Uptime's 15th Annual Global Data Center Survey Results Show Both Commitment and Hesitancy as Industry Plans for Wider AI Usage, Climate Change Reporting, and the NVIDIA Revolution to Come
Uptime Institute today announced the release of its 15th Annual Global Data Center Survey 2025 revealing an innovative and resilient industry – one that is also facing rising costs, worsening power constraints, and challenges in meeting the demands for AI. As operators expand and modernize to meet power and density requirements, they must address availability, efficiency, staffing challenges, supply chain delays, and unpredictable technological advances.
'Our data shows operators are tasked with managing a lot of big strategic challenges at the same time. These include anticipating multiple technological changes, planning for expansion in spite of major constraints on power availability, and preparing for and supporting unpredictable AI workload demand,' said Andy Lawrence, Executive Director of Research, Uptime Institute. 'This is a time where senior level experience is critical. But for the first time, more operators are finding it harder to recruit and retain senior people than people at an earlier stage of their career. There is a management shortage, with many experienced leaders retiring just as another phase of dramatic growth gets underway.'
Roughly one-third of data center owners and operators currently perform some AI training or inference, and a significantly greater proportion plan to do so in the future. But much of this is early stage and cautious. Uncertainty over the appropriate or likely venues for AI workloads, and apprehension over the power demands of projected NVIDIA GPU systems, is likely contributing to capacity concerns.
Now in its 15th year, Uptime Institute's annual survey is the most comprehensive and longest-running study of its kind. The findings of this report highlight the practices and experiences of data center owners and operators in the areas of resiliency, sustainability, efficiency, staffing, cloud, and artificial intelligence.
Key findings from the 2025 report include:
Cost issues remain the top concern for digital infrastructure management teams in 2025 — but worries around forecasting future capacity requirements have grown significantly.
Average PUE levels show little change for the sixth consecutive year, with improvements constrained by legacy infrastructure and some climate specific limitations to efficient cooling.
Average server rack power densities continue to rise, with greater adoption of racks in the 10–30 kW range. Few facilities exceed 30 kW, and extreme densities are as yet rare.
The collection and reporting of key sustainability metrics have not improved in 2025, which is likely due in part to commercial pressures to support AI, and easing regulatory pressure in some regions.
Trust in AI for data center operations depends on the use case: most would allow its use for analyzing sensor data and predictive maintenance tasks, but not configuration changes, controlling equipment, or staffing issues.
Impactful data center outages are gradually becoming less frequent — but one in ten still cause serious or severe disruption, underscoring the need for continued investment.
Enterprises continue to adopt hybrid IT strategies, spanning cloud, colocation on-premises data centers. On-premises data centers remain foundational for those with large, mission critical processing needs, with 45% of IT workloads still residing in corporate facilities.
Staffing challenges persist in 2025. Nearly two-thirds of operators report difficulty retaining staff, finding qualified candidates, or both.
About the Survey:
Uptime conducted this year's Annual Global Data Center Survey online and via email from April to May 2025 and collected responses from more than 800 data center owners and operators. For the third consecutive year, Uptime's survey asked data center operators to identify their management team's top concerns related to digital infrastructure. In 2025, new response options were added to reflect the evolving challenges surrounding power availability, supply chain disruptions, and demand for AI.
The survey participants represent a wide range of industry verticals in multiple countries. Nearly half (43%) are located in North America and Europe. Approximately one in five respondents work for professional IT / data center service providers — that is, staff with operational or executive responsibilities for a third-party data center, such as those offering colocation, wholesale, software, or cloud computing services.
