Guests: Ari Weil | Matt Butcher
Companies: Akamai | Fermyon
Show Name: An Eye on AI
Topic: Edge Computing, Kubernetes

The edge computing landscape just shifted. Akamai’s acquisition of Fermyon signals more than a consolidation play—it represents a fundamental rethinking of how applications will be built, deployed, and scaled in an AI-driven world. For developers tired of cold start penalties, security vulnerabilities, and inflexible runtimes, this deal brings WebAssembly-based serverless to a global network built for speed and resilience.

In a recent conversation, Ari Weil, VP of Product Marketing at Akamai, and Matt Butcher, CEO and Co-Founder of Fermyon, (now VP of Product at Akamai), walked through why this acquisition matters—not just for their companies, but for the broader shift toward edge-native application architectures.

From Helm to WebAssembly: Fermyon’s Journey

Matt Butcher’s career reads like a roadmap of cloud-native evolution. As one of the original creators of Helm, the package manager for Kubernetes, Butcher helped define how developers deploy and manage containerized applications. But containers, for all their strengths, weren’t designed for the kind of ephemeral, event-driven workloads emerging at the edge.

“Serverless-style workloads—where there’s no server running all the time—needed a specialized compute layer,” Butcher explained. “Docker is great for long-running processes, but we felt like workloads that start on demand, handle a request, and shut down needed something faster and lighter.”

That realization led Butcher and nine colleagues to leave Microsoft on the same day, four and a half years ago, to start Fermyon. Their goal: build the next wave of serverless computing using WebAssembly. The result was Spin, an open-source developer tool now part of the Cloud Native Computing Foundation (CNCF), and later SpinKube, a Kubernetes runtime optimized for WebAssembly workloads.

But speed alone wasn’t enough. “If you can cold start in under half a millisecond but still face 300 milliseconds of network delay, it doesn’t matter how fast your runtime is,” Butcher said. That insight led Fermyon into partnership with Akamai—a company with over 4,300 edge locations worldwide.

The Customer Problem: Beyond Lightweight Functions

Akamai has been in the edge computing business for decades, evolving from content delivery to dynamic application logic. Early solutions like EdgeSide Includes, XML-based configuration, and eventually EdgeWorkers gave developers ways to run code closer to users. But as applications became more complex, those tools hit limits.

“Developers use different languages. Languages evolve quickly. We needed something where they could code in the language of their choice and consider not just lightweight processing, but full edge-distributed applications,” Weil said.

That’s where Fermyon’s approach stood out. By using WebAssembly—a portable, language-agnostic runtime originally designed for the browser—Fermyon enabled developers to write in Rust, Go, Python, JavaScript, or any language that compiles to WebAssembly. More importantly, Spin’s developer experience was designed to get from “blinking cursor to deployed application in two minutes or less.”

“We wanted to make the initial experience so smooth that developers immediately moved on to solving real problems, not wrestling with tooling,” Butcher noted.

For Akamai’s customers—spanning commerce, manufacturing, travel, and publishing—the combination of Fermyon’s runtime flexibility and Akamai’s global reach opened new possibilities. “It was a match made in heaven,” Weil said. “But the biggest thing holding us back was that we weren’t one company. That’s why the acquisition made sense.”

What Integration Actually Means for Developers

The technical integration goes deeper than just hosting Fermyon functions on Akamai’s network. Akamai’s platform now includes Infrastructure-as-a-Service (IaaS) through its Linode acquisition, managed container services, CDN capabilities, bot protection, API security, and AI inference tools. Fermyon functions can now tap into all of it.

“My team has always been compute nerds,” Butcher said. “Now we’re landing in an environment where we can wire up optimized storage, relational databases, high-performance AI inferencing, and security services—all without reinventing the wheel.”

Weil emphasized that security isn’t bolted on—it’s integrated from the start. “We want security to be part of how you design and develop, not something you retrofit after deployment,” she said. Akamai’s approach includes continuous API discovery, automated firewall rule configuration, and observability that helps developers decide which policies to apply based on real traffic patterns.

WebAssembly’s sandboxing model adds another layer of defense. Originally designed to run untrusted code in web browsers, WebAssembly isolates workloads more effectively than traditional container runtimes. “It was built for the most attack-heavy surface we’ve ever invented—the browser,” Butcher said. “We took that battle-hardened security model and brought it server-side.”

Real-Time Use Cases: From Retail to AI Inference

The practical applications are already emerging. Retailers are using Fermyon functions to dynamically render pages based on whether the visitor is a search engine crawler or a human shopper. As zero-click searches powered by large language models (LLMs) become more common, ranking well means serving AI-optimized content instantly.

“Retailers are thinking about how to create engagement videos, social media ads, and conversational content that feeds LLMs with legitimacy,” Weil explained. “It’s no longer about blue links on Google—it’s about what the LLM decides to surface.”

Manufacturing and supply chain companies are exploring how to run AI-powered optimization at the edge, reducing fraud and improving logistics without routing everything through centralized data centers. The key advantage: inference happens close to the data source, cutting latency and reducing cloud egress costs.

Butcher pointed to a longer-term vision where applications intelligently distribute themselves. “WebAssembly’s component model allows you to run one piece of code close to the database and another piece close to the user, optimizing for both,” he said. “That’s a fundamentally new way to think about application architecture.”

Open Source at the Core

Despite the acquisition, Fermyon’s open-source commitments remain intact. Spin and SpinKube are now CNCF projects under open governance. Fermyon also continues contributing to the Bytecode Alliance, an industry consortium that develops WebAssembly standards and reference implementations.

“Developers need open-source tooling,” Butcher said. “It gives them the freedom to see how things work, fix bugs, and add features. That’s something Akamai shares.”

Akamai has been a CNCF sponsor and recently added support for the Linux kernel on its platform. “With Matt coming in, we’re looking at how we can improve community engagement across our portfolio,” Weil added.

What Comes Next

The combined roadmap focuses on three areas: tighter developer tooling, expanded API access, and new deployment models for edge functions. Akamai’s managed container service, which can deploy applications across thousands of edge locations, will integrate with Fermyon’s runtime to enable truly distributed workloads.

“We’re not just thinking about how developers build apps today—we’re thinking about how they’ll build them as AI tools evolve,” Weil said. “We’re bridging the gap from toy-level AI coding tools to real-time agent deployment based on business requirements.”

For Butcher, the vision has been consistent since day one. “We’ve always believed compute should be distributed, not centralized,” he said. “Now we have the infrastructure to prove it.”