Data Is Expensive to Store. And Companies Keep Making More of It

Businesses are generating an exponential amount of data as they integrate AI, 5G, and the Internet of Things (lot) into their tech stack.

This torrent is crushing current storage capacity and guzzling energy as the power needed to run data centers and servers soars. In turn, companies may face a range of new costs that they have yet to fully grasp.

Tirias Research estimates that data center server infrastructure operating costs from generative AI alone will exceed $76 billion by 2028. The costs to the environment could be just as severe.

While data generation surges, and enterprise storage needs are expanding exponentially, the current infrastructure to house data is out-of-date and inefficient.

Companies that don’t overhaul their storage solutions to account for this impending influx may find themselves with high power bills, congested data flows and incompatible security techniques.

Without a plan, energy consumption and carbon emissions will rise in lockstep with data production and utilization.

“Inefficient data center infrastructure will result in companies’ emitting increasing amounts of carbon as compute demand increases,” says Patrick Cassleman, Sr. Director of the Sustainability Center of Excellence at Intel.

Up to 73% of data created by enterprises goes unused but is stored anyway, according to Forrester. AI is part of the problem, but it may also be part of the solution.

Processes with built-in AI capabilities are helping to improve data compression and energy use. The 4th Gen Intel® Xeon® Scalable processor, for example, has a range of AI abilities that make data storage more efficient, and less costly to businesses and the environment.

Companies will either have to generate less data or become more efficient regarding how that data is stored to limit the financial burden and environmental impact of emerging technologies.

The 4th Gen Intel® Xeon® Scalable processors can consolidate five servers worth of data into a single server, freeing up data center rack space by 80% and reducing CO2 emissions by 60%.

By upgrading to processors that are both purpose-built to handle workflows from generative AI, and incorporate the capabilities of AI, data center operators get more control over their environment, particularly around power consumption.

With a new feature called Optimized Power Mode, the 4th Gen Intel® Xeon® AI accelerator engine anticipates the power needs of its customers, enabling servers to save power in the utilization range that customers run the most, providing up to 20% of power socket saving with little to no performance impact, according to Cassleman. The Quick Assist feature also speeds up crypto processing for secure storage.

Mostly, though, enterprises will need to become more economical about what data gets stored and for how long. More efficient processors are one very important step in this equation.

"AI can help determine what data is valuable, necessary, and of high enough quality to warrant storage,” Cassleman says. “Superfluous data can simply be discarded, saving both cost and energy."