AI challenges for wholesale connectivity providers
Today, wholesale connectivity providers typically deliver services of around 1-100Gbps for most enterprise customers, depending on the application, with 400Gbps services mainly reserved for data centres, hyperscalers and ISPs.
What drives bandwidth and connectivity?
As AI becomes mainstream, providers will need to shift increasingly to higher bandwidth, specialist connectivity services to meet the following challenges:
IoT and data aggregation
With the rise of IoT and 5G, IoT devices, from autonomous cars to smart cities, generate increasingly large volumes of data. Data is now being created at the edge, requiring a more complex mix of high capacity connectivity options to aggregate and process it efficiently.
AI training and inference
Training AI models, the initial stage of development of an LLM, requires extremely large datasets to be moved into the cloud.
AI inference, when a model produces its own conclusions, require vast though varying bandwidths.
Flexible, high-speed connectivity up to 400Gbps and beyond will be vital to handle this growing demand.
Data centre backhaul
As new core and edge data centres come online, there is a growing need for increased backhaul capacity to transport data efficiently between sites. Data centre interconnects (DCI) are increasingly recognised as one of the key applications requiring high capacity connectivity.
