By CIO Insider Team
 

Lenovo has announced its new artificial intelligence (AI) infrastructure initiative, the Wanquan Hybrid Token Factory (WHTF) in China aimed at addressing the growing demand for AI computing services. The company plans to establish its first facility in Zhangjiakou, Hebei province, integrating high-performance computing infrastructure, AI models and cloud services to support enterprise AI adoption.

The facility aims to offer businesses scalable and cost-effective computing solutions that support their generative AI applications, high-volume requests, and heavy workloads. The hybrid computing model's goal is to make AI infrastructure more accessible without enterprises having to invest in and maintain their own large-scale systems.

Key Highlights

• Lenovo Introduces Hybrid AI Infrastructure to Support Enterprise Computing
• Zhangjiakou Facility Targets Initial Computing Capacity of 15,000 Petaflops
• Hybrid Computing Model Aims to Reduce AI Processing Costs

The first phase of the Zhangjiakou facility will feature a computing cluster of 15,000 petaflops, as well as an intelligent computing centre powering about 100 megawatts. Upon launch, the facility will be able to process over 200 million tokens per second and over a quadrillion tokens each year. It will also be able to handle up to 100,000 concurrent users.

Also Read: Lenovo to Invest $ 1 Billion in AI Infrastructure

The infrastructure will be linked to the Lenovo xCloud intelligent cloud platform as the central management system for private and public clouds that allocate computing capacity based on workload demands. The integration will enable enterprises to use computing resources as they need; helping them to control their infrastructure costs and facilitates AI application deployment.

The launch follows a surge in the use of AI in China, especially during AI inference workloads. AI inference is the action of a trained AI model that is used to generate answers, content, and complete user requests. The need for computing power to power these interactions is steadily growing as more businesses and individuals utilize generative AI applications.

China's Centre for Information Industry Development reported the use of AI tokens has reached more than 175 trillion a day in the country. The volume continues to grow and is straining the computing infrastructure and represents a challenge for enterprises with respect to computing power, availability and expenses.

Also Read: Ambitious AI Innovations Lenovo Unveiled at Tech World 2024

Lenovo's hybrid computing solution will save over 70% in processing the one million tokens. The company is also aiming to achieve an output of the token exceeding 99.9% and a power usage effectiveness (PUE) ratio of less than 1.1 for its data centre operations. These targets highlight the project's focus on computing efficiency, infrastructure reliability and energy management.

The company also anticipates that by 2030, the facility's main results will be over ten times better.
The development highlights the rising prominence of AI data centres, high-performance computing, and cloud infrastructure within enterprise computing.