Vercel
Usage-based + per-seat B2B SaaS: reselling abstracted AWS compute/bandwidth/storage at a markup, plus Pro ($20/seat/mo) and Enterprise plans and v0 AI-builder subscriptions
Valuation stepped ~1.1B (2021) to 2.5B to 3.25B (2024) to 9.3B (2025); the ~2.9x jump from the 2024 Series E to the 2025 Series F tracks the AI-agent revenue surge. The Sep 2025 round paired ~$300M of primary capital with a ~$300M employee/early-investor secondary tender at the same $9.3B mark.
Earnings, margins, COGS & capex
Asset-light PaaS that abstracts AWS serverless infrastructure and resells it with developer tooling on top. Revenue compounding from ~$144M (2024) to a ~$340M run-rate (early 2026) on the back of Next.js ubiquity and a fast-ramping AI product line (v0). Gross margins are estimated healthy (~70%+) but structurally capped by the AWS-passthrough cost base and by heavy free-tier/agent usage.
Income statement — where each revenue dollar goes
% of revenueOf every $1 of revenue, ~30¢ is cost of goods and ~70¢ operating expense, leaving ~0¢ of operating profit.
Revenue trend
Margins
Estimated improving with scale; capped by AWS passthrough
n/a
n/a
COGS structure
Dominated by underlying AWS compute, bandwidth, and storage that Vercel resells, plus AI inference passthrough via the AI Gateway (60+ models). Free/Hobby tier and surging agent-initiated usage (>30% of weekly deployments) consume infra ahead of monetization, pressuring incremental margin.
Capex
Minimal direct capex — no owned data centers; infrastructure is rented from AWS and expensed through COGS. Investment is in R&D/headcount and AI compute commitments, not physical plant.
Latest earnings
n/a (no consensus)
No formal guidance. CEO Guillermo Rauch signaled IPO-readiness (Apr 2026) citing the AI-agent revenue surge but gave no timeline; no filed S-1 as of Jul 2026.
- Revenue run-rate (Mar 2026)
- ~$340M
- ARR growth since early 2024
- ~240% ($100M to $340M)
- v0 users
- 4M+
- Agent-initiated deployments
- >30% of weekly
- Implied revenue multiple
- ~27x run-rate at $9.3B
Growth drivers
- v0 AI app builder — 4M+ users, est. ~$42M ARR (Sacra estimate) within ~1yr of launch; Teams/Enterprise >50% of v0 revenue
- Agent-driven deployments — >30% of weekly deploys are agent-initiated (up ~1,000% in six months); Vercel positions as the host for AI-generated apps
- Next.js ubiquity — 500M+ downloads in the trailing 12 months; powers frontends of Grok, Claude, and Cursor
- AI SDK / AI Gateway — 3M+ weekly downloads; unified access to 60+ models as a monetizable inference layer
- Enterprise upsell — land-with-Pro, expand-to-Enterprise motion across AI-native and brand customers
Bull & bear
Vercel is the toll booth on AI-generated web apps: it owns the framework (Next.js) developers and agents build in, and increasingly the builder (v0) and the host, compounding fast while software-margin AI products scale.
- Next.js is the default frontend framework for the AI era — 500M+ downloads/yr and the frontend of Grok, Claude, and Cursor — giving Vercel unmatched top-of-funnel
- AI is a tailwind, not a threat: >30% of weekly deploys are already agent-initiated and up ~1,000% in six months; more agents building apps = more Vercel hosting
- v0 went from zero to an estimated ~$42M ARR and 4M+ users in ~a year, evidence Vercel can monetize the build layer, not just hosting
- ~240% ARR growth to a ~$340M run-rate is a rare growth profile, and v0's software margins run above the core reselling business
- Deep-pocketed, patient cap table (Accel, GIC, BlackRock, Khosla, General Catalyst) and a stated IPO path give optionality and staying power
Vercel is a premium wrapper on AWS whose margin and moat both get squeezed as Cloudflare and the hyperscalers commoditize hosting and AI tools route deployments around it.
- The core model resells AWS at a markup — a structurally thin, contestable margin; users routinely cite far higher cost vs raw cloud and defect at scale
- Cloudflare (NET) wins on price at every tier (egress ~$0 vs Vercel ~$0.15/GB, ~$150/TB above the included tier) and owns its own edge — a durable cost advantage Vercel can't match
- ~$9.3B on a ~$340M run-rate (~27x) prices in years of high growth; any deceleration compresses the multiple hard
- AI-tool disintermediation: Claude Code already accounts for ~75% of the agentic deploy activity Vercel sees — the agents, not Vercel, may own the customer relationship
- Framework concentration — the whole thesis rests on Next.js/React staying dominant
- No audited financials, undisclosed gross/operating/FCF margins, and no filed S-1 — the profitability picture is a black box
What it is worth
Private last-round mark + revenue-multiple sanity check (no public price). $9.3B Series F (Sep 30 2025) against a ~$340M Mar-2026 run-rate implies ~27x run-rate revenue.
$4-6B
Cloudflare/hyperscaler price war compresses margin, growth decelerates, and the ~27x multiple re-rates toward infrastructure comps on a disappointing S-1.
~$9-11B
growth normalizes, multiple compresses modestly as it matures; IPO in range of the private mark.
Sustained high growth + AI-app-hosting land-grab + margin expansion support a $15-20B+ IPO/next-round mark; agents become a durable volume engine.
