A recent McKinsey survey has shed light on how leading firms are integrating artificial intelligence (AI) tools into their operations. These findings point to the promise—and peril—on the opening frontier of AI infrastructure. Businesses big and small are understandably enthused about the opportunities AI technologies can unlock. Yet, industry leaders are becoming increasingly preoccupied with the capacity-drawing limits of data centers that sustain this growth.
Satya Nadella, the CEO of Microsoft, to speak out about the lack of data center space. He argues that this is a bigger problem than semiconductor supply. He stated, “It’s not a supply issue of chips; it’s the fact that I don’t have warm shells to plug into.” This viewpoint highlights the critical importance of infrastructure that’s able to stay ahead of the rapidly growing demand for AI services.
The growing complexity of the supply chain that actually makes AI services possible further muddies predictions on future need. Data centers, which need a very long lead time for planning and building, deal with a complicated and ever-changing gauntlet of uncertainties. According to industry experts, it is difficult to ascertain how much data center space will be necessary in the coming years. And nothing will impact these needs more than the continuing evolution of user behavior and possible new technology advances.
Last week, McKinsey published their latest survey on the state of AI. It’s an indicator that the top firms are heavily investing to build out their firm-specific AI capabilities. What’s obvious is how much worse many are struggling with infrastructure limitations that might swallow their fortunes. Our survey findings underscore a growing imperative for private companies to strategically prepare for their eventual infrastructure requirements. This is particularly critical as they embrace new technologies and policies.
Meta is attempting to address these challenges directly. As an indication of their commitment to this agenda, they recently released a three-year $600 billion plan for infrastructure investment. This investment will help us get on equal footing to conduct responsible and equitable AI work. It ensures that we don’t fall behind our competitive global landscape in the fast-changing technology economy.
This is a big moment in the evolution of data center financing. An Oracle-connected data center campus in New Mexico has already landed up to $18 billion in credit from a consortium of 20 banks. This funding signal further shows the confidence that investors have in the future of AI infrastructure, even amidst all of the unknowns that exist.
Unlike stadiums or other entertainment venues data centers usually require several years of construction resulting in uncertainty as to whether they will keep pace with rapidly evolving technology. As Russell Brandom, a technology journalist with over a decade of experience in the industry, points out, “A lot will inevitably change between now and when data centers come online.” Given the pace of change in AI application even by 2028, there are serious questions about how well these facilities will keep up with growing demand.
Timing and capacity present their own risks. At the same time, speculation is swirling over the next big breakthrough in energy efficiency, semiconductor design and power transmission that might utterly transform the AI infrastructure game. These innovations could be key in shaping how companies use AI technologies in the future.
Innovators and industry leaders will meet in San Francisco at the TechCrunch Disrupt conference on October 13-15, 2026. Look for conversations based on these themes to lead the discussions. Technological innovations and smart investments in infrastructure will be critical. Companies need to be purposeful as they figure out AI’s role in their business.
