A10 Networks Unveils Advanced Solutions for AI Infrastructure Security and Performance

TL;DR:

• A10 Networks is introducing new solutions specifically designed to secure and optimize the performance of AI and LLM inference environments.
• A new AI firewall capability directly targets emerging cyber threats unique to LLMs, such as prompt injections and sensitive data disclosure, by inspecting traffic at the prompt level.
• The company is also launching predictive performance analytics, which function as an early warning system to proactively identify and mitigate network congestion before it impacts service.
• These GPU-accelerated solutions aim to address the core challenges of latency, security, and operational complexity inherent in the widespread deployment of AI infrastructure.

As organizations worldwide accelerate the deployment of new artificial intelligence applications and construct new AI-ready data centers to enhance operational excellence, the need for robust supporting infrastructure has become paramount. These environments demand ultra-high performance to deliver the real-time responses expected from AI and Large Language Model (LLM) inference tasks, alongside novel cybersecurity solutions to safeguard them from an evolving threat landscape.

To address these nascent challenges, A10 Networks (NYSE: ATEN) is demonstrating new AI firewall and predictive performance capabilities designed to prepare and protect these new AI ecosystems. The solutions are being showcased at the upcoming Interop Tokyo conference, themed “AI Steps into Reality,” taking place from June 11-13, 2025.

Proactively Preventing and Mitigating AI-Specific Cyber Threats

A10 Networks is announcing a significant advancement in cybersecurity with its new AI firewall capabilities. This solution can be deployed in front of the APIs or URLs that expose LLMs to users, whether they are custom-built models or developed atop commercial platforms like OpenAI or Anthropic. Constructed on an edge-optimized architecture with GPU-enabled hardware, these capabilities are engineered to protect AI LLMs at high performance and can be integrated into any infrastructure as an incremental security layer.

These capabilities are designed to prevent, detect, and mitigate AI-level threats. They enable customers to test their AI inference models against known vulnerabilities and help eliminate them using A10’s proprietary LLM safeguarding techniques. Critically, the AI firewall inspects both request and response traffic at the prompt level, allowing it to detect threats such as prompt injections and the unauthorized disclosure of sensitive information. By enforcing specific security policies at this granular level, it effectively mitigates these unique and potentially damaging threats.

Delivering a Real-Time Experience with Predictive Performance

Beyond security, A10 Networks continues to focus on delivering superior performance and resilience for AI and LLM-enabled applications. This is accomplished by offloading processor-intensive tasks like TLS/SSL decryption, implementing advanced caching, and optimizing traffic routing. These functions are complemented by the delivery of actionable insights that help network operators maximize availability and performance.

A key innovation in this domain is a new capability for the early detection of network performance issues, which functions much like an early warning system. This system helps identify near-term network congestion or capacity deficiencies, empowering customers to take proactive measures before an issue becomes critical and impacts user experience. This predictive capability is instrumental in preventing unscheduled downtime and facilitating strategic planning for optimal network performance. To achieve the necessary processing speed, the predictive performance analytics will run on A10 appliances powered by GPUs, enabling the rapid analysis of vast data sets to identify anomalies well ahead of time.

A Holistic Strategy for AI Infrastructure

Together, these integrated AI security and infrastructure performance capabilities offer a holistic solution. They are designed to simplify management, leverage broader intelligence for the accurate detection of threats, and ultimately help organizations deliver an optimal and secure customer experience for their AI-driven services.

“Enterprises are deploying and training AI and LLM inference models on-premises or in the cloud at a rapid pace. New capabilities must be developed to address three key challenges of these new environments: latency, security and operational complexity,” stated Dhrupad Trivedi, president and CEO of A10 Networks. “With over 20 years of experience in securing and delivering applications, we are expanding our capabilities to deliver on these needs to provide resilience, high performance and security for AI and LLM infrastructures.”

For more information on the company’s strategy, you can learn more about A10’s strategy for securing AI environments on their official website.