Industry executives and experts share their predictions for 2024. Read them in this 16th annual VMblog.com series exclusive.
From Data Centers to Wisdom Centers - Architecting an AI-Infused Future
By Melissa Reali-Elliott, Content Marketing Manager of DC BLOX
It's no secret that AI has spurred new demand for data
centers and associated infrastructure. Synergy Research Group recently
predicted that hyperscale data center capacity, driven by AI workloads, will
nearly triple in the next six years. Many existing data centers, including
multi-tenant data centers, will have to be upgraded to handle the processing
requirements of AI. Companies that can implement flexible data center
infrastructure strategies, such as adding modular builds, will be best
positioned to harness the AI revolution, resulting in the anticipated and
necessary growth.
Traditionally, AI systems have been trained on large amounts
of data to identify patterns. Generative AI goes a step further, using complex
systems and models to generate new outputs in the form of images, text, or
audio using natural language prompts. This new model promises massive social
and economic value due to its ability to handle multiple repetitive tasks
quickly and accurately, combined with this ability to break down communication
barriers between humans and machines.
Data centers are already undergoing a profound
metamorphosis, transitioning from conventional repositories to advanced hubs.
AI capabilities promise increased processing capabilities as well as the
infusion of intelligence into every facet of operations.
Rethinking designs:
The crux of this transformation lies in a departure from
dealing with individual data points to cultivating a comprehensive, integrated,
and interconnected ecosystem. The envisioned data center of the future stands
as a nexus where artificial and automated intelligence converge, marking a
significant enhancement in AI capabilities. Anticipation of AI and HPC
requirements has prompted both hyperscale operators and colocation providers to
scale up power for new builds, including high-density power supplies to servers
of 30-50 kW/rack or higher, as opposed to the traditional 5 kw/rack.
In response to this surging demand,
data center operators worldwide are rethinking their architecture and
deployment plans. The impact is profound; hyperscale operators, in particular,
are not merely upgrading equipment, but dramatically altering their designs to
accommodate the unique demands posed by AI workloads. This evolution extends
beyond equipment enhancements, necessitating the creation of flexible, adaptive
infrastructures capable of seamlessly handling the intricate nuances of AI
training and inferencing compute needs. It is evident this will mean more
advanced cooling for heat-intensive processes as well as faster server
innovation.
In addition to power considerations, connectivity will be a
critical factor in designing data centers for AI. High-bandwidth network ports
are essential for seamless data upload, exchange, and inference, especially
given that curated datasets are sourced the world around, requiring even more
infrastructure builds for data transfer, such as cross-continental subsea
cables. For Generative AI applications, low-latency connectivity to inference
points is imperative, particularly for real-time applications and immersive AI
experiences such as AR/VR.
This multifaceted approach to design changes underscores the
intricate balance required to build data centers that not only meet the
immediate demands of AI but are also poised for the ever-evolving landscape of
advanced computing, in a world that was already power-constrained. In fact,
McKinsey projects that data center demand will reach 35 gigawatts by 2030, up
from 17 gigawatts in 2022. Effective power management will remain fundamental
to data center operations, even as we add new processing needs for AI.
Automation as a
Catalyst:
The rapid evolution of technology brings with it an
increased demand for automation. In the face of supply chain issues,
inflationary pressures, and labor shortages, automation becomes not just a
convenience, but a necessity. Automation, in various forms such as scheduling,
monitoring, and maintenance, serves as the catalyst for increased efficiency
and productivity.
Data center automation already exists in many forms,
including scheduling, monitoring, maintenance, application delivery and
troubleshooting inefficiencies in networks and infrastructure. Such automation
gives workers more time to focus on critical, higher-value tasks, further
accelerating data center modernization. This automation trend will continue in
2024, in the pursuit of increased data center agility and operational
efficiency.
AI solves the problem of extracting data from legacy
industrial assets as well as the challenges with interfacing with hundreds of
protocols, vast numbers of connected sensors, disparate data sources and
incompatible data formats that exist in the wild. This is the point at which
industrial equipment and factory systems meet the digital world, resolving
complexities much faster than the human brain can process.
