This article covers how Cloudflare built a distributed network across 200 cities to become a cybersecurity giant. Co-founder Zatlyn sees the network as a hard-to-replicate moat, and the company is now evolving into edge computing (Workers), turning the internet into a computer. Key holding: Cloudflare, whose market cap grew from $5B to $25B+ since IPO, blocking 72B cyber attacks daily.
Michelle Zatlyn (co-founder and COO of Cloudflare) discussed in the interview the company’s journey starting from Project Honeypot, explaining how it makes over 25 million websites more secure and faster through a distributed network. Cloudflare currently has a market capitalization of approximately
Michelle Zatlyn (Cloudflare co-founder and COO) elaborated in the interview on the company's journey from Project Honeypot to becoming a cybersecurity giant with a market cap exceeding $25 billion. The core theme is: Zatlyn believes that Cloudflare's distributed network is the core moat. Through the "traffic for space" strategy, it has established nodes in 200 cities, and is evolving from network security protection to an edge computing platform (Workers), with the goal of turning the network itself into a computer.
Zatlyn believes that the key to early-stage entrepreneurship lies in finding "mosquitoes" rather than "sharks"—that is, problems that truly bother users, not superficial glitz. Cloudflare's origin came from a survey: small business owners were surprisingly furious about web spammers, with one user replying, "web spammers make me believe in the death penalty." Yet at the time, there was no technical solution for small businesses; "big companies have good cybersecurity solutions, but everyone else doesn't."
Zatlyn points out that when your technical solution can solve a pain point that users already know and feel strongly about, product-market fit (PMF) comes naturally. Although Cloudflare's early product was crude (no analytics), user feedback was excellent because they could directly perceive value through a 50% drop in server load.
Zatlyn emphasizes that building a global network across 200 cities is Cloudflare's core moat and the hardest barrier for competitors to replicate. The reasons are:
| Comparison Dimension | Cloudflare | Traditional Cloud Provider (AWS/Azure) |
|---|---|---|
| Number of nodes | 200 cities | Approximately 20–30 regions (each with multiple availability zones) |
| Node positioning | Close to users (edge) | Close to data centers (centralized) |
| Expansion driver | Traffic invitation (ISP proactive) | Capital investment (self-built data centers) |
| Network density | Very high (e.g., 3 nodes in Pakistan) | Relatively low (each region covers a large area) |
Zatlyn believes that Cloudflare's distributed network advantage is difficult to replicate in the short term, because one needs a large amount of traffic first to be invited by ISPs, and obtaining traffic requires having a network first—this constitutes a chicken-and-egg problem. Falsification signal: If a competitor (e.g., Amazon) forcibly builds a network of similar density through massive capital investment and can obtain the same level of ISP cooperation, the moat may be eroded.
Zatlyn proposes that Sun Microsystems' prophecy "the network is the computer" is being realized within Cloudflare's network. She points out that Cloudflare Workers allows developers to deploy code on edge nodes in 200 cities, rather than running it on traditional servers. This creates a "third place to write code" — neither server-side nor client/browser, but at the network edge.
Zatlyn illustrates the value of edge computing with a specific case: a large content company needs to apply regional watermark verification to its content. The traditional approach handles it either server-side (incurring a latency penalty) or client-side (which is impossible). Through Cloudflare Workers, the watermark verification logic can run on edge nodes in 200 cities, staying close to the user while maintaining control, turning the problem from a "boardroom-level discussion difficulty" into a directly executable solution.
Zatlyn acknowledges that the main current limitation is distributed storage — it is difficult to run a large distributed database across 200 nodes. She believes that once the storage issue is resolved, it will trigger a paradigm shift in computing architecture: the discussion of "what is suitable to write on the server, what is suitable to write on the client, and what is suitable to write at the network layer" will be fundamentally transformed. She expects this transition to take 5-7 years.
Zatlyn shared a key decision: during the Series C round, Cloudflare received a high-valuation investment offer from a well-known venture capital firm (valuation exceeding $1 billion), but the investor repeatedly advised the company to cancel its free service. The Cloudflare team insisted that the free service was key to acquiring traffic ("We need traffic to reach ISPs in Pakistan"), and ultimately rejected the offer, instead accepting investment from Union Square Ventures at "less than half the valuation."
