Slowing down AI innovation isn’t a realistic solution to reduce the massive energy consumption of expanding data centers and digital infrastructure. The AI genie is out of the bottle — and it’s the biggest technology event of our generation, probably multiple generations.
A slowdown wouldn’t just create competitive economic risks, but major security risks, too, as bad actors and nation-states seek an advantage. The fact remains, however, that grids around the world are already strained by data volume and movement, and despite talk of nuclear fusion advances, there’s no single silver bullet for today’s imminent tech power crisis.
What’s more, AI isn’t the only culprit contributing to the overload. The increasing data, processing and movement that accompanies IoT, blockchain, social platforms, gaming and other technologies are also part of the energy problem.
Reducing consumption, cutting carbon emissions and making businesses more sustainable – meaning more environmentally friendly, which makes them more competitive – requires an ecosystem-wide approach that integrates sustainable technology and AI every step of the way.
SustOps – aka GreenOps or CarbonOps – is a new concept spanning technology, business intelligence and other key, data-driven sustainability practices that looks to aid in the process.
From DevOps to SustOps
About 15 years ago, those who write code and develop applications and those who deploy and operate those applications with IT infrastructure knew they had a treacherous waterfall problem — drowning people, performance and sometimes profit. DevOps joined those two worlds together with agile, integrated workflows that made software production faster, flexible and more collaborative. DevSecOps built upon that by emphasizing the importance of designing in security. AIOps and FinOps then appeared on the scene, the former using AI and machine learning to automate and optimize critical operational tasks, and the latter where engineering, finance and business teams collaborate to maximize the value of operations for software development, data management and cloud use. These are valid and meaningful “Ops” transformations, not jargon, that try to improve the ways we build and run technology.
SustOps doesn’t replace DevOps, DevSecOps, AIOps and FinOps. It’s a fourth – potentially overarching – dimension, requiring us to look at operational processes and practices through a business intelligence lens. SustOps will see data center administrators using insights and making trade-offs around operations that may look cost-efficient, but carry a higher net cost due to emissions. SustOps involves monitoring and analyzing data, often with powerful AI, to deliver actionable information on sustainable technical workflows and allow teams to collect relevant data from a variety of systems they might not have been able to combine before, or thought to do so. Now visible for decision-making, real-time reports, analytics and recommendations can trigger automated Scope 1, 2 and 3 actions (reducing direct and indirect emissions) and even Scope 4 actions (avoiding emissions) by connecting data across the data center.
Sustainable hardware and software choices for maximum impact
A prerequisite for SustOps and the larger ecosystem of decarbonization is choosing advanced hardware and software built to decrease physical data sprawl and data’s footprint in terms of storage, compression, deduplication and efficient data movement. All hardware isn’t the same, and what can seem like “little things” make a big difference.
For example, a device delivering 80% energy efficiency versus 75%, at scale, makes a major difference in dollars and kilowatts. Using better data compression technology to store the same data, but with 60% less footprint, has an immediate cascading effect on reducing power consumption and carbon emissions.
Combining different technologies like IoT sensors and digital twins can make a data center more sustainable, for example, by automatically detecting temperature changes that trigger different, optimized modes in a cooling solution. Digital twins paired with augmented reality can also help with sustainability by enabling data center administrators to move through a data center, spot what’s wrong, do root cause analysis and recommend actions to take – all without setting foot inside the data center.
Implementing a horizontal platform layer over all operations enables teams to manage data centers in an integrated fashion focused on sustainable outcomes: reduced energy use, emissions, carbon footprint and planetary impact.
Green coding for data flow and resource optimization
A remarkable and relatively new area for sustainability lies in code itself. AI is now poised to help optimize every line of code and program at scale in a much more elegant, efficient and environmentally sensitive manner. Code running software in the data center and applications in the cloud is often extremely inefficient and energy-intensive. Enter green coding, which is both improved by AI and can improve how AI is added to applications and deployed, reducing compute, storage and network resource usage and optimizing the movement of data across pipelines. Directly impacting energy consumption, it can cut away unnecessary dependencies during software development, improve build management, eliminate excess versioning, decrease repetitive tasks and consolidate tool use along the software supply chain.
Deploying intelligent algorithms also improves data flow and sustainability. For example, algorithms can detect idle states and automatically put devices into a sleep or power-saving mode, then wake them only as needed. They can instantaneously trigger action, sometimes neglected by humans, to regulate what’s in hot storage versus cold storage.
AI for power management and overall integration
AI and data analytics can help optimize smart grid systems for data centers as optimized usage and reserve can help destress a smart grid. A smart grid allows for a two-way flow of energy and information, with digital communication technology on both ends. Energy moves from the source to the consumer — but the consumer is not just a consumer. With local storage technology, smart appliances and smart meters, a data center consumer can hold and return energy back to the source, depending on the needs of the grid.
Another way AI is solving its own energy problem and driving power management innovation is through better battery technology. Researchers recently used AI to design N2116, a new solid-state electrolyte that requires 70% less lithium than competing battery designs. Batteries like this are on the immediate horizon and will allow for a much more cost-efficient and safer backup power supply for data centers. Since batteries are required to store renewable energy from wind turbines and solar panels, they would also serve as a necessary part of a greener electric grid, along with advanced reconductored power lines. On the nuclear front, small modular reactors (SMRs) and microreactors that leverage nuclear fission to generate power will likely be part of a carbon-free future. But it’s really clean, safe nuclear fusion — advanced by AI but still decades away — that will mark the next energy revolution.
AI helps with far more than just power management. It acts as an integrator – looking across software, infrastructure and power and cooling to make the right trade-offs, predict when maintenance will be needed and drive dramatic savings through automation of response.
Circularity across the value chain and outcome-based SLAs
Aligning all of the tech and practices across the data ecosystem with the circular economy is crucial. From the cradle in product conception, R&D and raw materials sourcing to the grave in waste management, carbon capture, recycling and waste heat reuse, circularity needs to be planned for and can be improved by end-to-end data visibility across the value chain.
The move toward SustOps also goes hand in hand with the move toward IT solution providers selling outcomes vs. just products or services. For example, an SLA might guarantee a 30% power savings compared to an enterprise’s existing or an alternative installation. Now, the strength of IT solution providers will be based on their ability to stand behind guarantees, and enterprises will be freed from worrying about individual piece parts of their IT environments.
Powerful new approaches like SustOps will allow us to holistically and meaningfully control the data center’s energy use now and into the future.
More than ever, business growth and customer satisfaction require data-driven sustainability that can answer the challenges of climate change that are upon us.
