Thank you to everyone who participated in the Quantum Computing EXPO. This year, quantum computing attracted an unprecedented number of attendees, and I believe we set a new record for visitors. We prepared 2,000 flyers, all of which were distributed. On the final day, we ran out of flyers by noon due to the high demand. Although it's not entirely clear why quantum computing received so much attention, it’s evident that significant changes are on the horizon, suggesting a transformative phase for the industry.
Popular Products at the Quantum Computing EXPO
Among the most popular products were the RAG system and semiconductor quantum computers.
RAG System:
Our company has been leveraging NVIDIA GPUs for quantum computing simulations through a system called cuQuantum. Utilizing our extensive GPU resources, we developed services tailored for generative AI, resulting in the RAG system.
The key feature of our RAG system is its affordability, offered at just 5,500 yen per month. There is a growing demand for cost-effective, on-demand services rather than continuous use.
We also showcased a combination of local LLMs and RAG, allowing for customization. Our online cloud development environment is available at 20,000 to 50,000 yen per month, with on-premises solutions costing several million yen depending on the machine specifications.
The RAG system was particularly popular for various applications, including government manuals and corporate onboarding training.
Semiconductor Quantum Computers:
Although not yet released, our semiconductor quantum computer prototypes garnered significant interest, particularly regarding the chips and cryogenic equipment.
Semiconductor quantum computers represent the latest approach in quantum computing, enabling the creation of numerous quantum bits (qubits) and quantum chips at a lower cost. For instance, Intel released a 12-qubit system in 2023 and is now working on a 100-qubit system. These systems can house millions to billions of qubits on a single chip due to the small transistor size.
Control and readout circuits can also be created using CMOS technology, allowing for SoC and chiplet integration. This eliminates the need for room-temperature control devices, reducing communication overhead and improving error correction speeds.
Additional Highlights:
- The cryogenic temperatures of 1-3K facilitate compact designs.
- Current semiconductor fabrication technologies enable the production of these quantum computers, which have rapidly progressed from 2-3 qubits to thousands, thanks to advancements in semiconductor technology.
- Domestically, development focuses on scaling from single-digit qubits to tens and hundreds.
- Emerging microwave-free silicon technology promises further error reduction and improved operability.
Moreover, we have developed a domestically produced cryogenic system dedicated to quantum computers, eliminating reliance on imported equipment and enabling local chip and system development.
Other Popular Attractions:
1. Quantum Seminar Platform:
A platform for easily learning about quantum computing and AI.
2. Tensor Networks:
Our cloud-based quantum machine learning solutions gained significant interest, particularly for the lightweight and fast deep learning capabilities using tensor networks. These were especially popular for edge applications like robots and satellites, as well as for compressing large models like LLMs.
The success story from Hakuhodo DY Holdings, showcasing the lightweight and fast 3D deep learning with NeRF, was particularly well-received.
Quantum computing and AI continue to experience high demand. The lecture on FTQC by QuEra was also a major highlight, generating numerous questions and discussions. We look forward to seeing you again at our exhibits in the fall and next spring. Thank you!