What comes after smartphones, with Evan Spiegel of Snap
Summary
Snap is entering 2026 at what Evan Spiegel calls a “crucible moment”: nearly 1 billion monthly active users, approaching net-income profitability, and consumer Specs launching later this year after 12 years of investment. The company is simultaneously funding hardware and transforming its software operation with AI.
Specs aim to occupy the empty space between camera-equipped “AirPods” and pass-through VR: normal-glasses wearability with the capabilities of a spatial computer. Spiegel expects portable televisions and work displays to migrate first, but the larger bet is on new shared experiences—laser tag, backyard dinosaurs, collaborative design—that “bring computing into the world.”
Snap’s hardware moat is its ownership of the full stack, from Lens Studio and Lens Core to a Linux-based operating system, waveguides, projectors, and IP-sensitive components manufactured in-house in the US and UK. Spiegel says Android is “way too bloated” for glasses and argues that hardware patents are much easier to defend than software innovations in the AI era.
AI is changing software development and the platform thesis: more than two-thirds of new code at Snap is now written by AI. Lens Studio is becoming more agentic, while bespoke applications can be created rapidly enough that Spiegel calls the traditional “killer app” a mirage; cheap software creation may also weaken the App Store’s million- or two-million-app lock-in.
The core business combines immense engagement with a fast-growing second revenue leg: more than 1 trillion selfies in a year, 25 million Snapchat+ subscribers, and direct revenue running above $1 billion. Advertising remains the other major stream, including Promoted Places and closed-loop measurement, while Lens+ monetizes computationally expensive AI creation and editing.
Snap’s defining product claim is that messaging has different network economics from public social media: one to three close friends can deliver most of the value, while incremental public connections may reduce willingness to share. Its answer was to “separate social from media,” keeping private relationships distinct from moderated Discover content rather than stuffing a friend graph to generate feed inventory.
Spiegel rejects blanket youth-technology policy, arguing for developmentally appropriate parental controls at the operating-system level and insisting that young people must learn AI tools. On safety, John Collison presses that “the algorithm is the content” for each user; Spiegel’s answer is that dangerous material should be stopped through clear content rules at the content level.
As AI makes software production cheaper, Snap sees distribution, product judgment, and proprietary hardware becoming more valuable. A service that once required 15 people can now be built with “half a person’s time” and distributed through Snapchat, while designers increasingly submit working code and Spiegel reviews hundreds of ideas weekly under a North Star of “durable value.”
Deep dive
1. Snap’s crucible moment joins scale, profitability, and a new platform
Spiegel’s 2026 update: Snap is nearing 1 billion monthly active users and “on the verge of net income profitability,” even while investing meaningfully in hardware. Consumer Specs arrive later this year after 12 years of work, alongside an AI-driven transformation of the business.
He does not predict the smartphone’s disappearance. Instead, it may become the most important “legacy” device in a broader computing mix, with glasses taking use cases where an always-available, hands-free, world-aware computer fits better than a phone in a pocket.
Spiegel expects the first migrations to be screen-based: a huge television anywhere, a desktop display or laptop-sized workspace while traveling, or other large-screen use cases. Collison adds that an early use case could be working ergonomically on a flight without hunching over a laptop. Why buy several televisions, Spiegel asks, when lightweight glasses can produce “any size screen”?
Replacement is not the central ambition. Spiegel wants experiences that previously could not exist—laser tag outside, dinosaurs in the backyard, or building Lego together—because the real promise is computing that brings people together. The live demonstration closes the loop with shared virtual darts and “playing chess in the pub.”
2. Specs target spatial-computer capability without VR isolation
The technical target is demanding: combine “the wearability of normal glasses” with something like Vision Pro capability. Current smart glasses are, in Spiegel’s formulation, “like AirPods with a camera,” while high-capability VR remains physically and socially isolating.
Snap owns Lens Studio, the Lens Core rendering engine, its operating system, optical engine, waveguide, projector, and additional unannounced components. A decade optimizing Lens Core for low-end phones became useful preparation for the power and thermal constraints of fitting “a full-fledged immersive computing experience” into glasses.
Snap built its operating system from the ground up on Linux after finding it more performant than repurposing Android. Spiegel calls Android “way too bloated” and points to Google’s Project Aura compute pack as the kind of external hardware burden Snap considers unrealistic for everyday users.
Some of the most important, IP-sensitive components are built in-house in the US and UK beside R&D and engineering, shortening iteration cycles. Spiegel claims Specs’ waveguide resolution and performance exceed competitors and values the defensibility: “It’s way easier to protect our innovations and our ideas” in hardware than in AI-era software.
3. AI turns software abundance into the Specs platform strategy
“Claude is transforming software development, full stop,” Spiegel says; more than two-thirds of new Snap code is now AI-written. Snap benefits from lower production costs while its network effects may insulate it from the rapid cannibalization threatening some standalone software businesses.
The operator-computer relationship is changing too. Developers have already built Specs tools for monitoring agents, letting users check progress and return to their day. Collison’s joke captures the inversion: Claude works at the desktop while the human goes to Coachella and supervises through glasses.
Spiegel calls the “killer app” idea a mirage in 2026 because software can increasingly be bespoke. His concrete example: the Saturn calendar team replaced an enterprise bug-reporting service costing roughly $100,000 annually with a better custom tool built in one day.
Lens Studio is becoming more agentic, allowing people to imagine software and place it into the physical world quickly. That may overcome the historic barrier facing any new computing platform: competing with 1 million or 2 million entrenched App Store applications when new software can now be generated almost on demand.
