Running AI Workloads with WebAssembly - Wasm I/O 2025
The technology landscape is experiencing a seismic shift. As generative AI transforms everything from content creation to enterprise workflows, organizations are grappling with a fundamental challenge: how to deliver AI capabilities efficiently, securely, and at scale.
At the recent WASM I/O 2025 conference, industry leaders (including our own CTO Radu Matei) gathered to discuss the current state of AI compute - and WebAssembly’s role in moving it forward.
The Panel’s Key Takeaways
- Is it production ready? The consensus was clear: WebAssembly has moved beyond experimental use cases and is actively powering production AI workloads today.
- WebAssembly’s role as a universal abstraction layer for hardware and GPU/CPU architectures offer a lot of advantages to teams trying to address different deployment targets
- Security advantages of Wasm binaries (sandboxed execution model) are ideal for adding a fundamental layer of security across AI workloads
- Wasm is enabling exciting new deployment patterns, from edge devices and browsers to Kubernetes orchestration.
The panelists shared real-world examples ranging from drone control systems to enterprise-scale video processing pipelines, demonstrating that WebAssembly AI applications are already operating at significant scale.
As both WebAssembly adoption grows, more and more AI-specific use cases are appearing rapidly - like the panelists, Fermyon is very excited to see these technologies mature and look forward to next year’s Wasm I/O to see how these predictions play out.
Building AI Apps with Spin
For developers looking to explore WebAssembly for AI apps, Spin provides an ideal starting point.
Now a CNCF Sandbox project, Spin is an open-source framework for building serverless WebAssembly apps, abstracting away the complexity of Wasm runtime management, and providing built-in support for AI inference through integrations with popular models and inference engines.
Check out AI app code examples in our Developer Hub, and make sure you are following the instructions on how to set up your development environment.
Running AI workloads on Fermyon Wasm Functions
When you’re ready to move your Wasm applications to enterprise-scale deployment, Fermyon Wasm Functions delivers the optimal performance that production AI workloads demand.
Built on Akamai’s global edge infrastructure, the platform combines the developer experience of Spin with enterprise-grade cloud features - automatic scaling, global distribution, and built-in security controls.
