This year has marked a significant turning point in the global landscape of datacenter development, capturing attention across various media platforms and community discussions. From Arkansas to Southern California, Nevada, Pennsylvania, West Virginia, and most recently Box Elder, Utah, localities are grappling with the economic potential of datacenter expansion while voicing concerns about energy consumption, infrastructure demands, and the impact on residential areas. A similar narrative is unfolding in the UK, where OpenAI’s ambitious “Stargate UK” initiative has faced setbacks due to rising energy consumption issues and regulatory scrutiny.
Infrastructure Challenges in Hyperscale Datacenters
New hyperscale datacenters often encounter significant grid-connection bottlenecks, with delays extending up to seven years in certain markets. This timeline can be exacerbated by the need for additional transmission lines, substations, generation capacity, and transformers. According to McKinsey, global spending on datacenters could soar to an astonishing trillion by 2030, a figure that rivals the GDP of some of the world’s largest economies.
The integration of AI at scale has become a cornerstone of enterprise strategy and a focal point in global discussions. The energy demands associated with this technology are substantial, leading enterprise leaders to wager that the value generated by AI will ultimately outweigh the costs of the energy required to power it. This evolving landscape has given rise to a new metric for executives: intelligence per watt.
Energy Constraints on AI Ambitions
Currently, AI-driven datacenters account for approximately 1.5 percent of global electricity consumption, with projections from the International Energy Agency indicating that this figure could more than double by 2030, nearing three percent of total global electricity use. This consumption level surpasses that of several major industrial sectors, including agriculture. The urgency for additional datacenters, particularly those situated near energy sources, is becoming increasingly apparent, with IDC forecasting that by 2029, one billion agents will execute 217 billion daily actions.
If the necessary infrastructure for the first billion agents takes up to seven years to establish, the implications for scaling to two, three, or even eight billion agents present a challenge that the industry has yet to fully address. The growing disparity between enterprise aspirations and energy capacity is a pressing concern.
For enterprise leaders, this moment represents a critical juncture. With 95 percent of global enterprises aiming to develop their own AI and data platforms within the next 780 days, it is clear that AI, data, and energy must be viewed as interconnected components of a unified strategy. The more complex question remains: how can executives pursue their AI goals while simultaneously managing energy consumption effectively at scale?
Insights from the BFSI Sector
Banking, financial services, and insurance (BFSI) sectors have historically allocated a larger share of their revenue to technology investments compared to other industries. McKinsey estimates that IT spending in banking typically ranges from six to twelve percent of revenue, in contrast to the three to five percent seen in other sectors. The drive to harness new technological value, particularly through AI and agentic systems, is fostering a common operational language among CIOs, CTOs, and business leaders.
AI and data are increasingly recognized as vital competitive advantages, yet the associated energy costs introduce a new layer of complexity to technology decision-making. A mere 13 percent of global enterprises that are successfully leveraging AI and agentic systems are more likely to develop their data strategies with a focus on control, efficiency, and sustainability. This trend often manifests as repatriation—moving AI and data from hyperscaler environments into proprietary control planes, allowing organizations to manage information across diverse platforms, whether cloud-based or on-premises.
This approach is gaining traction among leading BFSI organizations, including Wells Fargo, Mastercard, HSBC, JPMorgan Chase, Bank of America, Citigroup, Goldman Sachs, BNP Paribas, ING, Crédit Agricole, UBS, and NatWest, all of which have publicly committed to carbon neutrality while striving to become sovereign AI and data platforms.
Optimizing Energy Efficiency with Postgres
As agents operate at the data layer, it becomes imperative to manage energy consumption at that same level, where much of the operational workload occurs. Failing to do so is akin to heating a space while leaving windows wide open during winter. By exerting control over the data layer, agents, and the broader data estate, enterprises can lay the groundwork for effective energy management.
Energy efficiency initiatives must commence within existing enterprise operations, which positions PostgreSQL®—the most widely adopted database among developers—as an ideal solution. EDB Postgres AI is specifically designed to tackle the energy-intensive challenges posed by modern datacenters, enhancing database and AI efficiency precisely where workloads are executed.
By minimizing core usage requirements and optimizing data-intensive operations such as search, retrieval, and vector indexing, EDB Postgres AI has the potential to reduce datacenter energy consumption by up to 81 percent and lower emissions by as much as 87 percent. The ambition to evolve into an AI and data platform hinges on a fundamental principle: AI and data sovereignty. Organizations embracing this model not only achieve a fivefold return on investment and deploy twice as many AI systems, but they also gain enhanced control, improved efficiency, and a more intelligent framework for designing and operating datacenters in the agentic era. The pathway to success lies in achieving sovereignty through Postgres, representing the most pragmatic approach to maximizing intelligence per watt.
Contributed by EDB.
