What’s really powering growth in AI

[ "Blog: Latest Insights" ]
[ ]
[ ]
  • Our Blog

Download PDF

Our equity analysts and technology investment teams have been following AI since its inception to identify key players and areas of growth. At its heart, opportunity in AI is driven by growth in the need for computing power.


In fact, four years from now, we estimate that computing demand will be nearly 10x the current level.

It’s been estimated that it may take $7 trillion in investment to support AI needs. Here are 4 key contributors:


More data centers


Our analysts have calculated that we’ll need 7x more data center square footage over the next four years to house high performance servers, storage systems and networking equipment.


Electrical content in AI data centers can be 3x-5x as high as traditional data centers, and electrical equipment companies are poised to meet the growing need for transformers, battery storage, switchgears, power distribution units and power management software.





Resources focused on power and energy


Data centers also have enormous energy requirements. We think consensus estimates for power needs are underestimated by almost 400%.






Significant demand for cooling technology


Data centers also generate a lot of heat, which will drive significant demand for cooling technology including air and liquid cooling solutions. We estimate that the liquid cooling market could increase by 5x by 2028 due to AI-driven demand.




AI will require the right people


There’s a global shortage of skilled talent who can help companies effectively implement AI tools. Consultancies that can provide these skills to companies may see significant growth.


Skills shortages prevent 70% of companies from keeping up with the pace of change.



Source: Columbia Threadneedle Investments as of 03/31/24.

Capex estimate was posited by Sam Altman, CEO of OpenAI. Computing demand is based on CAGR of PFLOPS, a measure of computing capacity that refers to one quadrillion floating point operations per second. Data center space needs are based on 1000 watts/sq ft. Staffing shortages are derived from the Nash Squared Digital Leadership Report, 2022.



Download this article as a PDF