Description
There continues to be intense debate about the impact of AI on the Internet infrastructure sector. A growing number of voices, particularly in the investment community, believe there is now an ‘AI bubble’ forming that will inevitably burst. The hundreds of billions and literally trillions of dollars that is pouring into digital infrastructure, and the pace at which it is happening, has raised alarm bells. This level of investment simply cannot be sustained and is fundamentally irrational, the sceptics argue. They go on to suggest the sector is being aggressively overbuilt, with supply set to exceed the demand that we can realistically expect to be coming from a nascent and unproven new technology. Interestingly however, the bearish sentiment that has emerged is not supported by a lot of grounded argument or nuance. The sceptics point to irrational exuberance and the lack of historical precedent: ‘we have not seen this before, so therefore it will not and cannot happen’. Meanwhile, some have pointed to the circular financing deals that have NVIDIA at the centre as evidence of a bubble. The maze of capital flows raises eyebrows to be sure. But are the flows based on a lack of demand and the absence of real end user uptake? Are GPUs sitting idle and not running applications? Have infrastructure providers procured too much land and energy that is not being consumed to support real live deployments? Are there speculative and empty data centres being constructed and brought online? The sceptics do not seem to have given these questions much thought. At such an early stage of development, it seems presumptive to just assume something is fundamentally wrong given a few things that seem slightly out of line or have not been seen in the past. Could it actually be that the hyperscale data centre and cloud infrastructure being built could actually be needed? Is a next-generation technology like AI destined to fail? Why is it impossible for AI to simply carry on and accelerate the building of hyperscale infrastructure? This report takes a closer look at why the sceptics have it wrong. AI is not a bubble. While there may be some bumps in the road, what we are more likely to see is a growth wave of infrastructure development that will far outlast the hype.
