At Computex 2025 in Taipei, NVIDIA delved into its stronghold on the AI and quantum computing capabilities, mentioning not only existing and upgraded solutions but also partnerships that will be essential to merge the hardware and software chain.
In his keynote, CEO Jensen Huang introduced the NVLink Fusion system, designed to enhance chip-to-chip communication and facilitate the development and deployment of AI tools.
This technology integrates third-party CPUs and AI chips into NVIDIA’s server platforms, shifting from its traditional full-stack AI solutions.
In addition to hardware advancements, NVIDIA announced plans to invest in Taiwan’s tech infrastructure, announcing an AI supercomputer in the country. These developments underscore the giant’s commitment to fostering global AI infrastructure, with Huang calling it an industry worth ‘Trillions of Dollars’.
The company is actively intensifying its efforts in quantum computing and AI integration. It is reportedly in advanced talks to invest in PsiQuantum, a US-based startup aiming to build a fault-tolerant quantum computer using photonic qubits and standard semiconductor manufacturing processes.
The deal would support PsiQuantum’s $750 million funding round and mark NVIDIA’s deeper push into quantum computing despite Huang’s earlier scepticism regarding the future of quantum. This move aligns with NVIDIA’s broader strategy to bridge quantum computing and AI, as demonstrated by its recent collaborations and infrastructure developments.
Taiwan’s Supercomputing Leap
In Taiwan, NVIDIA is collaborating with local manufacturers and the National Centre for High-Performance Computing (NCHC) to advance quantum research. A new AI supercomputer, built by ASUS, will feature over 1,700 NVIDIA GPUs, including the latest Blackwell Ultra systems, interconnected with NVIDIA Quantum InfiniBand networking. This cutting-edge infrastructure supports projects in climate science, quantum research, and large language model (LLM) development.
A key focus is the development of sovereign AI applications. The system will power platforms like Taiwan AI RAP, enabling the rapid creation of culturally relevant AI tools. One flagship effort, the Trustworthy AI Dialogue Engine (TAIDE), is building Taiwanese LLMs for natural language processing, translation, and customer service, which is already impacting sectors like education and healthcare.
For climate research, the NCHC is using NVIDIA’s Earth-2 platform and AI models such as CorrDiff and GraphCast (Google DeepMind) to enhance weather forecasting and climate simulations. The supercomputer’s power will significantly accelerate training and inference workloads in this domain.
On the quantum front, researchers are using NVIDIA’s CUDA-Q and cuQuantum to drive breakthroughs. Projects like the Quantum Molecular Generator and cuTN-QSVM have enabled simulations with up to 784 qubits, pushing the limits of hybrid quantum-classical computing.
Enter Foxconn and Cadence
Cadence Design Systems recently unveiled the Millennium M2000 supercomputer, powered by NVIDIA’s Blackwell GPUs. The company claims the system delivers up to 80 times higher performance and 20 times greater energy efficiency than previous CPU-based platforms, significantly reducing simulation times across industries.
The M2000 integrates NVIDIA HGX B200 systems and RTX PRO 6000 Server Edition GPUs with Cadence’s computational software and CUDA-X libraries, enabling complex simulations such as electronic design automation, system design, and molecular modelling.
Notably, engineers can now complete chip-level power integrity simulations in a single day, compared to the two weeks needed with traditional CPU clusters. NVIDIA plans to utilise 10 M2000 units to aid in developing next-generation chips and AI data-centre infrastructure.
In parallel, Foxconn, in collaboration with NVIDIA and the Taiwanese government, is constructing an AI factory supercomputer equipped with 10,000 NVIDIA Blackwell GPUs. Managed by Foxconn’s subsidiary, Big Innovation Company, this facility aims to provide AI cloud computing resources to researchers and enterprises, accelerating AI development across sectors.
The Taiwan National Science and Technology Council will invest in this supercomputer to support AI development and adoption across industries. TSMC plans to leverage this infrastructure to enhance its research and development capabilities.
Global Quantum-AI Collaborations
Beyond Taiwan, NVIDIA is advancing global initiatives to integrate quantum computing with AI. In Japan, the ABCI-Q supercomputer, developed by the National Institute of Advanced Industrial Science and Technology (AIST), is a significant step in this direction.
ABCI-Q is powered by over 2,000 NVIDIA H100 Tensor Core GPUs across over 500 nodes, interconnected via NVIDIA Quantum-2 InfiniBand. This infrastructure supports high-fidelity quantum simulations and is integrated with NVIDIA’s open-source CUDA-Q platform, facilitating hybrid quantum-classical computing.
ABCI-Q also incorporates diverse quantum processors, including Fujitsu’s superconducting qubit processor, QuEra Computing’s neutral atom quantum processor, and OptQC’s photonic processor. This integration enables hybrid quantum-GPU workloads across multiple qubit modalities, accelerating research in quantum error correction and application development.
In addition to quantum advancements, NVIDIA is expanding its hardware portfolio by introducing the Grace CPU C1. This single-socket, high-performance server platform is optimised for edge, telco, storage, and cloud deployments and delivers up to twice the energy efficiency of traditional CPUs.
The Grace CPU C1 is gaining traction among leading manufacturers, including Foxconn, Jabil, Lanner, MiTAC Computing, Supermicro, and Quanta Cloud Technology, which are developing systems that leverage its capabilities.
Not just this, Foxconn, in collaboration with NVIDIA, is also advancing innovative hospital solutions to address the global nursing shortage. Their Nurabot, a collaborative nursing robot, assists with tasks like transporting medication, reducing nurses’ workload by up to 30%.
Utilising NVIDIA’s AI and digital twin technologies, Foxconn is transforming healthcare delivery in Taiwan.
Strategic investments and collaborations underscore the company’s commitment to advancing quantum computing and AI. By developing robust infrastructure and supporting research initiatives, NVIDIA aims to accelerate breakthroughs in various scientific and industrial domains.
