Back

Landscape

Consumer AI Apps

Consumer AI usage is broad (nearly half of U.S. consumers) but shallow — only 4.5% have a paid subscription to ChatGPT, Gemini, or Claude, and spending follows a steep power law where the top 1% of payers account for 19.5% of all observed spend. The market's first real revenue base is prosumers paying for coding, creative, and productivity tools, while reaching mainstream consumers will likely require new business models beyond subscriptions — including ads, transaction fees, and personal agents.

Created Oct 6, 2026·Updated Oct 6, 2026

Recent Updates

Market Structure: Three LLMs, Three Games

ChatGPT remains the dominant consumer AI product with roughly 2x Gemini's web traffic and 6x Claude's. On mobile the gap is wider: 2.5x over Gemini and 14x over Claude in monthly active users. ChatGPT also leads on paid subscribers at 3x the rate of Claude or Gemini.

Claude has emerged as a clear #3, passing DeepSeek and Perplexity in traffic. It overtook Gemini on U.S. consumer subscribers in early 2026, driven by product launches (Cowork, Claude Design) and new models (Fable 5). Claude monetizes its premium tier aggressively: 7.3% of its consumer payers are on the $100/month Max plan, compared to 1.3% for Google and 1.1% for ChatGPT on equivalent tiers source(https://a16z.com/100-gen-ai-apps-7/).

The three companies are diverging in strategy. Anthropic focuses almost entirely on the prosumer (Claude Design, Code Review, Claude Science). Google leads on creative models (Lyria 3 Pro, Gemini Omni). OpenAI spans consumer and enterprise with Images 2.0, ChatGPT Work, Health, and Personal Finance tools. Overlap between subscriber bases is low — only 8% of U.S. ChatGPT subscribers also subscribe to Claude.

By mid-summer 2026, Claude's daily sessions and subscriber growth decelerated while ChatGPT re-accelerated with the July launch of GPT-5.6 Sol, Terra, and Luna.

Spending Follows a Power Law

For the first time, the report adds consumer spending rankings using YipitData e-receipt panels, revealing extreme concentration:

  • 4.5% of eligible U.S. consumers had an active paid AI subscription in August 2026 (up from 2.1% a year prior).
  • Only 13% of paying users pay for more than one AI tool.
  • The top 1% of payers account for 19.5% of all observed consumer AI spend — more than the bottom 50% combined (16.6%).
  • Top 1% average spend: $903/month on consumer cards, up 80% in 18 months. Median AI payer: $25/month with minimal growth source(https://a16z.com/100-gen-ai-apps-7/).

These power spenders disproportionately purchase automation and building tools (n8n, fal, Manus, Nous Research) and creative tools (Higgsfield, Figma, HeyGen). 29 of the top 50 vendors by spend do not appear on the traffic lists at all — spend and traffic measure genuinely different things.

Only seven companies make all three lists (web, mobile, spend): ChatGPT, Claude, Suno, Perplexity, Photoroom, Canva, and Notion.

Personal Assistants

Consumer AI agents have moved from unusable to mainstream contender in six months. Multiple startup agents report hundreds of thousands of users: Instinct, Tomo, Poke, Lindy, and Town. Incumbents responded with Meta's Muse, OpenAI's Dots, and xAI's Grok Bot.

The channel of choice is iMessage — Instinct launched there in August 2026 and reportedly passed 100,000 users in three weeks while still invite-only. Meta launched Muse on September 9 across browser, desktop, mobile, and WhatsApp, reaching 250,000 daily actives in its first week and 5M downloads in under a month.

Early agent usage mirrors the power-spender profile: the most discussed task domain across early user communities is coding and technical automation. Instinct reports 40% of users connect a personal credit card within three weeks, spending an average of $1,300/month through the platform — with half attributable to travel. The company claims over $1 billion in annualized transaction volume source(https://a16z.com/100-gen-ai-apps-7/).

Platform dynamics are forming: Amazon blocked Meta's Muse within two weeks, while Shopify, Instacart, OpenTable, Expedia, Ticketmaster, and Plaid signed official integrations.

Where Startups Still Win

Incumbents are prominent on the lists — Canva and Notion rank top 10 on web, Figma top 15, Superhuman (fka Grammarly) at #4 on spend. Google alone has five web entries. Despite this, startups find durable positions through four strategies:

  1. Own a differentiated model. Creative products built on proprietary models sustain pricing power even as general-purpose models improve. Suno (#19 web, #7 spend), ElevenLabs (#25 web, #10 spend), Midjourney, HeyGen, Kling, and Topaz Labs all make the spend list.

  2. Own a multi-model experience. Cursor, OpenRouter, Lovable, Base44, and Replit route users to the best model per task — an incentive alignment advantage over labs selling their own model.

  3. Own an audience with specific needs. OpenEvidence serves physicians (50–60% of U.S. physicians use it); Venice focuses on private AI. Content-policy gaps also create openings — excluded NSFW products would have composed over 20% of the web traffic list.

  4. Own an experience requiring interface change. Startups have fewer habits and revenue streams to protect. Hardware entrants like Plaud (#16 spend) sell a physical device plus subscription. Meta and OpenAI are expanding into hardware but haven't shipped yet.

Business Model Gap

Almost all consumer AI revenue comes from subscriptions (84% of AI-native web products) and usage charges/credits (64%). Only 14% have ads; 2% earn transaction or platform fees. This is an inversion of the pre-AI consumer internet, where 97.6% of Meta's 2025 revenue and 73.2% of Alphabet's came from advertising.

Subscriptions work for prosumers but cap mainstream adoption. Among the top 20 global consumer subscription products, 65% are media businesses (Netflix, Spotify, etc.). ChatGPT already ranks in the top 20 — illustrating how hard it is to build a massive paid subscriber base.

High model costs forced early AI companies to monetize directly rather than grow on ads, but this is shifting:

  • Ads: ChatGPT advertising hit a $1B annualized revenue run rate in August 2026 on 1.2B weekly active users. OpenEvidence also partially monetizes via ads.
  • Transaction fees: Both Instinct and Muse plan to earn fees on purchases made through their agents, though neither charges yet. An affiliate-fee or take-rate model could unlock usage-based revenue without subscription friction.
  • Falling costs: Open-source models and task-level model routing (expensive frontier models for hard tasks, cheap models for simple ones) will make ad-supported and free tiers more viable.

Sources