Zatlyn believes that this decision avoided long-term strategic conflicts and was a cornerstone of the company's "low-drama" culture. She emphasized that choosing investors is as important as choosing co-founders: "If you choose the wrong co-founder, you are already sailing against the wind."
Zatlyn introduced Cloudflare's capital allocation strategy: establishing two independent R&D teams — one focused on the core business (current revenue source) and the other (the "Emerging Technology Team") focused on future directions, not measured by quarterly revenue. She believes that most companies constantly sacrifice the latter between "known, revenue-generating businesses" and "future, uncertain opportunities," and Cloudflare avoids this tendency through physical separation (placing the future team in Austin) and different evaluation criteria.
| Target | Guest Attitude | Key Data |
|---|---|---|
| Cloudflare | Bullish (as its own company, discuss its moat and growth path) | 25M+ websites, 200 city nodes, 72B network attacks blocked per day, market cap grew from $5B at IPO to $25B+, only 3 marketing team members when revenue was $80M |
1. The "Shark and Mosquito" Rule: Zatlyn cites Bill Gates's research, pointing out that the "mosquito" (a small problem), which should be overlooked, is instead deadly. Entrepreneurs should focus on the "mosquito" (real user pain points), not the "shark" (superficial glories like fundraising and valuation). Support: Cloudflare's predecessor, Project Honeypot, directly originated from an exploration of the sources of network spam, not from fundraising competition.
2. The "Network Is the Computer" Paradigm: Zatlyn believes that Sun Microsystems's prophecy is being realized precisely because of Cloudflare's distributed computing capabilities. She believes that in the future, a large number of computing tasks will migrate from central servers to the network edge, ushering in a new application layer. Support: The Workers project allows code to run on nodes in 200 cities, solving problems such as watermark verification that "cannot be done on the server side (too slow) nor on the client side (impossible)", but it needs 5-7 years to mature.
3. Rejecting High-Valuation Investment Reflects Strategic Resolve: Zatlyn explicitly rejected an investment with a higher valuation because that investor suggested canceling the free service, which was key to acquiring traffic. Support: Cloudflare ultimately accepted investment from Union Square Ventures at "less than half the valuation", and Zatlyn considers this a decision that "saved us a lot of trouble".
4. Dual-Track R&D Is the Antidote to "Future Being Hollowed Out": Zatlyn points out that a single R&D team will naturally prioritize future projects later, because the current business has clear revenue. Support: Cloudflare established an independent team (the "Emerging Technology Team" in Austin), which is not measured by quarterly revenue and focuses on future directions such as Workers.
5. The "Chicken-and-Egg" Barrier of the Network Moat: Zatlyn emphasizes that building nodes in 200 cities requires having traffic first to be invited by ISPs, and acquiring traffic requires having a network first. Support: Cloudflare solves this problem with a "free service for traffic" model, which competitors find difficult to replicate because they need both network capabilities and a traffic base.
6. The "Traffic for Space" Financial Model: Zatlyn explains Cloudflare's cost advantage in expansion: local ISPs, seeing high traffic backhaul costs, proactively invite Cloudflare to set up, thereby reducing Cloudflare's operating costs. Support: The case of Pakistan — the ISP paid for part of the infrastructure cost, and Cloudflare traded localized services for lower costs.
7. Early Product Was Crude but PMF Was Perceptible: Zatlyn notes that the early Cloudflare product had no analytics, but users directly perceived its value through a 50% drop in server load. Support: This explains why "our embarrassing product" received "thank you" replies — users could verify its value through their own metrics.
8. Choosing a Co-Founder Is Like Choosing a Life Partner: Zatlyn believes that the wrong co-founder can "double all difficulties", while the right choice can reduce "drama". Support: Among Cloudflare's three co-founders, two are still active, and the third left due to health issues; the company has maintained a long-term stable cooperative relationship.