Page 14 - Connecting the Future How Connectivity and AI Unlock New Potential

Connecting the Future: How Connectivity and AI Unlock New Potential

1�2�7 Edge Computing and IoT Integration

As processing capacity and speed requirements grow, performing AI computing at consolidated
data centers and in the cloud becomes increasingly inefficient or impractical. Rather than accepting
transmission delays for these critical tasks, computing at the edge can reduce latency by allowing
data processing and analysis much closer to the data source, reducing transmission distance and
time required for data transfer. Devices such as smartphones and IoT smart sensors can also be
useful mid-points in this data transmission process. They enable both upstream applications that
filter prioritized data to centralized servers, and downstream applications that perform real-time
processing at the source.
Gartner predicts that 75% of data generated by companies will be created and processed outside
traditional data centers or cloud environments by 2025 as the computing power of smaller devices
continues to grow. This expanded market for edge computation promises to offer proportionally
33
greater impact to under-resourced markets and rural communities, where lower investment returns
on large data centers may hinder construction. By limiting data processing at the source rather
than consolidating it in data lakes, edge devices can reduce the investment costs of traditional data
infrastructure. The shortened journey of data processing also has security benefits: it minimizes
the risks of transmitting sensitive information, such as personal health records or financial transac-
34
tions, over long distances. However, it also underscores the importance of securing dispersed
edge computing devices and network endpoints, which can become vulnerable entry points if
not properly protected.
Still, powerful miniature computing elements will likely start at high price points, so private sector
development efforts should balance affordability with capability in AI-enabled edge devices to
improve access for a range of consumers rather than maximizing pure computing power.

1�2�8 Reconfiguring Network Architecture

Modernizing network architecture prepared for an AI economy can start with the reconfiguration of
existing network designs. The current cloud environment is often costly, involving purchasing many
routes and ports connected to different locations just to achieve the interconnectivity between
carrier-neutral data centers and multiple hyperscalers. This operation inevitably drives up the costs
and increases network latency. Some global communication and Internet service providers have
begun moving away from this classical design by upgrading their equipment and creating a direct

8