Juniper Networks has announced a new solution purpose-built for neocloud, traditional SPs and other AI cloud providers who are deploying and managing GPUaaS (GPU-as-a-Service) and AIaaS (AI-as-a-Service) offerings. The new solution accelerates time-to-market, simplifies operations, and lowers the cost of managing these services, all while delivering high-performance data center networks that are optimized for multi-tenant environments. This is the latest Networks for AI solution introduced by Juniper, who is experiencing strong momentum in the AI data center space as companies worldwide seek for industry-leading switching, routing, security and automated operations.
Enterprises are turning to neocloud, traditional SPs and other AI cloud providers for on-demand AI services to reduce costs and improve time-to-market. These companies build optimized infrastructure to deliver better performance and offer more competitive pricing, particularly by enabling a pay-as-you-go model for expensive hardware like GPUs, making AI development more accessible. Such providers are typically more agile and responsive, offering cutting-edge AI technologies with greater flexibility and customization. Some also focus on data sovereignty and privacy, catering to organizations with strict regulatory requirements.
Juniper’s solution provides a powerful and secure way to deploy highly optimized, cost-effective and multi-tenant cloud-based AI services. It delivers high-performing, scalable and secure networks for cloud-based AI services such as training and inference with Retrieval Augmented Generation (RAG). It is quick-to-deploy and easy-to-manage, providing time-to-market advantages to AI cloud providers. The solution today consists of QFX Series Switches, PTX Series Routers and SRX Series Firewalls managed via Juniper Apstra data center assurance software and Mist AI.
The solution delivers these key benefits:
Automation accelerates deployment and simplifies operations
By leveraging a combination of intent-based networking and Mist AI, Juniper simplifies Day 0/1/2+ operations and reduces ongoing costs by up to 85 percent in some instances, with up to 10x reduction in deployment time. Through an integration with Red Hat OpenShift, even more automation benefits can be achieved in Kubernetes environments.
Zero Trust Security and Multi-tenancy protects users and data
Juniper’s Zero Trust DC Security portfolio, along with EVPN VXLAN capabilities within Juniper switches, provide multi-tenancy capabilities, and protects the AI infrastructure, models and confidential data from internal and external threats. Juniper’s SRX 4700 next-generation firewall is a power-efficient, 1 U device designed for service providers, cloud providers and large enterprises. It delivers the industry’s highest firewall throughput per rack unit, up to 1.4 Tbps, and supports 400 Gbps interfaces with wire speed MACsec to safeguard data in motion.
Validated solutions provide confidence
With Juniper’s Ops4AI Lab, managed providers can validate their AI models prior to actual deployment across various networking, compute and storage platforms and configurations. In addition, Juniper Validated Designs (JVDs) exist for multi-vendor AI blueprints, including NVIDIA and AMD accelerated computing, Weka and VAST storage. These tools ensure confidence and expedite GPUaaS and AIaaS deployment times.
Open flexibility avoids vendor lock-in
Juniper has the only multi-vendor solution for DC fabric management and automation, making it the most flexible to design and simplest to manage. Juniper also has unique capabilities to optimize AI workload performance over Ethernet, allowing providers to use open and proven technologies and products that avoid vendor lock-in. With a runway to 1.6 Tbps/port switches and multi-vendor support for GPU-agnostic systems, Juniper reduces costs, assures faster innovation, maximizes design flexibility and avoids supply chain challenges.
Continued momentum in AI DC
As a leader in AI-Native Networking and AI technologies, Juniper’s early bet on “Networks for AI” is paying off with strong market success. First to ship 800G Ethernet switches among OEMs, Juniper now commands a leading 49 percent share of the 800G OEM market, according to 650 Group’s Q1-Q3 2024 revenue shipment report published in December 2024. With AI-optimized solutions, Open4AI Lab and white-glove service to customers, Juniper has been winning many AI DC customers. Juniper AI data center solutions are powering networks from the world’s largest ACP, with hundreds of thousands of GPUs, to enterprise deployments.
“Managed AI services, such as GPUaaS and AIaaS, have been growing rapidly. To capture the demand, neocloud, traditional SPs and other AI cloud providers need to move fast and deliver exceptional value. Juniper is excited to deliver a purpose-built solution for these providers, expediting their journey towards offering unique and compelling services with the performance, security, automation and open flexibility their customers demand,” says Praveen Jain, Senior Vice President & General Manager, Data Center & AI, Juniper Networks.