Learn More:
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Zawya
17 hours ago
- Zawya
Trend Micro Reinvents Proactive Security with Digital Twin Technology
Innovation at the speed of agentic AI, built on NVIDIA AI, provides a data-driven foundation for supercharged decision-making HONG KONG SAR - Media OutReach Newswire - 6 August 2025 - Trend Micro Incorporated (TYO: 4704; TSE: 4704), a global cybersecurity leader, today unveiled a groundbreaking new cyber resilience model that reinvents how enterprises approach proactive security. Trend's Digital Twin capabilities will empower enterprises to simulate real-world cyber threats, validate their defenses, and adapt policies in real time across complex and diverse digital environments. To learn more about how Trend's Digital Twin technology is reshaping proactive security, visit: Bartley Richardson, Senior Engineering Director, Agentic AI at NVIDIA:"In today's rapidly evolving cybersecurity landscape, organizations need proactive solutions that can anticipate and counter potential threats before they occur. Powered by NVIDIA NIM microservices, Trend's cybersecurity digital twins bring AI-driven protection to enterprise infrastructure." At the core of this model is Trend's advanced agentic AI and industry-first application of cybersecurity digital twin technology. By creating high-fidelity, continuously updated simulation of an organization's infrastructure, Trend enables security teams to visualize risk, test scenarios safely, and make rapid, data-driven decisions that improve resilience and reduce business disruptions. The shift from periodic assessments to continuous, intelligent simulation marks a major evolution in proactive cybersecurity. It allows organizations to stay ahead of adversaries, harden their environments to circumvent cybersecurity threats, and confidently secure mission-critical operations against an ever-changing threat landscape. The technology shifts security practices from static and reactive to dynamic and predictive, enabling measurable improvements across a range of critical use cases. Stuart Samples, Chief Technology Officer at Northeast Georgia Health System:"Trend's digital twin approach completely changes our ability to understand risk in real-time. It helps us catch threats we didn't even know to look for, allowing our team to focus less on firefighting and more on confidently driving innovation." Trend's cybersecurity Digital Twin technology is powered by NVIDIA accelerated computing and NVIDIA AI Enterprise software, including NVIDIA NIM microservices. This enterprise-grade software enables the delivery of agentic AI models, optimized inference, and secure, scalable deployment—combining the simplicity of APIs with the flexibility of self-hosted infrastructure to strengthen cybersecurity outcomes with AI-driven precision. Trend's Digital Twin model will help organizations proactively manage risk across their entire infrastructure—from on-prem to cloud, IT to OT, and legacy to next gen AI-powered systems. It comes as governments and industry leaders accelerate AI infrastructure investments. Trend's approach to digital twins enables customers to make better informed risk decisions and enhance security operations across their organization. Scenario planning: AI agents simulate threats and tactics within the Digital Twin, long before a real-world incident occurs. This helps to test current mitigation strategies in a continuous cycle, to improve resilience planning for sensitive and complex environments. Security investments: Decision makers can introduce new tools, policies, or architectural changes to see how they perform in the virtual environment. This results in better informed, data-driven investment decisions. Business resilience optimization: Digital Twin technology can be used to simulate business-critical failure scenarios to highlight how data flows, how decisions are made, and how disruptions ripple across IT/OT systems and teams. These real-time risk insights boost enterprise security planning capabilities without risking disruption to production systems. Frank Dickson, Group Vice President, Security & Trust at IDC:"As threats migrate to OT, proactive security is critical. Trend Micro's Digital Twin technology introduces a new operational model for enterprise cyber security: simulation. Sometimes even probing a production network can result in downtime, often making it extremely difficult to expose vulnerabilities and confirm compensating measures. A continuous cycle of adversary simulation and defensive validation becomes a value tool to ensure that organizations stay ahead of cyber threats, while being sensitive to the fragility of some operational environments." The announcement is the latest in a line of innovation-led projects from Trend designed to help network defenders harness AI, close skill gaps, and improve security outcomes. Most recently, Trend announced it would open source its Trend Cybertron AI