Source: voronoiapp.com Link: https://www.voronoiapp.com/technology/-US-Data-Center-Construction-Spending-Soars-Amid-AI-Investment-Surge-3294
Now, I want to start out by saying that AI is both fascinating and scary at the same time. There are certainly huge benefits that can be realized out of “good, well-trained” AI, promising everything from better healthcare to smarter cities. Heck, it is even being used to clean clothes better (my new LG washer has some basic AI built in to determine how best to wash a load of clothes). But under that excitement liessome definite challenges, the largest is around massive datacenters being built and forecasted to be built to meet the insatiable demand for AI. I find it kind of ironic that there is now this huge drive to build these datacenters.
Source: datacenterknowledge.com Link: https://www.datacenterknowledge.com/infrastructure/top-5-data-center-stories-week-of-march-20
Over the last two decades, data center growth had stayed relatively stable, largely due to virtualization. This innovation allowed organizations to maximize their existing hardware resources, effectively doing more with less. Certainly, virtualization lead to the creation of cloud (private, public and hybrid) but there were ways to do more green or climate sensitive builds. At one point at VMware, we had talked about creative a Sustainability standard/certification for partners, suppliers and customers — a testament to the industry’s previous commitment to green computing.
However, the landscape has shifted dramatically. During the pandemic we saw the rise of AI and has since skyrocketed. The projected growth over the next few years is even greater. In many ways, this is disheartening. All the benefits of virtualization will be negated and made worse by AI. This has introduced two new challenges: electrical grid strain and water scarcity.
So let’s start with grids, particularly in places like the US. Currently, they can barely handle the load currently placed on them, nevermind doubling, tripling or quadrupling demand. And while there are upgrades being done, they are likely not being done at a rate to adequately support the increasing demand that AI data centers will place on them. Add to that events like unexpected winter disasters in places like Texas in 2021; California’s brownouts, excessive heat waves and wildfires; and you have a recipe for disaster. (there are countless other disasters to be added here but these were the first two that came to mind so quickly).
The usage of electricity has increased to the point that in 2023 and 2024 AI have increased emissions by major clouds (think Microsoft, Amazon, Google, etc.) by double-digit numbers.
Source: The New Stack Link: https://thenewstack.io/how-much-energy-is-really-being-consumed-by-data-centers/
Electricity aside, data centers are huge consumers of clean water for cooling the servers. Water is something that is becoming more scarce. Usage of water by data centers has been increasing over the years and will likely accelerate. As I write this, there are significant fires burning in Los Angeles where there have been shortages of water. Now, currently, the water shortage is not due to data centers but rather due to drought conditions and elevation issues .(while the Palisades area, in the Hollywood Hills, ran out there was plenty of water in the lower elevations). This doesn’t mean that it couldn’t be an issue in the future. It’s something that we should address now rather than experience it in the future.
Source: The Hustle Link: https://thehustle.co/big-tech-s-thirst-for-ai-dominance-may-bring-literal-thirst-for-everyone-else
Some may think that I’m anti-AI. I’m not. I do believe, however, we need to be smart about how we use our resources and plan for the future.We stand at a crossroads. The choice isn’t between embracing AI or protecting our environment – it’s about finding ways to achieve both. I know that there are many designs being proposed and, in fact, there is a Sustainable Data Center Summit coming up in February. I would say that there are huge opportunities at finding solutions to allow for AI to exist and feed our insatiable desire to learn while balancing against the needs of the finite resources we have.
This requires that leadership in both government and corporations to take the lead on:
- Investment in renewable energy infrastructure
- development of more efficient cooling techniques, like Direct-to-Chip cooling and zero water data centers
- Smart grid management systems
- industry-wide commitment to sustainable practices
The future of AI doesn’t have to come at the expense of our planet. By acknowledging these challenges now and actively working to address them, we can ensure that the AI revolution advances responsibly and sustainably.
Leave a Reply