Data centers worldwide are increasingly struggling to keep up with AI demand. Costs, power, and staffing are the primary concerns.
Uptime Institute surveys data center operators annually to identify industry trends. This year’s report highlights the major challenges, largely linked to the ongoing growth and demand for AI computing power, which brings capacity concerns.
Increasing costs are seen as the biggest challenge, with predicting future capacity as a “new” headache for data center operators. The demand for AI capacity remains volatile. Operators are also concerned about the availability of sufficient power and potential fluctuations in energy prices, which could lead to additional costs.
read also
Belgium as a Surprising Frontrunner: AI Factories in the Making
Fewer Outages
Although data center outages often make headlines, they are becoming less frequent, according to the report. Half of the surveyed operators have been spared outages over the past three years. Only three percent experienced a “serious” incident according to Uptime’s scale.
On-premises data centers remain the most popular form: 45 percent of workloads run on self-managed infrastructure. Sixteen percent are housed in colocation data centers and ten percent in a private cloud. Despite rapid growth, the public cloud accounts for “only” eleven percent.
read also
European Cloud Providers See Revenue Growth, but not Market Share
Wanted: (Older) Staff
Staff shortages have been a recurring theme in the data center world for several years. Forty-six percent admit difficulty finding new staff, while 37 percent cannot retain employees. Uptime Institute notes a shift in demand towards senior profiles, whereas younger workers were sought in recent years.
According to Uptime, this may be due to more senior profiles retiring or leaving their employers for other reasons. A shortage of experienced workers not only causes short-term staffing issues but also prevents the transfer of senior knowledge to juniors to fill roles in the long term.
Cool Lovers of AI
Data centers are eager to welcome AI models and companies, but operators themselves are only lukewarm enthusiasts. According to the study, the data center industry is cautious in adopting AI, focusing mainly on tools that can increase efficiency or mitigate human errors.
Only fourteen percent of operators would trust AI with configuration changes. When it comes down to it, data centers primarily rely on their own human expertise.