Menu

The Environmental Impact of Accelerators

Overview: An emerging area of development in IT has been new accelerator devices such as DPUs and smart NICs that offload data flow operations away from the CPU. The main discussion around these technologies has been around removing bottlenecks and increasing performance. This blog examines data flow acceleration devices through the lens of a different potential benefit – environmental efficiency.

The Environmental Impact of Accelerators

Analyst Take: Data flow acceleration has become an area of growing interest within IT. The quest for greater performance is an ongoing, never-ending journey – and it is one with multiple parts. Datacenters have seen improvements in multiple areas that improve data access including high bandwidth networking, solid state storage devices, and NVMe protocols – both to devices and as a fabric. All of these advancements have accelerated performance and enabled faster data access, but as one area improves, the next area becomes the new bottleneck.

The current area of focus is on the CPU’s ability to handle data flow, such as storing and retrieving data. By offloading data handling functionalities, the CPU is freed up to make computations. This has led to the development of new data flow acceleration technologies, and there are multiple approaches to handling this issue including data processing units (DPUs), storage processing units (SPUs), smart NICs, and computational storage. Examples of these technologies include NVIDIA BlueField DPUs, Fungible DPUs (recently acquired by Microsoft), Nebulon SPU, and XILINX Computational Storage.

Evaluator Group has discussed in depth the benefits of many of these technologies with regard to the performance or economic drivers behind them. But, with the growing concern around sustainability initiatives in IT, organizations must now also weigh the environmental impact of new technologies, such as data flow accelerators.

An initial reaction may conclude that additional accelerators would have an increased environmental impact. New accelerators involve new hardware, each with its own environmental impact throughout its lifecycle. Adding new computational devices requires using more mined materials as well as adding another component that requires power. It also represents an additional piece of technology that will inevitably be refreshed and lead to additional e-waste.

So, are data flow accelerators simply a performance enhancer that come with an additional environmental strain? Not necessarily.

The environmental benefits of data flow acceleration become clearer when considering energy usage, and remembering what it is that these technologies are specifically trying to accomplish – offloading operations from the CPU. The biggest environmental concern around IT is typically energy usage. By offloading certain processes to specialized hardware designed specifically for those tasks, the processes can be achieved more efficiently and as a result, with less power consumption. Meanwhile, the CPU can be freed up to focus on – and consume power on – computation processes in which it’s better suited.

A similar comparison can be made with GPUs – an offload that is more commonly understood. GPUs are well known to be energy intensive devices. They are used, however, because they provide an efficiency over CPUs for certain tasks, specifically for parallel computations. When used correctly, GPUs can offer a greater use of power for certain workloads than forcing a CPU to perform tasks it is not designed for, all while freeing up the CPU for more appropriate tasks. DPUs, or other accelerators, can be used in a similar way – accomplishing specific tasks more efficiently than the CPU could.

Use of data flow accelerators is still a developing area, and certainly one to keep an eye on. The technology aims to move the latest performance bottleneck, and it may prove useful in increasing energy efficiency as well. There are multiple different approaches being developed, each of which functions differently and may provide different benefits or drawbacks. IT organizations should look to understand the performance, economic, and environmental characteristics of these different accelerators to understand which ones may be the most appropriate for their needs. Specialized accelerators will not be a universal solution to the environmental concerns plaguing IT, however when used appropriately they may become an additional tool to increase power efficiency.

More insights from Evaluator Group:

Check out Evaluator Group’s research on Data Flow Acceleration and Sustainability here:

Technical Insight: Addressing the Next Bottleneck in Storage

Storage, Sustainability, and ESG Reporting

ESG in IT: Here to Stay or the Latest Fad?

Disclosure: Evaluator Group, wholly owned by The Futurum Group, is a research and analyst firm that engages or has engaged in research, analysis and advisory services with many technology companies, including those mentioned in this article.  The author does not hold any equity positions with any company mentioned in this article.   Analysis and opinions expressed herein are specific to the analyst individually.

Author Information

Mitch comes to The Futurum Group through the acquisition of the Evaluator Group and is focused on the fast-paced and rapidly evolving areas of cloud computing and data storage. Mitch joined Evaluator Group in 2019 as a Research Associate covering numerous storage technologies and emerging IT trends.

With a passion for all things tech, Mitch brings deep technical knowledge and insight to The Futurum Group’s research by highlighting the latest in data center and information management solutions. Mitch’s coverage has spanned topics including primary and secondary storage, private and public clouds, networking fabrics, and more. With ever changing data technologies and rapidly emerging trends in today’s digital world, Mitch provides valuable insights into the IT landscape for enterprises, IT professionals, and technology enthusiasts alike.

Related Insights
Snowflake's SnowWork Targets the Gap Between Data Insight and Business Action
March 25, 2026

Snowflake’s SnowWork Targets the Gap Between Data Insight and Business Action

Brad Shimmin and Nick Patience explore Snowflake’s Project SnowWork and how the Agentic Enterprise Control Plane turns the AI Data Cloud into a "system of action" for autonomous workflows across...
Mistral Forge Takes Aim at RAG. But Who Actually Needs Custom Models
March 25, 2026

Mistral Forge Takes Aim at RAG. But Who Actually Needs Custom Models?

Nick Patience, AI Platforms Practice Lead at Futurum, examines Mistral Forge, a custom enterprise AI model training platform, and argues that while its approach is sound, the addressable market may...
Oracle Positions AI Database 26ai to Lead $1.2 Trillion Market by Bridging the Agentic Reasoning Gap
March 25, 2026

Oracle Positions AI Database 26ai to Lead $1.2 Trillion Market by Bridging the Agentic Reasoning Gap

Brad Shimmin and Keith Kirkpatrick of Futurum explore Oracle's pivot to agentic plumbing. Oracle is embedding autonomous reasoning directly into Oracle AI Database 26ai to solve the enterprise data latency...
Grounding the Agentic Mandate As the Semantic Layer Market Eyes 19% Growth, Microsoft Fabric IQ Targets Leaders Prioritizing AI Investment
March 20, 2026

Grounding the Agentic Mandate: As the Semantic Layer Market Eyes 19% Growth, Microsoft Fabric IQ Targets Leaders Prioritizing AI Investment

Brad Shimmin, VP and Practice Lead at Futurum, shares insights from FabCon and SQLCon 2026 on how Microsoft is leveraging the new Database Hub and Fabric IQ to unify transactional...
NVIDIA GTC 2026 Day 1 - Can NVIDIA’s Ecosystem Accelerate the Inference Inflection
March 18, 2026

NVIDIA GTC 2026 Day 1 – Can NVIDIA’s Ecosystem Accelerate the Inference Inflection?

Brendan Burke, Research Director at Futurum, breaks down NVIDIA GTC 2026 Day 1, highlighting the NVIDIA Vera Rubin platform, the $27B Nebius-Meta deal, and how partners like HPE and Micron...
NVIDIA Agent Toolkit
March 16, 2026

At GTC 2026, NVIDIA Stakes Its Claim on Autonomous Agent Infrastructure

Nick Patience and Mitch Ashley, analysts at Futurum, examine NVIDIA's Agent Toolkit announcements at GTC 2026, covering NemoClaw, AI-Q, the Nemotron Coalition, and what they mean for enterprise agentic AI...

Book a Demo

Newsletter Sign-up Form

Get important insights straight to your inbox, receive first looks at eBooks, exclusive event invitations, custom content, and more. We promise not to spam you or sell your name to anyone. You can always unsubscribe at any time.

All fields are required






Thank you, we received your request, a member of our team will be in contact with you.