5 Takeaways from the Bloomberg’s AI Infrastructure Discussion at the Qatar Economic Forum

At OCOLO, we try to bring you what the greatest minds in the global infrastructure space are discussing at any given moment. Today, we share takeaways from the 2025 Qatar Economic Forum, Powered by Bloomberg, a panel of industry titans Bloomberg’s Joumanna Bercetche convened to discuss how next-generation data centers are evolving to meet the demands of artificial intelligence and the critical role infrastructure and capital play in building AI’s backbone Data Center Leaders on Building AI’s Infrastructure 

Here are the leading complex trends, which OCOLO helps clients convert to actionable intelligence to gain a competitive edge in today’s increasingly challenging marketplace.

Takeaway 1: AI is Driving a Different Kind of Unprecedented Demand

Forget everything you thought you knew about data center growth. The panelists were unanimous: AI is not just adding to demand; it’s accelerating it at a staggering pace.

• The scale is astounding: Over the next four years, more than $1 trillion worth of new data center construction will be needed to meet the demand for AI. That’s roughly four times faster than cloud sector growth, which took 15 years to reach the same milestone.
• A Change in Architecture: Traditional cloud architecture uses fewer than 10 kilowatts, while new AI training models can require 132 kilowatts, with projections reaching 600 kilowatts per rack.
• The Next Frontier is Inference: Although the emphasis is on training large language models (LLMs), the true need lies in inference—the use of a trained model. With 90% of AI eventually residing on mobile devices, inference will need to be closer to the customer.

The game has changed. Your infrastructure strategy can no longer be based on yesterday’s cloud models. You need partners who understand the architectural demands of AI — from power-hungry training clusters to low-latency inference closer to the end user. With decades of global experience in digital infrastructure, no one understands the evolving data center infrastructure needs of Enterprises like OCOLO does.

Takeaway 2: The Biggest Constraint Isn’t Capital or Land—It’s Power

The industry isn’t heading for a bubble because there is no speculation on demand — the customers are already there. The real bottleneck is power.

• Global Power Grid Problems: The enormous power demands of AI computation were never intended to be handled by aging transmission grids. Data centers worldwide currently use around 60 gigawatts of power, a number estimated to  rise to 300 gigawatts.
• A Supply-Side Crisis: The industry’s current annual power output of 5-6 GW must rise to 22 GW annually to meet anticipated demand.
• Supply Chain Strain: There  may be a four-year waitlist for essential power components like gas turbines.

Finding powered land is the new gold rush. The availability of power is the single biggest gating factor for new projects. OCOLO’s global network of service providers includes more than 120 partners that help us keep our proprietary database of inventory up-to-date so we can seamlessly connect customers to properties with secured power and solutions to even the most complex supply chain logistics.

Takeaway 3: Innovation in Cooling is a Non-Negotiable

AI models’ massive power usage generates an incredible amount of heat, and the temperature is only going up.

• Legacy Systems Will Not Cut It: The air cooling methods that have been employed in data centers over the past 25 years are no longer sufficient. AI data centers demand new innovations, such as Direct Liquid to the Chip (DLC) cooling, which involves attaching a glycol device directly to the GPU to dissipate heat.
• Cooling is a foundation, not a feature. Ignoring this technological change would result in outmoded infrastructure that cannot manage the next wave of AI. While liquid cooling is not expected to fully replace air cooling technology, it is increasingly being implemented along with air cooling as AI power loads increase, and the ratio is moving in liquid cooling’s favor

Find the right partner for the data center cooling transition. Wherever you are, and whatever the size or purpose of your data center, your approach must include a plan for using new cooling technology to ensure efficiency and scalability. OCOLO can help, thanks to our extensive relationships with vendors servicing all aspects of digital infrastructure. Visit the Resource & Partners page of our website for a list of data center-adjacent equipment and service providers, or ask us for help in finding the right partner.

Takeaway 4: A New Capital Structure is Evolving

There is a crucial question emerging for investors in the digital infrastructure space: what is the residual value of a data center at the end of its 15-year lease?

• Location is the New Longevity: The answer is clear: location and interconnectivity are paramount. Legacy data centers, particularly in key interconnection hubs like Ashburn, Virginia, remain valuable and have experienced rent increases upon renewal.
• Capital is Adapting: The investment landscape for digital infrastructure is maturing, moving from venture capital to private equity and now to infrastructure funds and real estate vehicles. This “coming of age” is supported by a deep and maturing debt market, including securitized capital from insurance companies.

The capital is there, but it’s moving into new structures. Understanding where your project sits in the capital stack—and having a compelling story around location, tenancy, and technology — is essential to secure funding. Talk to OCOLO about what your funding needs are. In our decades in this business, we’ve seen a range of financing options, and we’re plugged in on the latest and greatest on both the equity and debt sides as well.

Takeaway 5: Global Growth is Widespread, but with Regional Nuances

Data center demand is strong everywhere – but the pace and drivers vary by region, country and even city. Planting flags requires time, patience, local knowledge, instincts and most of all, resources. Mistakes are costly and take years to materialize, which compounds the expense and competitive impact. Here are some broad guidelines:

• The U.S. and India: Primarily driven by hyperscale demand.
• Europe: Constrained by power availability. Fast growth in emerging DC markets
• Asia: Seeing strong enterprise demand and accelerating uptake of AI and cloud.
• China’s Catch-Up: China could pour more than $50 billion into AI capex in 2025 as it plays catch-up in building foundational LLMs.

There is no one-size-fits-all solution for global expansion. Your strategy needs to be tailored to regional dynamics, from power availability and regulations to the specific type of demand. You want a partner in this journey who knows the landscape because they’ve spent years getting to know the terrain. OCOLO provides the market intelligence, relationships, history and experience to help you make informed decisions about where and how to build, wherever you’re looking to grow.

Ready to Turn These Trends into Action?

The Bloomberg panel made one thing clear: the future of AI hinges on the infrastructure we build today. From sourcing powered land and advanced cooling systems to navigating complex capital structures, the challenges are real.

At OCOLO, we help you translate these complex market trends into a clear path forward. Our platform connects you to the right partners, properties and insights you need to build the backbone of AI – efficiently and strategically.