Sign in Subscribe
Deep Research

Top 10 Generative AI Application Startups (2020–2025) — Ranked by Revenue Growth

Kaizhi Tang

• 4 min read

Generative AI apps went from novelty to real revenue shockingly fast. In this list, I focus on application-layer companies founded in (roughly) 2020–2025 that show unusually strong revenue growth (or credible ARR run-rates), and I highlight the underlying product wedges that made the growth possible.

Executive summary

  • Two dominant growth engines: (1) viral consumer adoption (Midjourney, Character.AI), and (2) high-ACV enterprise contracts in regulated, workflow-heavy domains (Harvey, Writer).
  • The fastest growers often win by being a “UI for a new capability” (text→image, answer engine with citations, voice cloning) plus strong distribution (Discord/community, SEO, enterprise sales).
  • The market’s main trap is model commoditization. Durable winners pair the model with workflow lock-in, proprietary data/feedback loops, and/or platform surface area (API + tools + governance).
  • Revenue numbers are frequently annualized run-rates; treat them as directional unless audited.

1) Midjourney

  • Revenue (2024): ~$300M (reported), up from ~$50M (2022) and ~$200M (2023).
  • What it sells: text-to-image generation (consumer-first) with strong creator community.
  • Why it grew: extremely high perceived magic, rapid model iteration, and a distribution hack: Discord as the product shell.
  • Growth lesson: if your output is inherently shareable (images), community + virality can substitute for paid marketing.

2) Jasper

  • Revenue/ARR: reported ~$80M annual revenue (2022); later guidance implied slower growth amid market shifts.
  • What it sells: marketing copy + brand voice workflows (increasingly enterprise).
  • Why it grew: early “packaging” of GPT into a marketer-ready UI with templates, plus a clear ROI story (more content, faster).
  • Growth lesson: first-mover advantage fades quickly—defensibility comes from workflow depth (brand governance, approvals, integration).

3) Perplexity

  • ARR run-rate (early 2025): just under $100M (reported).
  • What it sells: an “answer engine” that combines LLMs with web search + citations; subscription tier for power users.
  • Why it grew: it attacked a real pain: search results pages became ad-heavy and low-signal; users want direct answers with sources.
  • Growth lesson: a strong wedge is “LLM + retrieval + citations,” but the long-term moat is distribution (default placement, browser, partnerships).

4) ElevenLabs

  • ARR (late 2024): reported $80–90M, up from roughly ~$25M at the start of 2023.
  • What it sells: best-in-class voice synthesis (self-serve + API).
  • Why it grew: quality jump was obvious to users; enterprise use-cases (dubbing, narration) have clear budgets.
  • Growth lesson: API surface area + enterprise trust (rights, safety, controls) converts a viral demo into a durable business.
  • ARR run-rate (Apr 2025): reported ~$75M.
  • What it sells: domain-specific LLM workflows for lawyers (drafting, review, research).
  • Why it grew: legal is high-value and text-heavy; firms pay for productivity. The product is positioned as “augmentation,” not replacement.
  • Growth lesson: regulated domains reward vendors that combine quality + privacy + auditability and can sell top-down.

6) Writer (Writer.com)

  • ARR (projected 2024): reported ~$50M.
  • What it sells: enterprise genAI platform (custom assistants, governance, style/terminology, secure knowledge).
  • Why it grew: it sold to the real buyer: enterprises that need control + compliance + ROI, not just “write better.”
  • Growth lesson: many enterprises won’t standardize on a tool unless it looks like a platform, not a point solution.

7) Copy.ai

  • ARR: reported ~$10M (2022); later reports suggest multi-x growth after an enterprise/workflow pivot.
  • What it sells: GTM content + workflow automation (sales/marketing sequences).
  • Why it grew: big top-of-funnel via freemium, then moved upmarket to capture bigger budgets.
  • Growth lesson: the playbook is common: B2C/SMB adoption → upmarket pivot once the category gets crowded.

8) Character.AI

  • Revenue (2024): reported ~$32M, up from ~$15M (2023), with monetization introduced only in mid-2023.
  • What it sells: consumer subscriptions for entertainment/companion chat with user-generated “characters.”
  • Why it grew: extremely strong engagement loops (role-play, community sharing, creation tools).
  • Growth lesson: if you can create high time-on-app, subscriptions can work even without enterprise buyers—but safety and content policy become existential.

9) Murf

  • Revenue/ARR (2025): reported ~$15M.
  • What it sells: voice-over studio (self-serve) + business plans.
  • Why it grew: huge global demand for narration; creators want speed and cost reduction.
  • Growth lesson: “good-enough” voice is a commodity; durable growth requires editing workflows, licensing, and integrations.

10) Regie.ai

  • ARR: undisclosed; reported 300% YoY ARR growth (2024).
  • What it sells: AI-assisted outbound sales content + orchestration (multi-channel sequences).
  • Why it grew: sales teams pay directly for anything that increases meetings booked.
  • Growth lesson: tying genAI to a measurable revenue KPI (reply rate, meetings, pipeline) is one of the strongest enterprise wedges.

Patterns that predict durable winners

  1. Workflow lock-in beats model quality. The model will be matched; your moat is the surrounding system: approvals, guardrails, memory, data connectors, and team governance.
  2. Distribution is the real platform. Discord/community, app stores, browsers, SEO, and enterprise channel partnerships matter as much as the model.
  3. Compliance features move you upmarket. Audit logs, permissioning, redaction, data residency, and evals are not “nice-to-haves” in enterprise.
  4. Monetization clarity: the cleanest businesses map to an obvious budget line—marketing, sales, legal, media localization, training.

What I would watch next (2026)

  • Agentic workflow products that can take action (not just generate text) in CRM, ticketing, and ops systems.
  • Vertical copilots with proprietary data flywheels (insurance, pharma, industrial).
  • A consolidation wave where model providers bundle apps, forcing application companies to differentiate via workflow depth and brand.

References