4. Camera engagement and subscriptions broaden Snap’s economics
Spiegel defines Snap’s primary competition as messaging services, with cameras second: when something happens, does the user open Snapchat or the lock-screen camera? As evidence of scale, he says more than 1 trillion selfies were captured on Snapchat over a year—far above what the team initially publicized.
Snap has two major revenue streams. Advertising includes Promoted Places on the map, with closed-loop measurement for retailers trying to drive store visits; direct revenue, developed over roughly two or three years, has reached 25 million Snapchat+ subscribers and a run rate above $1 billion.
Lens+ provides access to additional AI and editing tools on what Spiegel describes as one of the world’s largest image- and video-generation platforms. More broadly, Snapchat+ finances features requested by passionate users—such as themes—that previously could not win resources when they served only a small subset.
5. Close relationships overturn textbook network effects
Snapchat challenged the belief that a network’s value rises monotonically with its size. Spiegel’s counterclaim is that value comes from “who you actually talk to”—usually a spouse, a few friends, or close coworkers—not from hundreds of nominal connections.
Collison tests the payoff curve: perhaps utility rises quickly through 10 or 20 friends, plateaus, then declines. Spiegel goes further for messaging—one, two, or three friends may deliver most of the value—while insisting that messaging services and public social networks should not be treated as the same product.
Facebook’s feed economics pushed users to add friends so the service could generate more posts and advertising inventory. Eventually, people no longer knew who would see their content, inhibiting posting and accelerating the shift from friend updates toward public media.
Snap’s answer, articulated around 2016 or 2017, was to “separate social from media.” Users communicate privately with important friends; when that material runs out, moderated and AI-ranked Discover content supplies entertainment without forcing them to enlarge the intimate graph for inventory.
6. Ephemerality restores the way human communication normally works
Spiegel expects digital retention to mirror thousands of years of human behavior: “the vast majority of life is ephemeral,” while important moments are deliberately saved. People do not record every conversation or write down every thought; technology should not make permanent storage the default for ordinary expression.
Snapchat linked privacy and self-expression as early as 2012, when the prevailing vision was that everything would become open, connected, and permanent. Spiegel argues permanence affects what people are willing to say; disappearing communication lets someone send a silly photograph without polishing it for public likes or future scrutiny.
Spiegel imagines “the social network of the future” as friends and family physically hanging out while using glasses together—retaining technological capability without sacrificing the richer experience of actually being together.
7. Youth policy should preserve balance and AI literacy
Spiegel’s concern with blanket screen rules is that children differ developmentally and practically. His 15-year-old needed a phone early to coordinate between two households; younger children get limited situational use, such as construction videos during a haircut or occasional movies and Game Boy play.
He rejects the swing from pandemic-era “use screens all day long” enthusiasm to categorical abstinence. Parents should cultivate balance across art, friends, entertainment, and communication, making decisions appropriate to each child rather than outsourcing judgment to a universal age threshold.
The larger risk, in his view, is raising young people without competence in AI during a period of dramatic change. Learning these tools is “absolutely imperative,” both vocationally and creatively; the industry’s obligation is to build enough trust that skepticism does not prevent young people from participating.
8. App bans and algorithm debates miss the control layer
Australia’s under-16 restrictions strike Spiegel as structurally weak because they cover only a small set of apps. Young users can switch to clones or millions of alternatives, while app-level age verification is difficult because many minors lack government identification.
His preferred enforcement point is the operating system. Apple already lets parents control which apps a teenager uses and for how long; Spiegel uses those settings at home and wishes governments spent more time telling parents that these “much more resilient” controls already exist. Collison agrees that naming a few substitutable apps is clumsy.
On well-being, Spiegel cites “maybe three or more” independent studies, not funded by Snap, associating Snapchat with positive relationships and well-being, unlike TikTok or Instagram. His proposed mechanism is connection with real friends in a low-pressure environment—but he concedes that overuse of any technology can produce negative outcomes.
Collison’s pushback—worth keeping—is that recommendation determines the individual user’s actual channel, so “the algorithm is the content.” Spiegel redirects responsibility upstream: material promoting anorexia should violate the rules and not appear on the platform. Snap has moderated public content since Discover launched, initially at the cost of slower corpus growth.
9. Scarce distribution and product judgment become the durable assets
As AI makes companies and software easier to build, Spiegel expects resources to move from engineering toward distribution. Snap can now create a service with “half a team member’s time,” rather than assigning 15 people, then launch it through Snapchat—opening the possibility of a broader multi-app strategy.
Product work follows “it’s cool to be uncool”: usefulness and durable value matter more than chasing status. Spiegel spends a couple of hours weekly reviewing hundreds of sketches, prototypes, slides, and functioning products; designers increasingly resolve neglected backlog items by arriving with, “Here’s my PR.”
His most revealing reversal is Streaks. Spiegel fought the metric for years because Snap avoids user-facing follower, friend, like, and comment counts, but changed his mind after hearing how daily prompts preserved friendships through moves, divorces, and jobs: sometimes people need “an invitation to connect.”
Snap’s first geographic breakout reinforces how product and distribution interact. Norway’s early iPhone distribution, affordable connectivity, and relationship-oriented culture gave photo-heavy Snapchat dry kindling before the US; 15 years later it remains a leading market. Spiegel says leadership now requires recruiting exceptional people and helping them do “the absolute best work of their lives.”