How to win when software is not a moat | Evan Spiegel (Snapchat CEO)
TL;DR
Distribution, not product-market fit, is the real choke point in consumer social — Spiegel says TikTok won by spending billions to subsidize both sides of its marketplace and Threads won by piggybacking on Meta’s distribution, while Snap originally grew by connecting users to their closest friends rather than the biggest network.
Snap learned 15 years ago that software is not a moat — after seeing features copied repeatedly, Spiegel says the company shifted toward harder-to-clone defenses like creator and developer ecosystems, AR platforms with millions of lenses, and vertically integrated hardware.
Snap’s innovation engine is a tiny, flat design team moving at brutal speed — the design org is just 9-12 people, everyone presents work on day one, and Spiegel reviews hundreds of ideas weekly because “if you want to have a good idea, you have to have lots of ideas.”
Talking to users matters, but not by building exactly what they ask for — Spiegel used Stories as the example: users asked for a “send all” button, but Snap listened to the underlying pain around social pressure, permanence, and reverse-chronological feeds, then invented a different product entirely.
AI is making designers more powerful inside Snap, but only with guardrails — designers are increasingly shipping code, Snap has automated code review that has caught nearly 10,000 bugs, and internal agents can debug shake-to-report issues and suggest fixes while preserving reliability at nearly 1 billion-user scale.
2025 is a ‘crucible moment’ because Snap must prove both profitability and its hardware future — with over 1 billion monthly active users, roughly $6 billion in revenue, 25 million Snapchat+ subscribers, and Specs launching after 12 years of investment, Spiegel says the company now has to show it can be a strong profitable business and not just an ambitious one.
The Breakdown
Why social apps rarely stick: distribution beats a clever product
Spiegel opens with a blunt point: most founders obsess over product-market fit and underrate distribution. His examples are TikTok, which spent billions subsidizing viewers and creators, and Threads, which rode Meta’s install base. Snap’s own early insight was different from classic network-effects thinking: what mattered wasn’t connecting you to everyone, but to your best friend, partner, or spouse.
Getting copied constantly taught Snap a harsher lesson: software isn’t defensible
When Lenny asks how it feels to be copied, Spiegel shrugs with a good line: it’s better than making things no one wants to copy. But the deeper point is strategic — Snap learned early that patents and feature velocity weren’t enough, so it started building ecosystems and platforms that are much harder to clone than a UI pattern. That logic also pushed Snap toward hardware, where a vertically integrated AR stack is far more defensible than another software feature.
Why Snap keeps betting on glasses instead of phones
Spiegel’s case for Specs is surprisingly human, not technical: phones isolate us, pull us out of the moment, and turn people into “hunched over gremlins.” He frames AR glasses as a way to keep computing in the real world instead of behind a tiny screen — like escaping a “keyhole” view of reality. He also takes a swipe at heads-up displays, joking that when you check the corner screen, you’re basically staring at your friend’s crotch.
Snap’s innovation system: tiny design team, fast critiques, zero preciousness
The most concrete section is on how Snap actually generates ideas. Spiegel cites Safi Bahcall’s Loonshots, then explains Snap’s dual structure: a big operational company for reliability, plus a tiny flat design team of 9-12 people for experimentation. Every designer presents work on day one, everyone rotates across products, and Spiegel reviews new work weekly because the whole culture is built around volume, critique, and the belief that lots of ideas are the only route to a good one.
User feedback should inspire invention, not dictate it
Spiegel pushes back on the “don’t talk to users” school pretty hard. He says you absolutely should talk to users — deeply, for an hour or two — but you shouldn’t blindly ship what they request. His Stories story is the clearest example: users wanted a send-all button, but what Snap heard underneath was discomfort with pressure, permanence, likes, and feeds that told stories backward, which led to a new format built around 24-hour, chronological sharing.
Why Snap waited so long to hire PMs — and why that changed
Spiegel says Snap’s early resistance to product managers was really a defense of design. He didn’t want designers reduced to “make me some visuals” execution behind PM-authored specs, so designers were expected to do product thinking themselves. At today’s scale, though, trust and safety, legal review, data science, and coordination make PMs essential — not as mini-CEOs, but as connective tissue.
AI inside Snap: designers shipping code, agents catching bugs, workflows collapsing
This is where the interview gets very current. Spiegel says designers now routinely ship code, not because it’s required but because AI removes friction from creation; meanwhile Snap’s automated review systems have already caught close to 10,000 bugs. He also describes internal agents built with tools like Glean and Claude that scan dashboards and docs, summarize hotspots, debug issues, and increasingly compress whole workflows from product spec to risk review to go-to-market materials.
The ‘crucible moment’: Snap has scale, but now it has to prove the next chapter
Near the end, Spiegel gets unusually direct about the stakes. Snap is nearing Fortune 500 scale, has almost 1 billion MAUs, and is preparing to launch Specs after 12 years of investment — but it still has to prove it can be sustainably profitable. His most contrarian closing point is bigger than Snap: in AI, humanity matters more than technology because adoption is social, and he expects much more public pushback than many tech leaders do.