Artificial intelligence can do the world a lot of good, from revolutionizing productivity in several industries to enabling predictive analytics that could change operational efficiency. This has resulted in a massive compound annual growth rate (CAGR) of 27.67 percent, showing the increase in interest that artificial intelligence has had in the past year alone. Many people have begun to wonder whether this rapid growth is sustainable in the long term, however.
To support this expansion of artificial intelligence technology, we must also expand the infrastructure that powers it. AI models run on highly sophisticated, complex hardware that is not only expensive to acquire but also costly to maintain and operate. These tools are often hosted at large, centralized data centers that companies own to maintain their models. For example, OpenAI — the company behind the popular large language model ChatGPT — is looking to add between five and 10 new data centers in the coming years.
However, addressing this need is not as simple as building the data center and allowing it to do its thing. Establishing new data centers poses all sorts of challenges for artificial intelligence developers and the communities in which they are located, from environmental to social and political issues.
The environmental impact of AI infrastructure
Critics have increasingly called attention to the environmental effects of artificial intelligence. People are just beginning to realize the immense carbon footprint of AI technology because of the computing power it takes to power these models. Studies have found that a single prompt on ChatGPT requires 10x as much power to answer as a Google search. Over time, this adds up, especially as more people begin to use the technology.
Beyond their direct energy consumption, AI data centers also pose the threat of resource diversion. Because of the amount of power needed to support them, they put a serious strain on power grids. In many cities, power infrastructure may not be equipped to handle the new data centers’ power requirements, which can cause power disruptions to the community. To manage the increased demand, municipalities may need to upgrade their power infrastructure, which can be costly, with taxpayers often having to foot the bill.
Power grids aren’t the only resources that data centers stretch thin. They also comprise a ton of equipment, from graphics processing units (GPUs) to cooling systems required to maintain and operate these complex, technological systems. Estimates have found that these facilities, especially for large companies, can require as much as 500 to 800 acres of land, which takes up a significant footprint in the surrounding communities in which they are built.
The ethical impact of AI infrastructure
Another area where the expansion of artificial intelligence causes concern is its potential ethical implications. At this point, an uneasy alliance exists between artificial intelligence proponents and opponents thanks to keen oversight and strategies to mitigate its shortcomings. As the technology proliferates and more people begin to use it, however, maintaining strict oversight will become challenging.
One of the primary ethical concerns surrounding artificial intelligence is the security of the data that these models use. Critics have raised several data security concerns, including the fact that AI models sometimes use user inputs as part of their training data, meaning that any information a user shares with the model is liable to be reused and exposed. That is not to mention that new data centers introduce additional points of vulnerability that hackers and other wrongdoers can exploit to gain illicit access to the data that AI companies store.
Similarly, as artificial intelligence use continues to expand, there will be less oversight of the potential biases these models exhibit. Because AI technology still depends on existing data to power it, its output will reflect any biases present in the data it is trained on. Although there are ways to mitigate this bias, such as ensuring that training data is a diverse, representative data set, closely monitoring the output of AI when it is at such a large scale becomes difficult. Because of this, biased output can slip through.
Unfortunately, the disparity in resources makes this bias even more severe. Because data centers for AI computing are expensive to build and maintain, large corporations end up controlling most of the computing resources required to operate artificial intelligence models. So, many of the most powerful AI systems in the world serve private interests, not the public good, creating concern among the general public that this technology could be abused.
Overcoming the shortcomings of AI together
The goal of artificial intelligence developers, therefore, is to restore — or create — public trust in the technology. There is still widespread skepticism about the technology’s future, with more Americans reporting that they think the harm of AI outweighs its benefits than the opposite.
To establish greater trust, we must first democratize access to AI infrastructure. One of the greatest problems with AI infrastructure is its inaccessibility. The hardware required to operate advanced artificial intelligence — data centers, chips, and cloud resources — is centralized and costly, which makes it difficult for smaller players (such as developing nations) to compete. By developing lightweight, decentralized AI architectures that reduce dependency on hyperscalers, we can level the global playing field.
The public also values transparency, particularly regarding emerging technologies such as artificial intelligence. Issues like opaque training data, unexplained outcomes, and public misinformation about AI erode the public's trust in this technology. When leaders in the industry are more transparent around datasets, model behavior, and ownership structures, the public is more likely to offer their trust.
A key step in establishing this trust is ensuring consistent human oversight. Unintended bias and opaque decision-making are among the leading causes of mistrust for artificial intelligence technologies, especially in high-stakes environments like finance, healthcare, and national security. By implementing “human in the loop” oversight, where humans are involved in every part of the AI’s lifecycle, from training and evaluation to operation, leaders in AI can make the average consumer feel more comfortable with the performance of models.
Trust also requires collaboration across sectors, between various industries interested in developing artificial intelligence, including government, academia, and industry. By coordinating all of these stakeholders on issues like AI safety and fairness, leaders can come together for the joint development of ethical and technical standards — the first step in creating a united global governance framework to harmonize governance.
Currently, standards related to artificial intelligence use are a fragmented framework of laws and guidelines. Different jurisdictions have different expectations regarding AI ethics, data security, environmental impact, and other issues posed by artificial intelligence. Fragmented AI policies create loopholes that wrongdoers can exploit, making collaboration difficult; however, it is also crucial that the sovereignty of participating nations is respected.
As a result, the way to create public trust in this technology is through global cooperation. By bringing together key decision makers internationally, including AI developers, business leaders, and lawmakers, we can create a global standard that dictates the responsible use of artificial intelligence.
— Pedro is the Founder of VIVY TECH, a venture studio focused on building technology and AI-driven solutions across several industries like finance, defense, health, education, sports, and commerce. Born in Caracas and now based internationally, Pedro’s path has spanned a handful of countries. His academic journey is equally global—he graduated from Los Arcos in Caracas while also training at IMG Academy in Florida, and has studied at the University of Tampa, Universidad Alfonso X El Sabio in Madrid, Suffolk University in Boston, and Harvard University, where he is currently a visiting student. Alongside his studies, Pedro has competed in sports at elite levels, representing Venezuela’s national golf team and receiving recognition in the tech and international relations space.
Copyright © 2025 California Business Journal. All Rights Reserved.
For California Business Journal Disclaimers, go to /terms-conditions/.
