NETINT Technologies outlines a new operating model called VPU-as-a-Service for cloud video processing, which uses purpose-built Video Processing Units (VPUs) to enhance efficiency for complex and high-volume transcoding workloads. This approach aims to improve codec economics and facilitate practical AV1 adoption by separating the execution layer from the main workflow logic, offering specialized hardware acceleration via a cloud-style deployment. The article positions VPU-as-a-Service as a balance between flexible CPU-based transcoding and fully managed services, allowing engineering teams to retain workflow control while benefiting from dedicated video hardware.
The transition to compute-heavy codecs like AV1 has created an efficiency gap in traditional cloud transcoding. VPU-as-a-Service fills this by providing the cost advantages of ASICs with the elasticity of the cloud, allowing platforms to scale high-density live and VOD workloads without massive CPU expansion. This shift indicates a broader industry move toward workload-specific infrastructure, mirroring how GPUs have evolved for AI. For strategists, it turns codec selection from a bandwidth-only decision into a flexible infrastructure play. Watch for the emergence of similar 'as-a-Service' models for other specialized streaming tasks like computer vision or server-side ad insertion.
The rise of VPU-as-a-Service comes as major streaming platforms reach a tipping point with next-generation codecs. Per industry reports from early 2026, AV1 has effectively become the 'present' of streaming rather than just its future; Netflix confirms the codec now accounts for roughly 30% of its total viewing and is expected to become its most-used format by the end of the year. This surge is driven by near-universal device support, with approximately 88% of new large-screen devices certified since 2023 supporting native AV1 playback. These benchmarks put immense pressure on processing layers to manage AV1's high computational complexity without ballooning operational expenditures.
In response to this demand, Akamai expanded its partnership with NETINT in April 2026, launching high-density 'Accelerated Compute' instances featuring 8-card VPU plans. These instances, powered by NETINT Quadra T1U units, are designed to replicate high-density on-premises environments, allowing broadcasters to run up to 30 concurrent live 1080p streams per VPU. According to Akamai, this approach can reduce transcoding costs to a tenth of traditional CPU-based platforms. Concurrently, Synamedia has integrated similar elastic caching capabilities through its Fluid Edge CDN and NetActuate partnership, enabling streamers to deploy delivery and processing capacity across 45 global points of presence on demand.
This infrastructure evolution is also being recognized at the highest technical levels. In 2024, NETINT’s Quadra VPU received a Technical Emmy Award alongside Meta and Google for its contributions to cloud video acceleration. As hyperscalers and edge providers move away from general-purpose compute, the focus has shifted toward energy-smart silicon. Current industry data suggests that switching from CPU-only clusters to VPU-accelerated servers can remove nine out of ten physical servers from a data center, directly addressing the growing pressure on streaming services to reduce their carbon footprint and power consumption in the face of rising energy costs.
Add StreamingMeme as a preferred source on Google to see more of our streaming news at the top of your Search results.
Add as preferred source
