Are AI Data Centers Good or Bad?

Show Notes

A data center is a collection of computers and networking equipment housed in a centralized location. They provide cloud computing capabilities for everything from web hosting through social media apps.
- A common example is photo backups. It’s very common for someone to backup their photos to a service like Google Photos or OneDrive. When you do this, you’re storing your photos in an infrastructure that uses at least one data center.
- In a very simplified case of Google Photos: you take a photo with your phone, it uploads to one of Google’s data centers where your photo sits on a storage device somewhere - ready for you to access it from any of your internet connected devices.
- Some common data center complexities:
  - For security purposes your photo may be stored piecemeal across multiple data centers to prevent physical break-ins at a single data center becoming a data breach
  - For quick access, your photo may be stored in multiple data centers to allow you to quickly access it wherever you are geographically
Data centers are located in a variety of facilities from huge warehouse-like buildings, through smaller non-descript commercial structures that sit in amongst offices in business parks
Data centers have been around for decades and even with the rise of cloud computing in the 2010s we didn’t see much pushback against them… but the latest need for data centers has many communities, activists, and industry critics worried…

Both terms “hyperscale data centers” and “hyperscalers” have been around since before the AI boom

Hyperscale data centers are massive data centers that engineered for large-scale workloads across optimized networking capabilities - they offer extreme scalability capabilities
- They’re commonly used for AI, automation, data analysis, data storage, data processing, and other large computing needs
The scale disparity of hyperscaler versus a traditional data center is quite large
- In the podcast I listened to they were referring to customers as “Mega Watt customers, such as this is a 30 megawatt customer”
- When it comes to traditional data centers you may look at things like bandwidth, the amount of websites it could host, how many hypervisors you’ll need (this of course varies wildly depending on scale)

Hyperscalers are a term used to refer to cloud service providers (CSPs) that provide hyperscale data center services - such as Amazon’s AWS and Microsoft’s Azure.

In 2026 and the previous few years, AI computing (chatbots, image generation, code generation, etc.) has been made available to the public - and this computing is almost all done in the cloud… meaning that the need for AI data centers is growing exponentially
Due to the speed and scale of the need for data centers, we’re starting to see several issues including economic, environmental, and local (municipal) concerns.

This growth has been so explosive that we’re seeing solid state storage and RAM prices soar - causing common household tech like the Steam Deck, PlayStation 5, and personal computers to increase in price by a lot - mostly because these AI data centers are ordering equipment months or years in advance claiming stock that hasn’t even been manufactured yet at prices that have increased significantly due to the rule of supply & demand
- Critics argue that this alone could completely crush the consumer electronics industry
- Others say that this is the natural progression of tech as computing moves more on the cloud than locally - which would leave us consumers with simple terminals that can do little else other than log into a remote system somewhere (ie your Xbox doesn’t render a game, you stream a game from a data center)
Many jobs have been created by data center construction, however many are temporary as it takes a lot more people to construct a data center than it does to keep it up-and-running

Large data centers are commonly built in small municipalities where the amount of land required to build large facilities is plentiful - however, these municipalities are not often equipped to handle the influx of change these large projects bring about
- Many residents that once lived rural may find themselves right beside a massive data center that is spouting pollution from natural gas power generation, or creating a tonne of noise through normal operation
- Since these communities are often small, there aren’t a lot of houses and apartments for people to stay in, so housing prices and rents soar as workers flood the area during the data center construction
- There are also concerns that data centers are not providing enough tax revenue to the places where they exist and so the residents and other businesses are stuck footing the bill for the local infrastructure, schools, etc.
- If the municipality is providing grid power or local water - infrastructure can easily become strained effecting water pressure, causing power outages, and more

Underwater Data Center (UDC)
- Built underwater
- Can be powered by offshore generation (ie wind turbines)
- Doesn’t need freshwater and cuts land use
Space Data Centers
- According to Starcloud
  - “As launch costs fall, data centers in space will leverage continuous solar energy and radiative cooling, rapidly scaling to gigawatts while avoiding permitting constraints on Earth.”
  - Avoiding permitting constraints
  - Scalability