The Price of Intelligence: Nebius Tackles AI's Heavy Resource Toll

AMSTERDAM – July 30, 2026

The artificial intelligence revolution runs on a very real currency: power and water. As demand for generative AI drives an unprecedented build-out of data centers, the global economy is confronting the immense resource appetite of its new digital engine. Against this backdrop, AI cloud company Nebius (Nasdaq: NBIS) has released its 2025 Sustainability Report, offering not just a scorecard of its environmental impact but a strategic blueprint for how to manage it while scaling toward "multiple gigawatts" of compute capacity.

Engineering a More Efficient Cloud

The core of Nebius's argument is that sustainability is an engineering problem. The company's report details impressive efficiency metrics that place it well ahead of industry averages. Across its global portfolio, Nebius achieved an average Power Usage Effectiveness (PUE) of 1.25 in 2025. This figure, which measures how much energy drawn by a data center actually reaches the IT equipment, stands in stark contrast to the global industry average of 1.54. The difference isn't trivial; Nebius estimates its efficient design avoided 102 GWh of electricity consumption last year, enough to power nearly 10,000 American homes.

The company's advantage, according to its report, stems from a full-stack approach—designing servers, cooling systems, and cloud software as a single, optimized system. "How AI is built matters just as much as what it enables," said Daria Mukhortova, Head of Sustainability at Nebius. "Our job is to clearly report the impact of increasing demand for AI, and to use the data to improve how we build."

This philosophy extends to water consumption, another critical resource for cooling data centers. One of its liquid-cooled facilities recorded a Water Usage Efficiency (WUE) of just 0.018 liters per kilowatt-hour (L/kWh). Compared to the projected U.S. data center average of 0.45-0.49 L/kWh, the savings are dramatic. Projecting this forward, the company estimates its designs could avoid nearly 900 million gallons of water for each gigawatt of capacity it builds—a significant claim as it eyes massive expansion.

From Waste Heat to Community Wealth

While internal efficiencies address the cost and environmental footprint, Nebius is also looking outward, reframing its infrastructure as a potential community asset rather than just a resource drain. The most compelling example is its data center in Mäntsälä, Finland. In 2025, the facility’s heat recovery system exported 19.5 GWh of captured waste heat into the local district heating network. This symbiotic relationship provided tangible benefits, reducing heating costs for connected households by approximately 10%. At full capacity, the system is designed to export over 250 GWh annually, turning a computational byproduct into a valuable local utility.

This model of community integration is a core part of the company's public-facing strategy. "AI infrastructure is built in real communities, so we start with the people already there — school districts, community colleges, the trades and first responders," noted John Sutter, VP of Public Affairs at Nebius. The company is pairing its infrastructure investments with local partnerships, from supporting STEM education and skilled-trades apprenticeships to collaborating with universities. In 2025, the Nebius Academy partnered with 16 universities, including a notable initiative with Rowan University in New Jersey to create academic pathways in AI and cloud computing, aiming to prepare local talent for the jobs its facilities create.

The High Stakes of Scaling Responsibly

The sustainability metrics are particularly crucial when viewed against the sheer scale of Nebius's ambition. The company isn't just growing; it's expanding at a blistering pace. With Q1 2025 revenue up 385% year-over-year and a revised 2026 capital expenditure guidance soaring to between $20 billion and $25 billion, Nebius is in a race to build out a global AI platform. The company reports having already contracted over 3.5 gigawatts of power to fuel this expansion.

In this high-stakes environment, efficiency is not just an environmental goal—it is a fundamental pillar of economic survival. Every watt saved on cooling or idle compute translates directly to the bottom line, lowering the Total Cost of Ownership (TCO) and improving competitiveness. The transparency detailed in its report, which aligns with emerging European standards, is therefore both a nod to accountability and a strategic move. By openly reporting its resource usage, Nebius aims to build trust with communities and regulators, securing the social license needed to build at such a massive scale. This strategy is complemented by an aggressive renewable energy procurement plan, having already achieved 100% renewable electricity in Iceland, France, and the UK, while pursuing novel solutions like behind-the-meter fuel cells in the U.S.

The report from Nebius offers a detailed look at one company's attempt to reconcile the exponential growth of artificial intelligence with the finite resources of the physical world. By integrating efficiency into its core engineering and business strategy, the company is making a calculated bet that sustainable design is the only viable path to long-term leadership in the AI infrastructure market. The ultimate test, however, will be whether these impressive efficiency and community integration models can be replicated without compromise across the multiple gigawatts of power and billions in capital that Nebius plans to deploy in the coming years.