Artificial Intelligence March 29, 2026

Inside a16z’s Top 100 Gen AI Apps Report and Gemini’s Rapid Rise

ChatGPT commands 900 million weekly active users – more than 10% of the global population. But the AI landscape beneath that headline is shifting fast. Andreessen Horowitz’s fifth edition of its Top 100 Gen AI Consumer Apps report, authored by partner Olivia Moore and published in March 2026, reveals an ecosystem that has matured dramatically since the early days of ChatGPT, Midjourney, and Character.AI. The most striking development: Google’s Gemini has surged into a clear second place, narrowing the gap on mobile and growing paid subscribers at 258% year-over-year.

Three years into the generative AI consumer era, the report captures a market that is simultaneously consolidating around a few dominant platforms and splintering into specialized niches. AI startups generated over $1 billion in new revenue in 2025 alone, proving that application-layer companies can thrive even as foundation models grow more powerful. The distinction between “AI-first” products and traditional software has effectively dissolved – tools like CapCut (736 million mobile MAUs), Canva, and Notion now rely on generative AI for their most popular features, prompting a16z to broaden its ranking criteria for the first time.

ChatGPT Still Leads, But the Gap Is Closing

ChatGPT remains far and away the largest consumer AI product. On the web, it draws 2.7 times more monthly traffic than second-place Gemini. On mobile, it holds a 2.5x lead in monthly active users. Weekly active users grew by 500 million people over the past year to reach 900 million – a staggering figure that makes sustained growth increasingly difficult simply because of the shrinking pool of potential new users.

Yet the report signals that the category is widening. Roughly 20% of weekly ChatGPT web users also use Gemini in a given week, indicating that consumers are increasingly comfortable splitting their AI usage across platforms rather than committing to a single assistant. Sessions per user remain 1.3x higher on ChatGPT’s web product and 2.2x higher on mobile, but Gemini’s session counts are climbing.

Gemini’s Surge: What’s Driving the Numbers

Google’s rise to a firm second place didn’t happen by accident. Several product launches converged to accelerate Gemini’s trajectory.

The Nano Banana image model generated 200 million images and brought 10 million new users to Gemini in its first week alone. Veo 3 was widely regarded as the breakthrough moment for AI video, and Google Labs saw traffic spike more than 13% following its launch. Desktop users grew 155% year-over-year, compared to ChatGPT’s 23% desktop growth over the same period. Monthly visits climbed from 284 million in February to 700 million by mid-2025, though still well below ChatGPT’s 5.72 billion.

On mobile, Gemini’s advantage is structural. Nearly 90% of Gemini’s monthly active user base comes from Android devices, compared to 60% for ChatGPT. With over 3 billion Android devices worldwide, Google can embed AI directly into the operating system, turning passive device usage into active Gemini engagement without requiring users to seek out a standalone app.

Paid subscriber growth tells an equally compelling story. As of January 2026, Gemini was growing paid subscribers at 258% year-over-year – outpacing Claude’s 200% growth rate, though ChatGPT remains roughly 4x larger in absolute subscriber count. Gemini Pro users show 57% retention at month 12, approaching ChatGPT’s 68%, which challenges the assumption that Gemini’s growth is purely free-tier driven.

Metric ChatGPT Gemini Ratio
Web Monthly Traffic Leader (base) ~12% of ChatGPT 2.7x larger
Mobile MAU Leader (base) ~40-50% of ChatGPT 2.5x larger
Weekly Active Users 900 million ~35-40% of ChatGPT ~2.5x larger
Paid Subscriber YoY Growth Leader in scale 258% ChatGPT 4x larger absolute
Desktop User YoY Growth 23% 155% Gemini 6.7x faster
DAU/MAU Retention 36% 21% ChatGPT 1.7x higher
Month-12 Paid Retention 68% 57% ChatGPT 1.2x higher

Google’s Multi-Product Assault on the Top 100

For the first time, a16z was able to track Google’s AI products individually, and the results are striking. Google landed four tools in the top 100: Gemini at #2, AI Studio at #10, NotebookLM at #13, and Google Labs at #39. This unbundled strategy contrasts sharply with OpenAI’s consolidated ChatGPT approach, allowing Google to compete across multiple categories simultaneously – developer tools, research, consumer experimentation, and general-purpose AI assistance.

NotebookLM, which first went viral nearly a year ago, has grown steadily since gaining its own independent website. Google Labs hosts Veo 3 alongside experiments like Doppl (clothing try-on), Portraits (AI coaches), and Project Mariner (an agentic browser). The breadth of Google’s portfolio means it captures users at multiple entry points, even if no single product matches ChatGPT’s scale.

The Emerging Platform Wars: Apps, Lock-In, and Super-Apps

The report’s most forward-looking analysis concerns what happens when AI assistants evolve from chat windows into operating environments. Both ChatGPT and Claude have launched connector ecosystems – GPTs and Apps for ChatGPT, MCP integrations and Connectors for Claude – that let users build workflows on top of the assistant. Once someone configures their AI to talk to their calendar, email, and CRM, switching costs rise dramatically.

