Generative AI technologies, such as ChatGPT, Midjourney and Google’s Bard, have quickly become part of our everyday lives. Their premise is simple: the user prompts the AI system for images or text, and the system outputs the request while also “learning” from the user’s inputs. However, there’s nothing simple about the demand that AI puts on data centers. In fact, this demand is creating a critical need for new infrastructure. Data centers are being built solely for artificial intelligence capabilities, due to the unique requirements for technology, power and cooling.
Danny Martin, senior estimator of Data Centers and Mission Critical at PCL Construction, expects the immense demands of AI to create opportunities to evolve data centers. This could include an increase in the capabilities of colocation data centers – where bandwidth and equipment are rented out to customers — and edge data centers, which are smaller centers that provide greater concentration in high-user areas.
“AI is erupting. We’re hearing that half of all data centers will support AI programs by 2025,” Martin says. “Use of data centers had already increased in the past few years due to people spending more time online for work and entertainment.” As AI becomes more prevalent, requiring more and more storage for its algorithms and models, that demand on data centers will only increase. “This makes it necessary to create new infrastructure or upgrade existing infrastructure to support these data requirements,” Martin says.
The requirements for AI data centers vary depending on the specific needs and scale of the project. Common requirements include computational power, storage capacity, scalability, network infrastructure, data security, cooling, power management, reliability and redundancy, monitoring and analytics, compliance and regulations, and green initiatives. Better data center performance means better data transfer rates and low latency.
“Computational power is the first and foremost consideration of AI data center facilities,” Martin says. Since AI algorithms require vast amounts of processing power for complex models, data centers need to incorporate high-performance components to enable faster data processing, model training and AI-specific tasks. For faster access to external data sources and cloud services, seamless communication between these different components is critical.
“As data is generated, there are very specific solutions for storing and retrieving large datasets promptly and optimizing efficiency,” Martin says. “The data centers’ recovery systems and redundant infrastructure guarantee uninterrupted service and data integrity for the end user.”
However, all this data collection and retrieval requires immense computational power, which generates significant heat. Data centers need robust cooling infrastructure — including liquid cooling and optimized airflows — to ensure efficient cooling while minimizing energy consumption.
As a builder of data centers, PCL must accommodate all these factors while also prioritizing sustainability and embracing green initiatives. Implementing energy efficient technologies, such as power management systems and renewable energy sources, will help minimize the carbon footprint. Additionally, advanced monitoring and analytics tools optimize resource allocation, reducing unnecessary power consumption. PCL is at the forefront of the industry-wide push to set new standards and implement new technologies that reduce power and water usage and offset carbon emissions.
As passionate builders with extensive experience, the data center and mission critical team members at PCL understand the unique requirements of data centers and the critical role they play in the emergence of AI. They know that not every operating data center has the means to host AI, so they monitor trends, technologies and data center designs to provide the most effective infrastructure support available to meet the growing data and computational requirements.
Martin says staying accountable and drawing on PCL’s experience in the data center space includes flexibility, planning and collaboration: “This is the fun part. We get to lead new trends, find new solutions and draw on collaborative industry knowledge. PCL is known for transparency, building trust across the industry and setting new standards with our partners to achieve success.”
Designing data centers is as much about building relationships as it is the construction process, and PCL’s relationships in this market speak for themselves. “A business model built on integrity, commitment, honesty, and professionalism is what we all look for when choosing an organization to collaborate with, whether as a subcontractor or vendor,” says Jason Yankovitz of Salute Mission Critical. “When you can collaborate with an organization driving towards the same goals as you, while doing it the right way day in and day out, you truly win on multiple levels.”
The dynamic growth of AI will create opportunities for new research and development, and for more sustainable and efficient data center structures. PCL’s reputation of trust is vital for these new endeavors. As AI drives data center demand and user connectivity, PCL is ready to provide the necessary support for these projects. and its experienced team understands the need for evolution in the ever-changing landscape of both data centers and AI. Just as importantly, PCL knows that while the success of AI depends on groundbreaking algorithms, it also relies on a builder’s ability to provide solutions for clients.