NVIDIA, a leader in AI computing technology, produces real-time intelligence from data through its AI Factory, providing long-term innovation and competitive advantage to businesses and countries. NVIDIA helps businesses adapt to the rapidly changing era of AI and drive success through efficient and sustainable AI factories.

NVIDIA announced its 'AI Factory' at the GPU Technology Conference (GTC) held in San Jose, USA, and said it is leading the innovation of data centers for the next-generation AI era.

AI Factory presents a new concept that goes beyond the role of traditional data centers through large-scale data processing and real-time insights.

AI factories aim to produce large-scale intelligence by expanding the role of existing data centers that store and process data.

This will enable companies and countries to secure AI-based economic competitiveness, and will orchestrate the entire AI life cycle from data collection to training and fine-tuning to inference.

In particular, intelligence produced in AI factories leads to real-time prediction and automation, significantly increasing the speed of value creation.

AI factories can be integrated with or evolved from existing data centers, providing a customized approach based on a company’s business model.

AI factories are becoming a key infrastructure in which governments and businesses around the world are investing to drive economic growth, innovation, and efficiency.

AI Factory is designed to address the rapidly growing demand for AI inference models.

NVIDIA explains how AI inference is becoming more advanced through three scaling laws: pre-training, post-training, and test time scaling, and emphasizes that the optimized design of AI factories is essential to meeting these compute demands.

Specifically, with the NVIDIA Blackwell-based GB300 NVL72 rack-scale solution, AI Factory can increase throughput by up to 50x. It achieves inference output and sets new standards in efficiency and scale.

This computing power positions the AI Factory as a platform optimized for large-scale intelligent manufacturing.

NVIDIA supports AI factories with an integrated platform that includes hardware, software, ecosystem partners, and reference architectures.

This enables enterprises to deploy cost-effective, scalable, and high-performance AI factories.

NVIDIA DGX SuperPOD is an on-premises AI factory solution that helps enterprises quickly build and operate AI projects, while the cloud-based DGX Cloud provides a platform optimized for developing and deploying large-scale AI applications.

AI factories are rapidly being built in key regions such as Europe, India, and Japan, transforming various industries. The European Union is designing seven AI factories in cooperation with 17 member states, and Yota Data Services in India has built a platform to democratize high-end GPU resources. Japan and Norway are also accelerating industrial innovation and AI adoption through NVIDIA-based AI infrastructure.