model and agent framework in a bid to accelerate the development of autonomous cybersecurity agents. Its work has also expanded to cover protection for GenAI workloads and advances to sovereign AI infrastructure. Rachel Jin, Chief Enterprise Platform Officer at Trend: "Enterprises are struggling to defend complex, dynamic infrastructure environments from highly adaptive, AI-powered adversaries. Rapid shifts in threat actor tactics and IT infrastructure mean reactive, point-in-time risk assessments are no longer fit for purpose. Our Digital Twin technology empowers customers to simulate threats and safely validate security controls without touching production systems—to finally close the gap between digital transformation and defensive readiness." Hashtag: #trendmicro #trendvisionone #visionone #cybersecurity#digitaltwin The issuer is solely responsible for the content of this announcement. About Trend Micro Trend Micro, a global cybersecurity leader, helps make the world safe for exchanging digital information. Fueled by decades of security expertise, global threat research, and continuous innovation, Trend Micro's cybersecurity platform protects hundreds of thousands of organizations and millions of individuals across clouds, networks, devices, and endpoints. As a leader in cloud and enterprise cybersecurity, the platform delivers a powerful range of advanced threat defense techniques optimized for environments like AWS, Microsoft, and Google, and central visibility for better, faster detection and response. With 7,000 employees across 65 countries, Trend Micro enables organizations to simplify and secure their connected world. Trend Micro
The National
a day ago
- The National
Nvidia-backed 'Super Bowl of AI' is going to Washington
Nvidia announced on Tuesday that the company's flagship and much-anticipated GTC conference will take place in Washington this year. During the last graphics processing unit technology conference (GTC) in March, chief executive Jensen Huang, who has become one of the tech world's most recognisable figures in recent years, described the event as the 'Woodstock' and 'Super Bowl' of artificial intelligence. The company said it is the first time the GTC summit, scheduled for October 27-29, will take place in Washington. 'GTC DC brings together researchers, developers, business leaders, and policymakers in the heart of our nation's capital to explore breakthroughs in AI, robotics, life sciences, energy, quantum and 6G-advancing innovations vital to America's technological leadership,' said Mr Huang. During Nvidia's GTC conference hosted in California this year, Mr Huang introduced new chips and spoke of his vision for Agentic AI, autonomous vehicles, robotics and accelerated computing. The announcement comes after he strongly endorsed President Donald Trump's recently announced AI Action Plan during an event in Washington. Nvidia's chief executive has made trips to Washington over the course of this year, and more specifically the White House, where he has sought to reverse what he has described harmful chip export policies enacted under former US president Joe Biden. Mr Huang has been somewhat successful with regard to those lobbying efforts: the Trump administration announced plans last month to grant licences to Nvidia so it could resume sales of its H20 GPU to China. That decision, however, was met with criticism from some Senate Democrats and technology analysts, who described the H20 chip sales to China was 'an abrupt departure' from the administration's position in April that Beijing's access to the processors posed a serious national security risk. Last week, a representative for the US Department of Commerce defended the licences and said they were an improvement on recent years, claiming that the Biden administration had not imposed 'any restrictions on the H20 whatsoever and they flowed freely into China'. The representative added that the Trump White House was the first to implement a licence requirement for the exports to Beijing. 'The Trump administration will consider any H20 licence applications carefully,' the Commerce Department representative told The National. Nvidia also defended the company's push to sell the H20 to China. 'The H20 helps America win the support of developers worldwide, promoting America's economic and national security,' an Nvidia representative told The National. 'It does not enhance anyone's military capabilities, and the US government has full visibility and authority over every H20 transaction.' Despite continued controversy over exports to China, Nvidia has had many reasons to do a victory lap in recent weeks. In July, it become the first public company in the world to pass $4 trillion in market capitalisation. Nvidia has also played a major role in a global AI infrastructure project backed by the UAE, and in May, the company, along with Oracle and OpenAI, teamed up with Abu Dhabi's G42 to bolster an AI hub project called Stargate UAE.
