Fastly’s mission is to make the internet a better place for all experiences: fast, safe and engaging, according to Kip Compton, who joined the company as chief product officer in January 2024 before being promoted to CEO in June this year.
Fastly was founded by Artur Bergman, who, while working at user-generated content site Wikia, grew frustrated that existing content delivery networks (CDNs) were not dynamic enough to handle unpredictable updates and continued to serve stale content.
Bergman built and deployed a CDN technology to solve this, which became the foundation of Fastly.
Over time, security issues such as distributed denial-of-service (DDoS) attacks came to the fore. To protect the delivery of applications, content and application programming interfaces (APIs), Fastly added security to its platform, which has since become the fastest-growing part of its business.
Another issue was the trade-off between speed and user experience. While caching supported fast internet experiences, it became jarring for users to visit an e-commerce site without getting a personalised experience.
In response, Fastly created an edge compute capability so that personalisation could be delivered without sacrificing performance.
“Our mission has built up as the company focused on delivering great experiences and what constitutes a great experience,” said Compton.
Platform strategy
While security is the fastest-growing part of Fastly’s business, its CDN capabilities are still bringing in new customers.
“Our strategy is around the Fastly platform,” said Compton. “All of the services we offer are delivered on a unified global platform, and that is important to our customers because it means all of the services work together in a consistent way.
“Historically, CDN is what people came to us for, and that remains the largest part of our business by dollar value. But we are increasingly seeing customers start with security and then adding CDN or compute. Once they’re using the platform, they often start using additional services.”
Mike Shaver, Fastly’s vice-president of security products engineering, said the platform approach distinguishes Fastly from competitors by allowing it to offer a wide variety of well-integrated capabilities.
“I’ve stopped thinking about building security products, but rather about adding security capabilities to the Fastly platform,” he said. “We want to make sure that whatever Fastly capabilities you’re using to accelerate or protect your applications can be interchanged and combined.
Historically, CDN is what people came to us for, and that remains the largest part of our business by dollar value. But we are increasingly seeing customers start with security and then adding CDN or compute
Kip Compton, Fastly
“If you’re using our compute or delivery products, you can secure them the same way with our WAF [web application firewall] product, our anti-DDoS systems, or our AI bot management. All of those pieces can be layered according to the customer’s needs.”
This approach means developers and customers can learn the platform once and apply that knowledge across the entire portfolio.
“Integrating with Fastly is very much integrating with the web,” added Shaver. “We’re a company built on an open approach to technology – HTTP and all the standards that go with that – so products that meet us on those web standards are quite straightforward to integrate. We have a number of explicit partnerships, but also an implicit partnership with pretty much everything else that’s on the web.”
Shaver noted that Fastly’s compute services, which are integrated with the platform’s security capabilities, provide a flexible way to handle, route and pre-process HTTP requests using standard technologies like JavaScript and WebAssembly.
On observability, Shaver said Fastly aims for flexibility. “Our ambition is that you can choose to use our pane of glass to have a complete view of your systems … or you can export that data and use it in your preferred observability system,” he said.
AI, bots and the battle for content
A hot topic for the industry is the tension between the utility of training large language models (LLMs) on vast amounts of data and the rights of intellectual property owners. Compton said the issue came up in all of his recent customer meetings in Australia.
Customers with high-value content want companies like Fastly to protect it, give them insight into which bots are visiting their sites and for what purpose, and to provide the control to allow or block them.
“A lot of the customers we work with are taking very thoughtful approaches to how they monetise that content,” he said. “There’s a big difference between scraping for training data and fetching for real-time responses, and a lot of customers are looking at monetising those differently.”
The traditional ad-supported business model was compatible with search engines, as users would click through to a page, generating revenue for the content owner. This model breaks down with LLMs, which can deliver information to users directly, bypassing the original site and its monetisation.
“We’re in favour of innovation, but we believe it’s important for content owners to be compensated for their content so that their model is sustainable,” said Compton. “We’re really focused on giving them the tools they need to implement the business policies that they want.
“I’m optimistic that we’ll find a way to apportion that value … everyone’s trying to figure this out,” he added.
