Riyadh Calling
A question we often get is when will the true impact of AI be of tangible benefit to everyday people living their lives, doing their jobs, and pursuing activities that interest them. In our estimation we are still in an extremely exaggerated market. AI is the big theme in terms of the data center investment driver du jour – without question. But can you afford to wait, and see before investing?
Every technology company on the planet has overpromised benefits and either completely ignored or severely underestimated the costs of AI. This is causing extreme ‘fishing expedition’ activities all over the world – options on real estate, contracts for increased power generation, design proposals for networking infrastructure – distorting the data center spending reality. Due diligence is time consuming and costs real money to rationalize your plan.
Case in point, earlier this week the Saudis struck a deal with the Trump administration for a $20B investment in Data Center development. Yet, many people are amazed at a recent analysis of actual generative AI Use by Marc Zao-Sanders at the Harvard Business Review. In this summation, Therapy and Companionship top the list of use cases followed closely by Organize Life and Find Purpose. Even Sam Altman, AI Chief Pitch Person, admits publicly that these patterns are a reality for ChatGPT. To further substantiate what is happening, we can look to places like Japan where a glut of older people is extreme – seniors in Tokyo are talking with companionship robots for several years now. This activity certainly has value for these older folks, but at current benefit levels, is it worth a billion, tens of billions, or even a trillion dollars in fresh investment?
Extraordinarily, Enhance Learning and Generate Code comes in at fourth and fifth place, respectively. One would like to be optimistic about the future benefits, but dystopian doubts about living in harmony in communities of likeminded people that are suddenly free to create and express themselves like never before are ever present when one examines the observable behavior patterns. Just look at what happens at the average dinner table in the good ole USA. There is a nice family of four, all sitting down for a home cooked meal fresh out of the microwave, all looking at different screens on their mobile devices and interacting with people or bots not even physically present. Generative AI has the potential to make this pattern more severe and not less. As a civilization, is that what we want?
Generative AI, The Internet of AI, the ‘agentic enterprise,’ pick your flavor, will take about seven to ten years to really weave into key areas of the fabric of life. It is not a fixed in concrete destination. We have the opportunity to correct the course for how AI is applied, and we should intervene where it makes sense. Young people typically adopt first; they are also the most vulnerable. They have all tried it, but if you ask, aspiring scholars will readily tell you how bad AI can be at helping with their studies – especially STEM (Science Technology Engineering and Math for the artists in the room) oriented problems.
You can expect to start seeing marginal improvement in business process and productivity in the next three – four years. We can derive this timeline from corporations like IBM. At Think 2025, the Consulting group stipulated that the average AI project is approximately 40 months in duration from a professional services standpoint. That is not saying much about the actual realization of business value across the timeline, but it is a kind of measuring stick for how long behavioral transition really takes at least for large enterprises.
Application adoption and the promise of business value drives data center investment. The best Data Center design and planning exercises focus on this fact. Organizations that optimize their capital for sporadic and sometimes ‘lumpy’ adoption with surge protection that enables them to pounce on new opportunities are in the best position for success.