Redefining AI Application Delivery

In this discussion with F5’s Ahmed Guetari and Adam Judd, we explore the collaboration between F5 and NVIDIA, which leverages the innovative capabilities of NVIDIA’s BlueField-3 DPUs and F5’s BIG-IP Next platform. This partnership is set to redefine AI application delivery and security, particularly in the high-growth Indian market, by enabling better resource allocation, faster data processing, and robust security enhancements. With AI adoption surging across industries, F5 and NVIDIA’s synergy brings transformative solutions for service providers and enterprises, poised to drive India’s AI-led innovation.

Could you tell us about F5’s recent partnership with NVIDIA and the context behind it?

Ahmed Guetari: This partnership stems from a blend of shared goals and complementary strengths. F5 has a history of supporting customers through significant industry inflexions. We saw it during the dot-com era, the rise of application delivery controllers (ADC), and the recent acceleration of digital transformation, largely due to COVID-19. Another area close to my heart is 5G deployment, which, like AI, is cloud-native.

The AI workload, though much larger in scale, is very similar to 5G’s in terms of data management and efficiency needs. NVIDIA’s accelerated computing platform, powered by their Data Processing Unit (DPU), needed robust software for optimized performance, which we brought with our Big-IP solutions. By modernizing Big-IP and adapting it to DPU, we’re enabling more efficient deployment of AI applications, conserving energy, and allowing multi-tenancy, which helps customers better utilize and monetize their infrastructure.

What are the primary goals of this partnership, and how does it leverage both companies’ strengths?

Ahmed Guetari: NVIDIA is a leader in AI, and they wanted to enhance their customers’ success in accelerated computing. They’ve adopted a best-of-breed approach, working with partners who bring specialized software expertise to complement their hardware. F5’s history in deploying similar applications, combined with our best-of-breed software for managing application delivery, aligns with NVIDIA’s need for effective AI deployments.

India has been embracing AI in significant ways. Could you elaborate on why this collaboration was launched in India?

Adam Judd: India was chosen for the global announcement, highlighting the country’s rapid AI adoption. We’re seeing a massive engagement here—92% of knowledge workers have used AI tools like ChatGPT. The Indian government, through initiatives like the India AI mission, has also committed substantial investments to foster an AI ecosystem. This is why it made perfect sense to launch this collaboration in a region with an immense appetite for AI.

Ahmed Guetari: India is at an intersection of three critical factors. First is knowledge: a high proportion of knowledge workers here have already integrated AI into their work. Second, data volume in India has surged, with companies like Jio processing enormous amounts of sovereign data. Lastly, the government’s substantial investments and proactive stance on AI reinforce India’s readiness to lead in this space.

Are there any specific technologies or products integrated as part of this collaboration?

Ahmed Guetari: Yes, this partnership leverages F5’s Big-IP, which we’ve modernized over recent years to support cloud-native deployments and facilitate the infrastructure needs of major customers. While Big-IP is our flagship product in this collaboration, additional challenges, like security, will also need attention. We’re excited to continue addressing these needs as the partnership evolves.

Our collaboration with NVIDIA transforms how AI applications are managed and secured, empowering organizations to leverage unprecedented performance while meeting evolving security standards.

Beyond security, what other challenges do you foresee for organizations deploying AI in India?

Ahmed Guetari: Two primary challenges stand out. First, managing multiple large language models (LLMs) across sectors. Healthcare, finance, and government will each have specialized LLMs, and having the right infrastructure that can intelligently allocate resources to the relevant LLM is key.

Secondly, India is home to a thriving developer community, and it’s vital to harness their creativity without reinventing the wheel. By making technology accessible and easy to consume, we’re enabling developers to build on existing AI infrastructure rather than starting from scratch.

Adam Judd: Yes, and there’s a crucial distinction between the AI training phase, which is very GPU-intensive, and the inferencing phase, which directly benefits users. Given India’s vast geography, both centralized data centres and distributed micro data centres will play essential roles. This decentralized approach aligns with data localization policies and helps secure data in compliance with local regulations.

How does this collaboration specifically benefit telecom operators and service providers?

Ahmed Guetari: Globally, telcos are finding AI valuable for tasks like customer support, with many hours of calls that can be analysed to improve efficiency and reduce operating costs. Telcos are also increasingly looking to AI for software development augmentation, using AI to fill skill gaps, especially as retaining software talent has been challenging.

In Asia-Pacific, where data localization is stringent, some telcos are deploying AI infrastructure as a service. This is particularly relevant in India and Japan, where governments trust local operators to maintain compliance. We’re supporting them with solutions that align with these sovereign data requirements.

Could you quantify the performance improvements AI-enabled infrastructure brings compared to traditional setups?

Ahmed Guetari: The AI infrastructure’s performance boost is significant. Although specific figures depend on workload requirements, shifting from traditional CPUs to NVIDIA’s DPU-powered architecture with F5’s optimized software enables AI applications to run more efficiently. This not only enhances processing speed but also reduces energy consumption, aligning with green computing goals.

Ahmed Guetari

GM & VP of Products for Service Providers at F5

Adam Judd

SVP, APCJ at F5

aanchalg@cybermedia.co.in