This causes a transformative juncture wherein data centers
will both aid the industrial world by building robust AI infrastructure, but
also be aided by the results it generated.
Incorporating AI for
Efficiency:
Data center owners and operators are beginning to realize
the potential benefits of integrating AI into data center management. AI is
proving beneficial and increasingly being used in operational areas such as
predictive maintenance, anomaly detection, physical security, and filtering and
prioritizing alerts. We have also seen it used for optimization through IT
space temperature control, workload placement, and power optimization.
While many enterprises are still exploring the
possibilities of AI adoption, initial implementations often revolve around
simple tasks, such as administration, customer service, or other avenues to
augment productivity. But we will increasingly see executives push for more
strategic means to leverage AI capabilities to automate and manage mission
critical tasks. To do this effectively and reliably, operators will need to
invest in well-designed and purpose-trained AI models.
Applications of sophisticated machine learning are evident
in providing insights into critical infrastructure components, including power
distribution equipment. AI enables operators to collect vast datasets and
convert them into actionable intelligence, monitoring the efficiency of
components and interconnectivity, ultimately enhancing overall operational
efficiency. Increased data, when processed quickly, can make us more efficient.
Efficiency gains extend beyond facilities and into network
management, where AI tools will play a crucial role in achieving optimal
performance. Addressing concerns such as cyber attacks, AI will aid in
identifying irregularities, determining actionable insights, and mitigating
potential disruptions to security processes.
Generative AI holds potential in the data center industry as
a powerful outlining tool for creating initial documents, models, designs, and
calculations, especially in site planning phases, but this promise will take
longer to prove fruitful than the upcoming year. Despite its strengths, the
inherent limitations of generative AI require an ongoing necessity for human
expertise and oversight, as accuracy cannot be assumed or guaranteed. Even if
employing AI-based tools to draft documents, they must nevertheless undergo
thorough validation by experienced data center professionals before operational
use, meaning that some areas of AI capability still raise questions about the
balance between efficiency gains and risks that necessitate human validation.
In short, the rise of AI will not only impact the ways that
we meet new processing demands, but also how we reimagine operational processes
to streamline tasks, optimize resource allocation, and drive unparalleled speed
and efficiency. The focus is on developing agile, scalable, and cost-effective
solutions that align seamlessly with the ambitions of both a sustainable and
exponentially growing digitized future.
The AI Wisdom Center:
The AI journey our industry is embarking upon will
encapsulate a visionary shift in information processing. Generative AI
represents a paradigm shift that demands a holistic transformation in the way
we approach not only data processing and site design, but the very data itself.
By building for and then utilizing AI, data centers are not only preparing for
the challenges of a digitalized future but actively contributing increased
efficiencies that we will need in order to build the associated infrastructure.
The more information we have, the more we can be informed by
it to make better decisions more quickly to build for sustainable and
innovative growth. This shift signifies the evolution of data centers into
wisdom centers-hubs where data, once actionable intelligence, now evolves into
wisdom that will guide how we reshape the landscape and operational practices
of digital infrastructure.
##
ABOUT THE AUTHOR
Melissa Reali-Elliott has spent over 15 years marketing
digital technologies and is a self-professed data center nerd. She holds
degrees in Marketing, Economics, and Psychology from the University of Central
Florida. Over her career, she has supported organizations specializing in
software applications, IoT, RFID, industrial automation, and power distribution
to utility markets, including smart grid, smart city, and microgrid
applications, as well as critical infrastructure industries such as data centers.
She currently serves as the Content Marketing Manager for DC BLOX, a provider
of Tier III data centers throughout the Southeast US, along with connectivity
and fiber solutions that further build digital infrastructure. She is a regular
contributor to many industry publications and authors the Data-Centered column
for Mission Critical Magazine. She is also an Ambassador for the Nomad Futurist
Foundation, which empowers and educates the younger generations through
exposure to digital infrastructure technologies. As a result of her work, she
has been awarded a 2022 IM100 by iMasons and a 2023 Top 25 Women in Technology
by Mission Critical Magazine.