AI demands bandwidth
Over two-thirds of current AI users expect to quadruple their AI bandwidth within 18 months.
Let’s get physical: why the network is not one-size-fits-all
Let’s get physical: why the network is not one-size-fits-all
According to Omdia, 80% of enterprises already run at least one AI application, with most accessing a public AI model. This requires connectivity – and not just any connectivity will do. AI’s ability to reshape, empower and disrupt industries depends greatly on network performance.
In addition to primary AI use, 1/3 of global network traffic has become AI “enhanced”. Increasingly, AI is summarizing meetings, suggesting next-best actions, detecting trends, and creating content in real or near real time. An important consequence of retrofitting AI features into common business applications is the need for greater network reliability, lower latency and higher speed.
A 2025 Omdia survey of 419 large global enterprises found that greater reliability is now the number one priority driving enterprise networking decisions. Because internet performance can be uncertain in some regions of the world, some 60% of enterprises plan to rebalance from the public internet to business-grade connectivity options.
Key findings
Growing AI-enhanced traffic
Today – mostly without enterprises realizing it – a third of global network traffic has become AI enhanced.
Boosting bandwidth
8 out of 10 enterprises expect to double their cloud connectivity in the next 18 months, with dramatic rises in 100G connectivity planned.
Moving to business internet
60% of enterprises plan to rebalance from the public internet to business-grade options.
Read the Omdia white paper to discover the five reasons for enterprises to rethink connectivity in an AI-first world.