The two platforms are diverging in revealing ways. ChatGPT’s app directory includes 220 apps across 13 categories, with 85+ spanning Travel, Shopping, Food, Health & Wellness, Lifestyle, and Entertainment – consumer transaction categories featuring partners like Expedia, Instacart, and Zillow. Claude’s roughly 160 curated connectors plus 50 community-built MCP servers skew toward professional tools: financial data terminals like PitchBook and FactSet, developer infrastructure like Sentry and Supabase, and science tools like PubMed and Benchling. The two share just 41 apps in common – roughly 11% of the combined catalog.

OpenAI has signaled ambitions to make ChatGPT a consumer super-app, with plans for a “Sign in with ChatGPT” identity layer and advertising integration. The result may look less like the search wars – where one player took 90% – and more like the mobile OS wars, where two platforms with different philosophies both built trillion-dollar ecosystems.

Beyond the Big Three: Grok, Chinese Apps, and Vibe Coding

The report captures several other notable shifts. Elon Musk’s Grok went from zero mobile app users at the end of 2024 to upwards of 20 million monthly active users, ranking #23 on mobile. The July 2025 release of Grok 4 triggered a nearly 40% surge in xAI’s active user base.

Chinese-developed apps now account for an estimated 22 of the top 50 mobile AI apps, though only three primarily serve Chinese users. The concentration is heaviest in photo and video, where companies like Meitu produced five separate entries and ByteDance contributed four products including Doubao (the #4 mobile app) and Cici. Chinese video models have benefited from a deeper bench of video researchers and more permissive copyright rules, though Veo 3 became the first U.S. model to break this trend.

Vibe coding – the practice of using AI to generate entire applications from natural language descriptions – has exploded. In the previous ranking, only Bolt appeared on the web list. Now Lovable has surged to #22 (up from #55), and Replit debuted on the main list. Credit card panel data shows that cohorts of U.S.-based users on top vibe coding platforms maintain revenue retention upwards of 100% for several months post-signup, meaning cohorts are growing their overall spend even accounting for churn.

The Maturing Ecosystem: Thick Apps and Multi-Model Orchestration

A defining theme of the report is the emergence of what a16z calls “thick” AI apps – products that go far beyond wrapping a single model’s API. These applications feature multi-model orchestration, autonomy sliders that let users control how much agency the AI exercises, and sophisticated context engineering that creates deep specialization.

The generative media landscape illustrates this perfectly. Enterprise production deployments use a median of 14 different models, a stark contrast to the LLM market where OpenAI, Gemini, and Anthropic command 89% of enterprise wallet share. Producing a single polished creative asset often requires chaining together multiple specialized models – generating an image, removing the background, upscaling, recoloring, and applying style-consistent adjustments.

The shift from visible prompts to invisible AI scaffolding is accelerating. Mainstream apps increasingly eliminate the prompt box entirely, using proactive interventions – IDE refactors, CRM email drafts after calls, design tool variations generated automatically. Chat interfaces are becoming what one a16z partner calls “training wheels” for a future where AI operates behind the scenes.

What Comes Next: Agents, World Models, and the 2026 Gold Rush

The report positions 2026 as a potential “once-in-a-decade gold rush” in consumer technology. ChatGPT’s 900 million user base, combined with its Apps SDK and Apple’s mini-apps support, creates a distribution platform that rivals the early App Store era. OpenAI’s group messaging features and developer tools are designed to make ChatGPT the starting point for everything from shopping to health management.

Personal AI agents with persistent memory represent the next breakout category. Tools like OpenClaw are gaining traction by maintaining context across sessions, making the AI progressively more useful as it learns about the user. Context compounds – the more an LLM knows about you, the better results it provides and the more you use it, creating a retention flywheel that could prove structurally difficult to displace.

World models are pushing from prototype to product. Marble from World Labs generates persistent, interactive 3D environments from text or images, while DeepMind’s Genie 3 enables real-time explorable video. Applications span gaming, entertainment, simulation, and training autonomous systems – what one a16z partner describes as “generative Minecraft” where users can issue commands like “create a paintbrush that changes anything to pink.”

Key Takeaways

The fifth edition of a16z’s Top 100 Gen AI Consumer Apps report paints a picture of an ecosystem entering its next phase. ChatGPT’s dominance is real but not unassailable – Gemini is growing desktop users nearly seven times faster and converting paid subscribers at 258% year-over-year. The battle is no longer just about model quality; it’s about ecosystem depth, distribution advantages, and the race to become the default interface between consumers and the digital world. With over $1 billion in new AI app revenue generated in 2025 and less than 10% of enterprise potential realized, the opportunity ahead remains enormous – and the competitive dynamics are only intensifying.

Sources