Artificial intelligence computing company Blaize Holdings Inc. today announced the launch of Blaize AI Services, a new platform designed to help AI infrastructure providers and enterprises deploy production-ready, application-level AI services without building the underlying AI stack from scratch.

Many organizations have run an AI pilot, but far fewer have turned it into something that reliably delivers value at large scale. Blaize argues that the gap in getting from experiment to production is where the real cost and complexity kick in.

Blaize AI Services addresses the gap by combining modular application programming interfaces, hybrid computing and forward-deployed engineering into a single platform that makes AI easier to operationalize and scale. The result, according to Blaize, is lower cost per AI interaction, faster time to value and AI that grows as a repeatable business capability rather than an expensive one-off project.

The new AI Services offering also assists with the issue of bottlenecks in AI rollout, such as getting from working models to operational services that can scale up reliably and economically. Providers that want to make money from AI infrastructure are often left stitching together fragmented tools, specialized inference functions and operational workflows. In offering application-level APIs and deployment support that makes AI easier to operationalize, Blaize says, AI Services also helps address bottlenecks that occur.

Under the hood, AI Services is designed around hybrid inference economics to decompose high-level tasks intelligently and schedule the components across Blaize accelerators and GPUs based on cost, power and performance targets. The aim is to help improve utilization while supporting a broader range of workloads.

The open, hybrid architecture is built for heterogeneous environments, with standard APIs and tools that integrate into existing infrastructure rather than forcing platform replacement.

“AI adoption does not stall because of model availability,” said co-founder and Chief Executive Dinakar Munagala. “It stalls in the last mile between pilot and production. Blaize AI Services is designed to help our customers deploy application-level AI services faster, operate them more efficiently and monetize infrastructure through repeatable, revenue-generating offerings.”

The platform is expected to support application-level AI services across vision, video, document processing, speech and various extended use cases of multimodal workflows.

AI Services is also designed to help customers introduce flexible commercial models, such as usage-based pricing and outcome-based services to create a path to recurring AI revenue beyond traditional infrastructure leasing or hardware sales.

Key planned benefits include a faster path from pilot to production, lower infrastructure complexity, improved hybrid compute efficiency, new recurring revenue opportunities and deployment support for the last mile.

“Providers and enterprises do not need more disconnected AI components,” Munagala added. “They need a production-ready way to deliver business outcomes. With Blaize AI Services, we are packaging the core elements required to operationalize AI into modular services that can run across hybrid environments and scale with customer demand.”

Blaize notes that AI Services builds on its broader strategy to deliver practical AI infrastructure that connects silicon, software and deployment into a unified platform. The platform emphasizes open integration, sovereign deploy options and support for both provider and enterprise operating models, including cloud providers, data center operators, system integrators, government organizations and large enterprises.

Blaize is a Nasdaq-listed company that went public via a special-purpose acquisition company merger in January 2025.

Image: Blaize