then. Government target: 2028 - He references the U.S. wanting a useful quantum computer by this year. Ski resort timing anecdote: Twice a year - Jason says he visits Japan twice yearly for Foundry University and skiing.

Pivotal Quotes: "Hardware itself. Yes." — Kazu Nakashogi: He explains why quantum computing has taken so long: the hardware is extremely difficult to build. "We can only apply those kinds of problems to quantum computer or quantum calculations." — Kazu Nakashogi: He corrects the misconception that quantum computers simply speed up everything; they are specialized for certain problem classes. "We believe that we reach to like two of them meet each other's close to 2030." — Kazu Nakashogi: His estimate for when hardware capability and problem requirements may align for practical use.

Implications: Quantum is moving from pure research toward early deployment, but only for narrow, high-value tasks. The winners may be integrated hardware companies with strong university, government, and supply-chain ties, while security and encryption migration become urgent.