~27x run-rate is a premium AI-infra multiple — justified only if high growth persists for several years and gross margin holds/expands past the AWS-passthrough ceiling. No audited margins or S-1 to underwrite it; gross/operating/FCF margins are undisclosed.
SWOT
Strengths
- Owns Next.js, the de-facto React meta-framework — a distribution funnel no competitor can replicate cheaply
- Fast-growing AI product line (v0, AI SDK/Gateway, Agent) already an estimated ~20%+ of revenue
- Premier developer brand and best-in-class DX; frictionless git-to-deploy workflow
- Blue-chip AI-native customer roster (OpenAI, Perplexity, xAI) validates the 'AI cloud' pitch
Weaknesses
- Gross margin structurally capped by reselling AWS at a markup — customers can and do defect to raw cloud on cost
- Heavy dependence on AWS as both supplier and (via Amplify) competitor
- Well-documented 'bill shock' / markup complaints create churn risk at scale
- Rich ~27x run-rate valuation leaves little room for a growth stumble
Opportunities
- Become the default host for the exploding volume of AI-agent-generated apps
- Monetize inference as a middleware/gateway layer across 60+ models
- Move up-market into a broader 'AI cloud' platform beyond frontend hosting
- IPO window if AI-infra multiples stay elevated
Threats
- Cloudflare (NET) undercuts on price at every scale with Workers/Pages and owns its edge network
- AWS/Azure/Google can bundle comparable hosting and squeeze the reseller
- Framework-concentration risk if React/Next.js loses mindshare
- AI coding tools (Claude Code, Cursor, Lovable, Bolt) could route deployments elsewhere — Claude Code already drives ~75% of the agentic deploy activity Vercel observes
Moats, dependencies & bottlenecks
Moats
Controls the open-source framework developers and AI agents build in; unrivaled distribution funnel into paid hosting
Best-in-class git-to-deploy workflow and mindshare; sticky for teams but replicable by well-funded rivals
Vertically integrating build to deploy for AI apps; early lead but crowded field
Reselling AWS is the opposite of a cost moat — Cloudflare and hyperscalers underprice it
Dependencies
Infrastructure supplier Core compute/bandwidth/storage runs on AWS; AWS is also a competitor via Amplify — supplier-and-rival concentration
Technology / mindshare Entire funnel depends on React/Next.js staying the dominant frontend stack
Product input / partner AI SDK/Gateway resells access to 60+ models; v0 and Agent depend on frontier-model quality and pricing
Agent-initiated deploys are a top growth driver but route through third-party agents Vercel doesn't control
Advantages
- Owns the framework (Next.js) that seeds nearly all demand
- Highest-velocity AI-app deploy platform with agents as a tailwind
- Elite developer brand and DX
- Well-capitalized cap table and a stated IPO path
Weaknesses
- Thin, contestable infra-reseller margin
- AWS supplier/competitor dependency
- Documented bill-shock churn risk
- Valuation priced for perfection with no public financial transparency
Bottlenecks
- Gross-margin ceiling from AWS passthrough limits how much operating leverage scale can create
- Free/Hobby and agent usage consume infra ahead of monetization
- Enterprise sales motion must mature to convert self-serve developers into high-ACV contracts
- Price competition from Cloudflare caps pricing power on commodity hosting
Top signals & trends
Top signals
CEO signaled IPO-readiness Apr 2026 with no timeline; a filing would expose true margins/opex and reset the private mark
>30% of weekly deploys and climbing — direct read on the AI-app tailwind
Estimated ~20%+ of revenue and Enterprise-heavy — evidence the build layer monetizes
Aggressive NET pricing pressures Vercel's margin and pricing power
Framework health is the leading indicator for the whole funnel
Trends
More machine-built apps needing a host — the central bull case; already >30% of deploys
Opens a middleware monetization lane but margins on passthrough inference are thin
Cloudflare and hyperscalers driving hosting toward commodity pricing
Sustains the $9.3B mark and IPO optionality while sentiment holds
Cursor/Claude Code/Lovable could either feed or disintermediate Vercel's deploy layer
Ecosystem & competitor graph
Suppliers feed the company; customers pull from it. Line thickness shows the strength of each tie (supply-chain dependency, customer earnings contribution). Hover to isolate a tie.
Primary underlying compute/bandwidth/storage that Vercel resells
Model provider surfaced via AI Gateway and used in v0/Agent
Claude models via AI Gateway; Claude Code drives ~75% of the agentic deploy activity Vercel observes
Gemini among the 60+ models offered through the AI SDK/Gateway
Enterprise customer building on Vercel
AI-native enterprise customer
Enterprise brand customer
Next.js powers its product frontend
Workers/Pages undercut Vercel on price at every scale (zero egress vs ~$0.15/GB) and Cloudflare owns its own global edge network — the most direct and durable public competitor
Vercel's own supplier competing downstream; can bundle comparable hosting at cost
Closest direct private peer — Jamstack/frontend hosting, smaller and slower-growing
GitHub-integrated deploy and Azure hosting; owns the code-host layer above Vercel
AI app-builder rivals to v0; Lovable+v0 are only ~6% of the agentic deploys Vercel sees while Claude Code drives ~75%
Private AI-native build-and-host platform (~$9B valuation, Series D Mar 2026), overlaps v0 in agentic app creation