S9 E1: Matt Levine, CacheFly

Episode Summary

Executive Summary: Matt Levine tells the origin story of Cashfly, a CDN built by experimenting with anycast and distributed web servers before self-serve cloud products were common. He explains how the company grew from a two-pop data center MVP into a global infrastructure business, why clueful engineers and small-team efficiency mattered, and how recent trends like cloud, 5G, and larger content libraries are shifting CDN value toward automation and cache-miss performance.

Main Topics: Cashfly’s origin and early strategy (Priority: 5/5): Levine describes how a chance opportunity with Limelight and a fascination with anycast led him to build Cashfly as a CDN rather than a network provider. MVP architecture and technical hacks (Priority: 5/5): The first version used two pops, Lighttpd, FreeBSD, and a homegrown load balancer; the early system was more distributed web servers than a true cache. From enterprise-only to self-serve access (Priority: 5/5): Cashfly positioned itself as the anti-enterprise CDN, enabling small customers to sign up with a credit card and later revisiting self-serve after cloud-native competition changed expectations. Team building and the value of being 'clueful' (Priority: 4/5): Levine emphasizes hiring people who can make smart decisions in ambiguous situations and thrive without heavy process or documentation. Scalability, logistics, and COVID-era growth (Priority: 4/5): The company’s biggest scaling pain came from operations and supply-chain constraints during COVID, when demand spiked and hardware procurement/delivery became the bottleneck. Product evolution and future of CDN (Priority: 5/5): Cashfly is shifting toward automation and better cache-miss performance as video, 5G, and edge computing make traditional cache-hit narratives less representative of real user experience. Leadership lessons and founder advice (Priority: 4/5): Levine advises startups to use cloud infrastructure initially but plan an exit strategy, and to find a strong number two who can run operations so founders can stay focused on leverage.

Key Arguments: Anycast could power a CDN despite early skepticism; the company proved it in production after building and testing over time. Cashfly’s early differentiation was making CDN access available to everyone, not just large enterprise buyers. The business won by solving practical customer problems one by one, using support tickets and feedback to define the roadmap. Hiring for 'clueful' people mattered more than process because the company needed people who could act intelligently in novel situations. Scalability constraints often came from people and operations, not just software architecture. COVID revealed that infrastructure companies must also be logistics companies: supply chain, shipping, customs, and data-center access all became critical. Modern CDN competition is moving beyond cache-hit speed tests; user experience on cache misses will matter more as content libraries outgrow edge caches. Founders should use cloud early for speed, but design for migration off cloud once traction appears to preserve leverage and margins.

Data Points: CDN locations today: 68 cities - Levine says Cashfly caches content in 68 cities worldwide. Initial pops: 2 - The MVP began with two locations: west coast and east coast. Company launch year: 2005 - Cashfly launched as a retail brand around 2005. Early customer count: 3,000-4,000 customers - Levine estimates the first few years included a broad range of customers from individuals to enterprise brands. Cheapest plan: $15.99 - Cashfly offered self-serve plans starting at roughly this price. Enterprise entry point: ~$10,000 - Levine says the traditional CDN sales process effectively started around this level and required invoices/manual sales handling. Support burden duration: 5 years - Cashfly turned off self-signups for five years after cloud-era support overload. Competitor count exiting market: 2 recently; about 30 since launch - He notes two competitors recently left the CDN business, part of a broader long-term churn. COVID-era footprint growth: 35 to just under 70 locations - Cashfly expanded roughly from 35 locations after COVID to nearly 70 later. Time horizon early on: 5 to 8 years - Levine says the original plan in 2002 was not a long-term 20-year company in his mind.

Pivotal Quotes: "cdn is the new cross connect" — Matt Levine: He recalls a moment with Limelight founders that convinced him they were thinking about the space the wrong way. "the right idea at the wrong time is way worse than not having the idea at all" — Matt Levine: He explains how business frameworks and strategies depend on timing and company life cycle. "we grow to hire we don't hire to grow" — Matt Levine: He describes Cashfly’s preference for efficiency and carefully timed hiring.

Implications: The interview suggests CDN winners will combine automation, operational rigor, and efficient teams. For startups, cloud is the right starting point—but long-term advantage comes from owning infrastructure decisions, hiring wisely, and optimizing for real-world user experience rather than simplistic cache-hit benchmarks.