Data centres are the refineries of this data-driven world—but what happens when demand for these refineries fails to keep pace with rapid AI growth? The answer is a bottleneck across the industry, delaying the AI plans of several companies.
That situation has now become a reality, with several new data centre projects in the US running behind schedule.
A Financial Times report, based on data from SynMax, a satellite and AI analytics group, states that major projects from tech giants, including Microsoft and OpenAI, are running behind schedule and are likely to miss their completion deadlines by more than three months.
These upcoming data centres were supposed to handle the increasing demand for computational power needed to train and run AI models.
The delay is not only impacting companies dependent on these data centres for their AI plans, but also AI companies such as OpenAI and Meta, which have invested billions of dollars in developing AI technology. It is taking longer than expected for them to convert these investments into revenue and generate returns for investors.
What is causing the delays?
There are multiple reasons behind the delays. The Financial Times, quoting two construction executives, reports a shortage of specialist workers, including electricians and pipe fitters.
There are also challenges inherent in data centre operations. Data centres require massive amounts of power, with some upcoming facilities in the US consuming energy equivalent to the output of a nuclear power plant. Strained grid capacity, along with shortages of equipment such as gas turbines and transformers, is further slowing progress.
Data centres also require large spaces to host servers, cooling systems, and networking equipment. OpenAI is building a 1,200-acre data centre in the US. Due to the need for vast land, companies often build data centres in remote areas where land is cheaper. However, these locations are pushing labour costs up by as much as 30 percent.
Some AI companies are also working on multiple data centre projects simultaneously. The concentration of projects in certain regions is leading to competition between providers, as workers move from one project to another in search of better pay.
Financing pressures
There are also challenges in financing new data centre projects. Recently, a number of US banks retreated from their commitments to fund Oracle-linked data centre projects due to OpenAI’s outstanding $1.4 trillion in multi-year commitments.
“Financing at this scale is hard,” said Wes Cummins, chief executive of data centre operator Applied Digital, as reported by the Financial Times.
US-Iran war and its impact on data centres
At the same time, investments in new data centres in the Middle East are in limbo due to the ongoing conflict between the US and Iran, further exacerbating the situation for AI companies.
Iran’s Islamic Revolutionary Guard Corps (IRGC) has designated 18 major US technology firms, including Meta, Google, Apple, Microsoft, Nvidia, and Tesla, as “legitimate targets” for attack. Iran alleges these companies aid in the targeting of its officials and has threatened strikes against data centres and regional offices.
Meanwhile, representatives of companies that build and operate data centres for clients, including Nebius and Applied Digital, have said they expect to deliver their projects on time.
