Summary:
1. The growth of AI is driving demand for neocloud services that support GPU-intensive workloads.
2. Neocloud providers offer specialized GPUaaS for AI, machine learning, blockchain, gaming, and scientific modeling.
3. Neoclouds deliver high-performance GPUs, tailored solutions, and cost savings compared to traditional hyperscalers.
Article:
The expansion of artificial intelligence (AI) technologies is transforming the data center market, leading to a surge in the need for neocloud services that cater to the demands of GPU-intensive workloads. A recent study conducted by JLL indicates that the global neocloud segment is projected to experience an impressive 82% compound annual growth rate (CAGR) between 2021 and 2025 as enterprises seek to secure GPU capacity for their AI projects.
Neoclouds are specialized cloud providers that offer GPU-as-a-Service (GPUaaS) to businesses, specifically designed to support compute-intensive tasks such as AI training, machine learning inference, blockchain operations, gaming, and scientific simulations. These providers grant on-demand access to graphics processing units (GPUs) and customized service models that traditional hyperscalers may struggle to deliver promptly.
Muhd Syafiq, the director of JLL’s data center research in the Asia-Pacific region, highlights that neoclouds offer flexible and cost-effective access to high-performance GPUs, particularly for AI, scientific, and blockchain applications. Unlike conventional hyperscalers, neoclouds focus solely on meeting the high-density requirements of AI workloads, providing rapid deployment, bespoke solutions for AI projects, and competitive pricing.
Alexander Harrowell, a principal analyst specializing in advanced computing at Omdia, notes that neocloud providers are disrupting the industry by addressing supply and demand imbalances. These providers attract significant investment, often from GPU vendors or AI laboratories, and are seen as a form of vendor financing for GPU technologies. They play a crucial role in offering infrastructure optimized for high-power density and GPU-centric workloads, such as Nvidia NVL72 rack-scale systems.
The rise of neocloud data centers is driven by the escalating demand for AI training and inference infrastructure, emphasizing the need for specialized cooling and structural capabilities to support powerful GPU hardware. Neocloud providers can swiftly establish new sites to meet urgent demands, offering a cost-effective alternative to hyperscalers, which typically require years to expand their capacity. While neocloud adoption is on the rise, these specialized providers are expected to complement traditional hyperscalers by focusing on AI-intensive workloads, while hyperscalers continue to cater to a broader range of compute and storage services preferred by many enterprises.
The specialized infrastructure of neoclouds plays a key role in driving cost efficiency, providing extreme processing power required for intensive AI projects at a more affordable rate compared to general-purpose setups. By offering scalable and dedicated GPU power, neoclouds enable AI teams to access specialized resources swiftly, enhancing performance levels. Neoclouds generally offer lower prices and more flexible contracts than major hyperscalers, with potential cost reductions of up to 66% for GPU instances. Their contracts typically range from two to five years, providing greater agility compared to the longer lease terms associated with traditional data center clients.
In conclusion, neocloud providers are reshaping the data center landscape by offering tailored solutions for GPU-intensive workloads, cost savings, and flexible terms, making them an attractive option for startups, research teams, and enterprises seeking to optimize their AI projects efficiently.
