Scaling Your Media Workloads: Introducing Akamai’s New 8-Card VPU Plan

Sarah Walter

Apr 21, 2026

Sarah Walter

Sarah Walter

Written by

Sarah Walter

Sarah Walter is a Product Manager at Akamai, leading cloud compute products with a focus on GPU and accelerated instances. With more than a decade of experience, she brings deep expertise in cloud infrastructure, video processing, and product lifecycle management. She is passionate about building at the intersection of people and technology, and she thrives on solving complex challenges and scaling innovative products that serve media, streaming, and enterprise workloads globally.

Share

Akamai has been powering media for more than 20 years, from our global content delivery network to security to distributed cloud. To support our deep partnership with content and media companies, we’re continually looking for ways to provide the next generation of media workload performance.

In 2024, we introduced our partnership with NETINT to build the first video processing unit (VPU)–powered video encoding in the cloud and launched our one- and two-card VPU plans in 2025.

We are thrilled to announce that we have expanded our Accelerated Compute Instances with the release of our new 8-card VPU plan, powered by NETINT Quadra T1U VPUs. This new offering is designed specifically for media organizations dealing with massive scale to bring maximum transcoding density and flexible compute power to our distributed cloud platform.

A “host's worth” of capacity for on-prem users

Many of our media and streaming customers are already using NETINT VPUs on-premises or within their colocation spaces. For teams accustomed to managing a full hardware capacity, migrating to the cloud can sometimes feel like a compromise in density.

Our new 8-card VPU plan changes that. It is the perfect fit for customers looking to replicate their high-density, on-prem environments in the cloud, allowing customers to harness specialized video processing hardware without the operational complexity and heavy up-front CapEx of hardware ownership.

Right-size your compute and maximize parallel streams

Now Akamai Cloud customers can right-size their compute infrastructure across three carefully tailored VPU plan sizes. This tiered approach allows you to seamlessly scale up or down based on your exact application performance requirements.

By packing eight specialized application-specific integrated circuits (ASICs) into a single instance, this plan unlocks unprecedented stream density. While our initial beta customers were able to achieve up to 30 concurrent live streams on our smaller setups, the massive scale of the 8-card plan has enabled our customers to reliably achieve more than 320 parallel streams simultaneously. This makes it an absolute powerhouse for high-volume video on demand (VOD), live event streaming, and user-generated content platforms.

Robust compute for audio, deinterlacing, and beyond

NETINT VPUs are incredibly efficient at fixed-function video processing, bringing native, hardware-accelerated AV1, HEVC, and H.264 encoding and decoding to your pipeline. However, a complete media workflow requires more than just video compression. That is where the 8-card plan truly shines. 

In addition to the massive VPU density, this plan equips you with a vast amount of general-purpose compute capacity. This helps ensure that you have all the CPU power necessary to handle the intensive media tasks that VPUs natively can't do, such as:

  • Deinterlacing video feeds

  • Processing and encoding complex audio 

  • Dynamic manifest manipulation and server-side ad insertion (SSAI)

  • General application routing and request management

By dedicating the VPUs to media transcoding only, you can completely free up your CPU resources to perform these other mission-critical operations without risking application bottlenecking.

Next steps 

Want to learn more about how Akamai and NETINT are working together to bring video processing at scale to the cloud? Swing by our booth at the NAB Show. Our product team will be on hand with a demo of how these cards work and to answer any questions. When we return, we will share an update on the stories and ways that technology is being used to create and share media globally. 

Get started today

By combining the hardware-accelerated power of an 8-card VPU instance with Akamai’s industry-leading egress costs of just US$0.005 (half a penny) per GB, you can significantly cut your media pipeline's operating costs and scale across regions while delivering peak quality to your viewers.

Ready to test the ultimate cloud transcoding engine? Log in to your Cloud Manager to deploy an Accelerated Compute Instance to see it for yourself.

New to Akamai Cloud? Apply now to get started with complementary cloud credits. 

Sarah Walter

Apr 21, 2026

Sarah Walter

Sarah Walter

Written by

Sarah Walter

Sarah Walter is a Product Manager at Akamai, leading cloud compute products with a focus on GPU and accelerated instances. With more than a decade of experience, she brings deep expertise in cloud infrastructure, video processing, and product lifecycle management. She is passionate about building at the intersection of people and technology, and she thrives on solving complex challenges and scaling innovative products that serve media, streaming, and enterprise workloads globally.

Tags

Share

Related Blog Posts

Cloud
Observability for Akamai Cloud: Get Started with Akamai Cloud Pulse
April 23, 2026
Learn how to use Akamai Cloud Pulse to monitor infrastructure. Explore how to collect and export metrics for Object Storage and NodeBalancers.
Cloud
Why Managed Agents Needs Distributed Infrastructure
April 10, 2026
Anthropic’s Claude Managed Agents requires distributed cloud and security. See how Akamai Inference Cloud and API Security power and protect autonomous AI agents.
Cloud
Scale Faster: A Practical Guide to Building with Akamai Block Storage
April 07, 2026
Akamai Block Storage provides high-performance storage to your cloud virtual machines and containers. Here's what it's used for and how to set it up.