NVIDIA & Akamai: Bringing 'AI at the Speed of Now'

Akamai Technologies is making a significant infrastructure investment, announcing the acquisition of thousands of NVIDIA Blackwell GPUs as part of an ambitious plan to build one of the world's most widely distributed AI platforms.

The expansion will see Akamai integrate next-generation NVIDIA Blackwell GPUs into its global footprint, which exceeds 4,000 edge locations.

Rather than concentrating compute power in a handful of hyperscale data centres, the company is embedding AI inference capacity deep within its distributed cloud platform, marking a notable shift in how enterprise AI infrastructure could be deployed.

The focus is on inference, the stage where trained models are put to work in real-world applications.

By positioning high-performance GPUs closer to end users and connected devices, Akamai aims to reduce latency by up to two-and-a-half times, lower bandwidth costs and allow organisations to process and act on data where it is created.

According to a report by MIT Technology Review, more than half of organisations cited latency as the primary barrier to scaling AI.

Addressing the inference challenge

The deployment is designed to support use cases that rely on immediate responses, including autonomous machines, digital health systems, industrial automation and financial fraud detection.

In these environments, even small delays can limit effectiveness, making proximity of compute a potential competitive advantage.

"While hyperscalers continue to push the boundaries of AI training, Akamai is focused on meeting the unique demands of the inference era," says Adam Karon, Chief Operating Officer and General Manager of Cloud Technology Group at Akamai.

"Centralised AI factories remain essential for building models, but bringing those models to life at scale requires a decentralised nervous system. By distributing inference-optimised compute across our global fabric, Akamai isn't just adding capacity.

"We're providing the scale, at minimal latency, that is required to move AI from the laboratory to the street corner and the hospital bed – where the work happens, where the data lives and where the ROI is realised."

Distributed infrastructure for production workloads

Akamai's approach reflects a broader shift in how enterprises are thinking about AI infrastructure. While hyperscale providers continue to invest heavily in large training clusters, businesses are increasingly concerned with how to run models efficiently in production environments.

By embedding NVIDIA Blackwell GPUs directly into its distributed cloud infrastructure, Akamai is positioning itself as an alternative to centralised AI factories and claims businesses could save up to 86% on AI inference compared to traditional hyperscale infrastructure.

The Blackwell architecture is designed to deliver high throughput and improved energy efficiency for demanding AI inference workloads, making it well suited to geographically dispersed deployments.

The company says its architecture dynamically routes workloads to optimal GPU clusters within its network, enabling businesses to fine tune and deploy large language models closer to regional users. This could help address data sovereignty and compliance requirements alongside performance objectives.

Rethinking AI deployment models

The decision to deploy thousands of NVIDIA Blackwell GPUs signals Akamai's ambition to play a central role in the next phase of AI infrastructure development. Instead of competing directly on large-scale model training, the company is betting that inference at the edge will define the commercial impact of AI technology.

For developers, the platform is intended to provide scalable GPU access without the complexity of managing infrastructure across multiple regions. For enterprises, the appeal lies in predictable performance and reduced data transfer costs when serving AI-powered applications at scale.

As AI systems move from experimentation to embedded services in healthcare, manufacturing, retail and finance, the need for distributed, low-latency compute is expected to intensify. By extending GPU acceleration across its global network, Akamai is seeking to ensure that AI models are not only powerful but also practical in everyday operational settings.

If successful, the rollout could help reshape how technology leaders think about deploying AI infrastructure, shifting the emphasis from centralised capacity to globally distributed performance.