Page 13 - Connecting the Future How Connectivity and AI Unlock New Potential

Connecting the Future: How Connectivity and AI Unlock New Potential

However, satellite communications can still play a role in advancing AI adoption through last-
26
mile connectivity to remote regions. Unlike terrestrial fiber and microwave towers, stable and
redundant satellite relays can connect edge users with IXPs or data processing centers, enabling
underserved regions to access the benefits of AI applications at an affordable cost.

Internet Exchange Points
Internet Exchange Points (IXPs) play critical roles in ensuring uninterrupted data transfers
as traffic passes between local networks. IXPs connect independent telecommunications
systems with peering data sharing agreements, allowing companies to reduce latency by
utilizing competitor-owned infrastructure. Investment in IXPs outside of data centers now
offer lower-cost options for underserved markets, enabling affordable local data transfer,
rather than travelling vast distances to reach large data centers.
Governments are crucial to the success of IXPs, which thrive when connecting local
networks to the Internet’s core. By promoting competition and creating favorable invest-
ment conditions with stable regulations and reliable energy sources, countries can attract
innovative market participants to build the necessary peering networks. 27
IXPs are critical to build strong, accessible infrastructure, especially in underserved
markets, to support widespread AI adoption.

1�2�6 Cloud Services & Data Centers

Cloud servers and their associated software and data offer efficiencies and scaling power. These
virtualized computing systems operating at large data centers offer lower cost and more reliable
data storage, processing, and exchange to multiple companies on a single server. The outsized
benefits of cloud computing contribute significantly to the acceleration of AI development. Compu-
tational power, depth and data diversity has been used for model training, data analysis, and
application employment across a variety of fields, from healthcare to financial services. 28
Without access to cloud servers, organizations in these underdeveloped markets face higher costs
of energy and server operations while lacking access to the technical benefits of scaled comput-
ing. Rectifying this divide requires both increased private sector investment in server sites and
accompanying government investment and policy support to cultivate infrastructure environments
friendly to access cloud-based computing. 29
Data centers now play critical roles in digital networks, supporting AI applications that require
greater computing support like Natural Language Processing (NLP) or computer vision. McKinsey
estimates demand for data centers to grow at an average rate of 33% per year until 2030, far
outstripping construction and integration efforts underway. 30
Several critical bottlenecks hinder the construction of data centers from meeting current and
projected data processing demands. First, new sites typically require new energy sources, as
the high energy demands at a single location could place pressure on existing energy grids with
possible negative effects on local communities. Second, permit acquisition processes for new
31
data center sites slow construction and burden companies with regulatory requirements. While
governments have begun taking encouraging steps to expand power generation, promote clean
energy at data center sites, and simplify permitting for new projects, these efforts must be accel-
erated to support the rapid scaling and performance demands of modern data centers. 32

7