Executive insights on AI infrastructure from Uptime Institute’s global survey.
AI adoption is no longer theoretical. It’s here, accelerating, and already reshaping how organisations operate, innovate, and compete.
The Uptime Institute’s 2025 AI Infrastructure Survey provides an urgent executive briefing: the strategic build-out for AI is in full swing, and proactive engagement is paramount to securing future competitive advantage. Based on comprehensive data from 519 global data centre owners/operators, the survey reveals a market adapting at speed, driving significant shifts in capital expenditure and demanding innovative approaches to manage AI’s extreme density and power requirements.
The findings point to one clear message: if you’re not scaling your environment now, you’re falling behind.
Why your AI infrastructure needs to scale by 2026
AI workloads are rapidly becoming mainstream:
- 32% of operators are already running AI inference workloads
- 45% plan to implement them in the near future
That’s nearly 4 in 5 organisations either deploying or preparing for AI, confirming that AI isn’t an experiment. It’s a competitive advantage.
And powering it requires more than compute; it needs reimagined infrastructure built for density, performance, and scale.
Why your AI strategy must stay local
Public cloud is often discussed as the go-to model — but the survey shows a different reality:
- 46% of AI workloads are hosted on-premises
- 34% run in colocation environments
- Only 14% rely primarily on the public cloud
Why this shift? The survey highlights the top drivers for keeping AI workloads closer to home:
- Data Sovereignty (46%): Protect IP, ensure compliance, and reduce risk
- Reuse of Existing Infrastructure (50%): Maximise ROI on current facilities
- Power Availability (37%) & Data Proximity (29%): Critical for real-time performance
- Cost (30%): Cloud OpEx for 24/7 AI can become unsustainable
Bottom Line: AI deployment is as much about location, control, and cost predictability as it is about compute power.
Power and cooling: AI’s infrastructure breaking point
AI is not only data-hungry; it’s power- and heat-intensive. The Uptime report shows:
- 27% of AI training racks already exceed 50kW
- Even inference racks now hit the 31 to 50kW range
To meet this demand:
- 52% of data centres are upgrading power infrastructure
- 51% are modernising cooling systems
The consequences of ignoring this shift are clear: delays, downtime, and capped growth.
NEXTDC is already enabling environments with up to 600kW per rack and advanced liquid and immersion cooling technologies designed for the AI era.
NEXTDC
Connectivity and ecosystem: AI Factories need more than power
The location of AI workloads matters — not just for performance, but for capability.
NEXTDC offers:
- Data centres in every Australian capital
- Direct access to subsea cable systems, enabling faster regional AI workloads
- Dense ecosystems that bring together hyperscalers, sovereign cloud, research institutions, and high-performance digital platforms
Whether you’re training large models or serving real-time AI applications, your infrastructure must be physically and digitally close to your data, your partners, and your users.
Strategic drivers: Why leaders are investing now
The reasons behind this infrastructure race go far beyond IT. According to Uptime’s survey, the top reasons organisations are investing in AI infrastructure include:
- 50%: Improve operational efficiency
- 49%: Enable new products and services
- 41%: Enhance customer experience
- 28%: Boost employee productivity
- 25%: Differentiate in the market
This isn’t about maintaining status quo — it’s about using AI to lead.
Bottom Line: What CIOs and CTOs Need to Know
Your infrastructure choices now will define your organisation’s AI potential for the next decade.
Ask yourself:
- Can your racks support 30kW+ densities?
- Is your cooling ready for continuous inference and model training?
- Is your infrastructure located where your AI data needs to live?
- Are you connected to the digital ecosystems that enable AI to thrive?
AI is already benchmarking your readiness. Will your infrastructure keep up?
Want to explore the full article?
Read the full Uptime Institute 2025 AI Infrastructure Survey.
NEXTDC