The National
a day ago
- The National
How data centres could help the Gulf meet its climate targets
Data centres that power the Gulf region's burgeoning cloud and AI systems are accounting for an ever-increasing proportion of electricity consumption. If managed well, this nominal increase in the need for power generation can be transformed into a tool for the greater use of renewable energy, helping the GCC countries to realise their ambitious climate goals. Saudi Arabia, the UAE and other Gulf countries are aggressively investing in AI as they seek to maintain their competitiveness in a rapidly evolving global economy. One of the key issues that policymakers must grapple with is how to supply AI infrastructure with the power it needs to operate. Despite the recency of these novel technologies, they already account for almost 2 per cent of global electricity consumption; this figure is set to grow exponentially over the coming decades as AI is integrated into every walk of life. A parallel and ostensibly independent challenge that governments are contending with is the need to decrease carbon emissions as part of the global fight against climate change. The Gulf states have committed to realising carbon neutrality within the next four decades, driving heavy investments in renewable energies such as solar and wind. They also see this as an opportunity to develop a comparative advantage in the probable successor to fossil fuels, as reflected in Abu Dhabi's Masdar becoming a global supplier of cutting-edge clean energies. Although the technological advancements in renewable energy have been breath-taking, two key challenges persist. The first is intermittent generation combined with a lack of battery infrastructure. The Sun does not shine all day, nor does the wind blow throughout it, and so electricity demand is invariably out of sync with the unpredictable supply profile that these technologies provide. This leads to significant waste and to the need to invest in backup, fossil fuel-powered generators to avoid supply disruptions. The second problem is that renewable energies cause voltage and frequency fluctuations, which in turn lead to instability in the electricity grid. The resulting blackouts can be damaging economically, and also constitute a national security threat given the power-intensive and integrated nature of most modern defence systems. As a result of these two difficulties, renewables' installed capacity has not kept pace with the advancement in generation technology. However, a new paper published by Massachusetts Institute of Technology economists Christopher Knittel, Juan Ramon Senga and Shen Wang offers policymakers a way of killing two birds with one stone. The scholars begin by noting that one of the key advantages of data centres is the inherent flexibility of their electricity demand profile. While some baseline processes must occur on a continuous basis throughout the day, a lot of the computationally intensive – and hence electricity-hungry – operations can easily be delayed by a few hours or more without disrupting overall performance. An example would be a large language model such as ChatGPT reading and understanding a voluminous new tranche of material, such as all of the manuscripts in a collection that has recently been digitised. As more renewable energy is integrated into the generation profile, AI data centres can be instructed to synchronise their flexible operations – and hence their electricity demand – with the renewable energy output As long as AI data centres continue to account for a small proportion of electricity consumption, this flexibility will remain inconsequential. However, if – as expected – they transition into being among the biggest sources of energy demand, the flexibility can be exploited to mitigate the intermittency and grid instability problems caused by renewable energy. More specifically, as more renewable energy is integrated into the generation profile, AI data centres can be instructed to synchronise their flexible operations – and hence their electricity demand – with the renewable energy output. This diminishes the likelihood of unconsumed electricity being wasted and decreases the need to invest in expensive battery technologies. Moreover, the voltage and frequency surges that destabilise the grid are likely to decline in incidence. As a result, renewable energies become much more reliable components of the energy mix, and governments can more confidently invest in them without fearing the fragility that has so far limited the transition away from fossil fuels. This solution is by no means a silver bullet. A key weakness is that it depends on a large growth in electricity demand occurring and being matched by a comparable expansion in electricity generation as a prerequisite to shifting the balance towards renewables. However, this challenge will almost certainly be mitigated by complementary technological advancements in other areas, such as battery technology and smart grid management. As a result, Gulf countries can exploit this mechanism as part of a multi-pronged approach to greening their energy mix. In short, the rise of AI and the accompanying expansion of data centres need not be a burden on the GCC's climate ambitions. On the contrary, as policymakers embrace a strategic view of electricity demand management, flexible data centres can become powerful enablers of a greener grid. By aligning the region's digital and environmental transformations, the Gulf can turn a looming challenge into a unique opportunity – advancing both its technological leadership and its commitment to climate responsibility.
