As artificial intelligence technology becomes increasingly prevalent, enterprises in the Asia-Pacific region are facing the challenge of rising inference costs. To seek more economical and efficient solutions, more and more companies are beginning to migrate their AI infrastructure to the edge.
1
Advantages of Edge Computing
Cost Reduction: By moving the inference process closer to the user at the edge, companies can significantly reduce data transmission costs, especially in compute-intensive applications like image generation. For example, businesses in India and Vietnam incur much lower costs when deploying image generation models at the edge compared to centralized cloud solutions.
Reduced Latency: Edge inference can shorten the distance data needs to travel, allowing models to respond more quickly, which is crucial for industries that are highly sensitive to latency.
Increased Efficiency: Distributing inference across multiple sites can avoid routing massive amounts of data between large cloud centers, thereby improving overall efficiency.
2
Industry Applications
The industries with the strongest demand for edge inference include retail, e-commerce, and finance, all of which have high requirements for revenue, security, and user engagement.
3
Challenges and Responses
As inference moves to the edge, companies need to adopt new operational management approaches and adapt to a more decentralized AI lifecycle, including updating models across multiple sites. Additionally, data governance and security become more complex, requiring companies to strengthen protections against API vulnerabilities, data pipelines, as well as fraud and bot attacks.
4
Collaboration and Development
To address the growing AI workloads, cloud service providers and GPU manufacturers are strengthening collaborations. The partnership between Akamai and NVIDIA aims to build an “AI delivery network” that distributes inference across multiple sites rather than concentrating it in a few regions. This collaboration will help drive the development and application of edge AI technologies.