A10 launches AI Gateway to control model use & cost

A10 Networks has launched A10 AI Gateway, with availability expected in the fourth quarter of 2026.

The product is designed as a central management layer for organisations using multiple AI models, agents and applications. It brings routing, cost management and governance into one system as companies seek tighter oversight of expanding AI use across their operations.

Businesses have been adding generative AI tools and AI agents to internal and customer-facing software at a rapid pace. That expansion has created a more fragmented environment, with different teams using different model providers and making it harder for management and security teams to track spending, usage patterns and policy compliance.

A10 AI Gateway is intended to address that by giving companies visibility into which models are being used, who is using them and what each request costs. It also applies access controls and usage policies across departments to help organisations manage both budgets and security.

Control layer

The gateway uses identity-based access policies linked to local directory systems, allowing companies to assign different models and rules to different user groups. In practice, that means an engineering team and a finance team could be directed to separate models or be subject to different usage limits.

The system also analyses incoming AI requests and routes them accordingly. Simpler tasks can be sent to lower-cost models, while more demanding tasks can be directed to models better suited to advanced reasoning.

Cost controls are also central to the product. A10 said the gateway tracks spending for each request in dollar terms, supports budget controls by model and team, and lets customers set hard limits, alerts and request-rate controls.

That reflects a broader shift in the AI market, where companies are moving from experimental deployments to operational use and facing pressure to justify spending. As more organisations use several model providers at once, finance and technology leaders are increasingly focused on governance tools that show where money is going and whether internal rules are being followed.

Security focus

A10 is linking the gateway to its broader AI security portfolio, including TrojAI, A10 AI Firewall and ThreatX. The company said this gives customers oversight across pre-deployment testing, runtime protection and controls at the network and application programming interface layers.

The product can be deployed as software on its own or with hardware, and it can run inside a customer's own environment. That includes on-premises systems, private cloud settings and air-gapped environments, which are often used by organisations with stricter data-handling or sovereignty requirements.

Keeping the gateway inside a customer-controlled environment is a notable part of the offering. Many businesses in regulated sectors are weighing the benefits of AI tools against concerns over where data is processed, how model access is governed and whether external services create new security or compliance risks.

Dhrupad Trivedi, President and Chief Executive Officer of A10 Networks, outlined the company's view of the market in a statement accompanying the launch.

"Enterprises are being asked to move fast on AI and stay in control at the same time. AI use is spreading faster than teams can discover, govern, or secure it, driving up cost and leaving no single place to set the rules. The A10 AI Gateway gives organizations one intelligent control plane to see how AI is being used, to govern who can access which models, and match every request to the right model at the right cost. Our integrated portfolio gives customers a complete, sovereign way to put AI to work at scale. That is how enterprises adopt AI with confidence instead of trading speed for control," said Trivedi.

A10, based in San Jose, California, sells network and security products and says it serves more than 7,000 customers worldwide. The launch places the company more directly in the emerging market for AI operations management, where vendors are trying to help businesses monitor model use, control costs and apply security rules as AI becomes part of day-to-day corporate technology estates.