Impact in Action | Q3 2025
The rise of artificial intelligence has been both swift and staggering. In only a few years, tools like ChatGPT, Claude, or Gemini have transformed AI from an abstract concept into a visible force in daily life, business strategy, and global markets. What began as an experiment in human-like conversation has evolved into vast computational networks that generate text and code, inform decisions across industries, and provide answers to a wide array of daily questions. Yet this newfound “intelligence” we celebrate doesn’t exist in the cloud; it’s anchored firmly in the physical world. Every AI model relies on enormous data centers humming with electricity and cooled by water systems that stretch local infrastructure.
The rapid expansion of AI data facilities has quietly become one of the most consequential environmental stories of the decade. As investors and individuals, we face a dual challenge. We want to harness AI’s economic and efficiency potential, yet we also need to face the reality of its growing energy appetite. This dynamic tension between technological ambition and environmental responsibility can support the development of systems that are both powerful and sustainable.
The Beating Heart of Artificial Intelligence
The revolution in computing that made AI possible began with Nvidia’s invention of the graphics processing unit (GPU) in 1999. GPUs unlocked the ability for computing to be faster and more efficient, creating the foundation for today’s generative AI systems and further innovation. Moore’s Law, which predicted that computing capacity would double roughly every two years, has given way to “Huang’s Law,” named after Nvidia CEO Jensen Huang, describing the exponential performance gains of modern chips.
Nvidia’s latest Blackwell chip can perform 30 times faster than its predecessor while being up to 25 times more energy efficient – yet each chip still consumes around 1,200 watts, roughly the same as an average U.S. household. Multiply that by tens of thousands of chips per data center, and the scale of the energy challenge comes into view.
The Growing Appetite of Data Centers
While AI companies emphasize efficiency gains, history shows that technological improvements often lead to greater overall consumption. In Wisconsin alone, two proposed AI facilities are projected to consume more electricity than all residential homes in the state combined.[1] These hyperscale data centers, which are 10 to 20 times larger than traditional ones, are now central to the growth of AI. This surge is contributing to rising electricity prices across multiple states, reversing a decade of relative stability.
The environmental implications are multifaceted. Beyond electricity, AI systems also require massive quantities of water to cool processors. A single data center can use hundreds of thousands of gallons daily. In drought-prone regions, this compounds an already delicate resource balance, even lowering the water table.[2] In all regions, data centers drive up utility costs for local residents sharing the same water and power grid.[3]
Policy and Path Forward
Governments and regulators have shifted into response mode relatively quickly. In the U.S., new rules under the Department of Energy are exploring how to classify and track AI-related demand within the broader grid system. The European Union’s AI Act now requires high-intensity AI systems to disclose energy use and environmental impact. The key for both policymakers and investors is measurement: understanding the balance between the value AI creates and the resources it consumes.
In response to concerns about sustainability and rising energy costs, data centers are increasingly pairing with renewable energy projects and battery storage to stabilize their energy supply. Microsoft and Google have signed multi-decade power purchase agreements that are helping finance new solar, wind, and geothermal capacity. Innovations in liquid cooling, closed cooling systems, chip architecture, and “AI for climate” applications – where AI itself is used to optimize grid efficiency or forecast renewable generation – offer promising examples of circular progress.
Impact Investing in the Age of AI
For impact-oriented investors, this moment echoes past transitions. The first wave of socially responsible investing sought to avoid harm; the next aimed to encourage best-in-class practices. Today, investors are looking to create a positive impact by engaging companies on sustainability goals while maintaining competitive returns.
At North Berkeley Wealth Management, we believe that progress and responsibility are not opposing forces. The rapid rise of AI reminds us that our investments shape the infrastructure of tomorrow. By channeling capital toward innovation that reduces emissions and enhances resilience, we can help ensure that intelligence – artificial or otherwise – is powered responsibly.
Resources
[1] AI data centers in Wisconsin will use more energy than all homes in state combined. Clean Wisconsin
[2] AI is Draining Water from Areas that Need it Most. Bloomberg
[3] AI Data Centers are Sending Power Bills Soaring. Bloomberg
