The race to build artificial intelligence infrastructure has quietly shifted from a competition over computing chips to a scramble for something far more fundamental: reliable, affordable electricity. While headlines focus on graphics processing units (GPUs) and model performance, the companies actually winning the AI infrastructure race are those controlling access to power generation, transmission capacity, and the physical facilities to house energy-hungry servers.

Why Is Power Becoming the New Bottleneck in AI?

The International Energy Agency projects that global electricity consumption from data centers will more than double by 2030 to roughly 945 terawatt-hours, with the McKinsey consulting firm expecting U.S. data center electricity demand to account for nearly 12% of total U.S. power demand by the end of the decade. This explosive growth stems from the fact that modern AI servers consume enormous amounts of electricity, and nearly all of that electricity eventually becomes heat. As data centers become denser and more powerful, traditional cooling methods are reaching their limits.

For years, investors assumed semiconductors would be the defining constraint of the AI era. Increasingly, it looks like electricity will be the real limiting factor. CoreWeave, originally a cryptocurrency mining operation that transformed itself into one of the world's largest independent providers of AI computing infrastructure, reported a record $99.4 billion revenue backlog in the first quarter of 2026 and generated $2.08 billion in quarterly revenue, up 112% year-over-year. The company disclosed that its available 2026 capacity was effectively sold out, underscoring just how constrained compute resources remain.

Which Companies Are Winning the Infrastructure Race?

Several companies are positioning themselves at the center of AI's infrastructure constraints. Bitzero, which listed on the Nasdaq under the ticker AIBZ on June 9, 2026, controls more than 1 gigawatt of planned data center capacity across Norway, Finland, and North Dakota, with its Nordic assets powered by low-cost renewable energy. The company secured much of that capacity before regulators began tightening restrictions on new large-scale data center development, giving it a significant competitive advantage.

In May 2026, Bitzero signed a binding letter agreement with AI cloud provider OneQode for a 15-year lease covering the full 110 megawatts of its Namsskogan, Norway site, with total potential revenue of approximately $2.6 billion, or roughly 20 times Bitzero's current market cap at the time of listing. Based on current internal assumptions, Bitzero estimates site net operating income margins of approximately 85% at full capacity, implying around $151 million in annual net operating income once the facility is fully deployed.

Vertiv, a company specializing in the infrastructure that keeps data centers operational, reported 30% year-over-year revenue growth in the first quarter of 2026, with adjusted earnings per share jumping 83%. The company provides power systems, thermal management, liquid cooling technologies, and equipment racks that prevent AI chips from overheating. Management raised full-year guidance and highlighted continued strength in high-density computing environments where advanced cooling and power-management systems are becoming essential infrastructure rather than optional upgrades.

How Are Data Centers Adapting to Power and Cooling Challenges?

• Liquid Cooling Adoption: As AI server densities continue to rise, operators are increasingly adopting liquid cooling and immersion cooling technologies to manage heat more efficiently and support high-performance computing workloads, with industry forecasts projecting rapid growth in the liquid-cooling market.
• Renewable Energy Integration: Operators are incorporating renewable energy into their long-term strategies, with the continent's abundant solar, wind, and hydropower resources offering opportunities to strengthen energy security while lowering operating costs.
• Strategic Location Selection: Companies are prioritizing facilities in regions with cheap, clean power and cold climates that reduce cooling costs, such as Nordic countries and areas with existing grid connections.

Bitzero's strategy exemplifies this shift. CEO Mohammed Bakhashwain summed up the company's approach plainly: "Build where clean energy is produced". The company's Kokemäki, Finland campus has planned capacity of between 600 megawatts and 1,000 megawatts at full buildout, with engineering support from Red Engineering Design Ltd., a global data center consultancy and strategic partner of NVIDIA in designing next-generation AI factories. An initial phase of up to 80 megawatts is targeted for service delivery in 2027, and the site already has a confirmed 400-kilovolt high-voltage grid connection.

What Does This Mean for Global Data Center Investment?

The global data center infrastructure market is experiencing rapid growth. The market was valued at $297.07 billion in 2025 and is projected to reach $752.12 billion by 2034, registering a compound annual growth rate of 10.9% over the 2026 to 2034 forecast period. This expansion is driven by cloud adoption, artificial intelligence and high-performance computing workload growth, and rising digital transformation investments.

Africa's data center sector is expected to attract an additional $8.76 billion in investment by 2031, with investment in data centers projected to grow at a compound annual growth rate of 15.76% through 2031. The colocation segment is expected to expand even faster, with a projected compound annual growth rate of 23.74%, as businesses increasingly seek scalable, carrier-neutral infrastructure. South Africa, Kenya, Nigeria, and Egypt are expected to remain Africa's leading data center investment hubs, while countries such as Morocco, Ghana, and Tanzania are emerging as new growth markets.

The rapid adoption of emerging technologies such as artificial intelligence, machine learning, big data, and the Internet of Things across Africa will increase demand for AI-ready data centers. International technology companies including Microsoft and Google have announced multi-billion-dollar investments in AI infrastructure and data centers across Africa. NVIDIA plans to build an "AI factory" in Kenya over the next three to four years and deploy 12,000 graphics processing units in new facilities.

How to Evaluate Data Center Infrastructure Investments

• Power Access and Capacity: Assess whether a facility has confirmed access to reliable power generation and transmission capacity, including high-voltage grid connections and renewable energy sources that can support long-term growth without regulatory constraints.
• Cooling and Thermal Management: Evaluate whether the data center employs advanced cooling technologies such as liquid cooling or immersion cooling systems, which are increasingly essential for supporting high-density GPU deployments and reducing operational costs.
• Geographic and Regulatory Positioning: Consider whether the facility is located in regions with favorable regulatory environments, low construction costs, cold climates that reduce cooling expenses, and existing submarine cable networks for international connectivity.

The shift in AI infrastructure priorities reflects a fundamental reality: computing power is now abundant relative to the electricity needed to run it. Securing data centers today is less about capital than it is about access to power generation and transmission capacity. Bitzero has that advantage. Most of its competitors are still trying to get it.

As AI workloads continue to grow, the companies that control reliable, affordable electricity and the infrastructure to deliver it will likely shape the next decade of artificial intelligence development far more than those focused solely on chip design or model architecture.