How artificial intelligence is changing data center design

With global spending on AI systems set to double between 2023 and 2026, it’s clear that data center capacity will increase rapidly to meet demand.

However, surprisingly many data center operators have hit the brakes on new projects and slowed investment over the past year, with vacant capacity in London falling by 6.3% in 2022-23.

What’s behind this counter-intuitive trend? To explain this, we need to understand some of the issues related to artificial intelligence computing and the infrastructure that supports it

How artificial intelligence is changing data center infrastructure

Data centers have always been built around The CPU power supply is expanded to handle traditional computing workloads. However, AI computing requires a GPU-driven rack, which consumes more power, dissipates more heat, and takes up more space than a CPU of the same capacity. In practice, this means AI computing power often requires more power connections or alternative cooling systems. Can be rephrased as: What this essentially means is that AI computing power typically requires more power connections or alternative cooling systems. Since this is embedded infrastructure, it is built into the data center complex. structures, replacement costs are often extremely high, if not completely economically impossible.

In practice, operators must commit to how much space in new data centers is dedicated to AI versus traditional computing

If you make this mistake and overcommit to AI, Potentially leaving data center operators with permanently underutilized and unprofitable capacity

This problem is exacerbated by the fact that the AI ​​market is still in its infancy, with Gartner claiming it is currently in a hype cycle where expectations are too high of the peak. As a result, many operators are choosing to hesitate in the design phase rather than commit prematurely to the proportion of AI computing in new data center projects.

Take a holistic approach during the design phase

However, operators are acutely aware that if they do not take the risk of delaying investment, they will lose market share and competitive advantage. However, this is a tall order given that many of the fundamentals of data center infrastructure are being rewritten in real time.

To balance the needs of first movers and offset risks, operators need to design their data centers to be artificially The era of intelligent computing offers maximum efficiency and flexibility. This requires a new, holistic approach to design.

1. Involve more stakeholders

No matter how operators decide to differentiate between artificial intelligence and traditional computing, data center sites with artificial intelligence computing capabilities will be much more complex than traditional facilities . Higher complexity often means more points of failure, especially since AI computing has more demands than traditional computing.

So, to ensure uptime over the life of a site and reduce the risk of costly problems, teams need to be more thorough in the data center planning phase.

In particular, the design phase should seek input from the wider team and expertise at the beginning of the project. In addition to seeking expertise in power and cooling, designers should work early with operations, cabling, and security teams to understand potential sources of bottlenecks or failures

2. Integrating Artificial Intelligence into Data Center Operations

Since operators now have AI computing in the field, they should leverage their ability to leverage AI to improve operational efficiency. Artificial intelligence has a long history of adoption in data centers, with the technology able to execute workflows with extremely high precision and quality. For example, artificial intelligence can help with:

• Power usage monitoring and distribution
• Hardware fault detection and prediction By proactively using this technology at every stage of the data center lifecycle, operators can significantly improve operational efficiency and robustness. Artificial intelligence is well-suited to help solve new challenges encountered in adopting the novel and complex layout of these next-generation data centers, such as avoiding false economics through fault detection and predictive maintenance
• 3 Artificial intelligence puts a greater load on data centers during peak hours, such as during training runs or when running enterprise-grade models in production. At this time, AI computing will often far exceed the traditionally expected limits on power consumption, cooling requirements, and data throughput. At the most basic level, this means that the underlying materials of the data center are subjected to more A lot of pressure. If these underlying materials or components are not of high quality, it means they are more susceptible to failure. Since AI computing means a dramatic increase in the number of components and connections in a site, this means that cheaper and lower quality materials that work well in traditional sites may cause the data center running AI computing to grind to a halt
• To avoid false economic risks of saving money, operators should avoid purchasing inferior materials such as substandard cables. These materials are prone to failure and require frequent replacement. More seriously, failure of substandard materials and components often results in plant shutdowns or slowdowns, impacting profitability

Solving Infrastructure Challenges

Although the infrastructure requirements for AI computing may be the main reason why operators delay investment, in the long run, this will not be the case

As the market As uncertainty increases, companies will be split between traditional computing and artificial intelligence computing, moving closer to their "golden fields"

In this case, companies should ensure that they have everything in place in website operations Possible advantages in order to succeed as they learn and grow

This means designing holistically from the beginning, leveraging AI itself to discover new efficiencies in their sites, and investing in investments that can meet the needs of AI Calculate greater demand for quality materials.

The above is the detailed content of How artificial intelligence is changing data center design. For more information, please follow other related articles on the PHP Chinese website!