The Top 100 Consumer AI Apps | The a16z Show

Despite the perception of a «yellow war» on tech Twitter, the foundation model race is less zero-sum than it appears. ChatGPT maintains overwhelming dominance — 2.7× larger than Gemini on web, 2.5× on mobile, and nearly 30× larger than Claude on web — yet each platform is carving a sustainable niche. Claude has embraced the proumer market with tools like co-work, Claude in Excel, and Claude in PowerPoint; its app store features premium data sources, research tools, and financial analytics with only 11% overlap with ChatGPT's marketplace-focused directory. Gemini, meanwhile, has staked its claim on creative tools: user growth correlates almost perfectly with releases of its image and video models like Nano Banana and Imagen 3.

The most striking shift is ChatGPT's evolution toward a «Google-like» monetization strategy. Rather than relying solely on subscriptions, OpenAI is building infrastructure to monetize through ads and transaction fees from travel, shopping, and other consumer purchases. This approach mirrors Google's trajectory and positions ChatGPT to capture value from users who may never subscribe but generate revenue through their activity. For developers, the promise of «authentication with ChatGPT» — where users bring their memory, tokens, and identity to third-party apps — could create powerful network effects, locking in OpenAI's 900 million signups while reducing inference costs for developers.

Context is emerging as the ultimate compounding advantage. Products like ChatGPT group chats introduce social switching costs; app store ecosystems create developer lock-in; and cross-product authentication enables personalization at scale. The question is no longer which model is «best» in the abstract, but which platforms can build the deepest moats through user data, developer ecosystems, and workflow integration. All three are betting on different answers — and all three may be right.

The creative tools landscape has matured dramatically since the early dominance of Midjourney. Standalone image generators are declining as ChatGPT and Gemini's built-in models handle commodity use cases — memes, infographics, basic marketing visuals. The survivors, like Ideogram and Midjourney, differentiate through aesthetic sophistication or complex workflows unavailable in general-purpose chatbots. Meanwhile, music (Suno), voice (11 Labs), and video tools continue to thrive because the major model companies have underinvested in these modalities. Video remains fragmented: Chinese models like Canes 2 outperform US offerings due to unrestricted training data, forcing platforms like Korea to offer multi-model access.

Desktop apps represent a major shift in how AI is consumed. Products like Cursor, Whisper Flow, and Granola deliver experiences impossible in a browser: deep file system integration, always-on ambient assistance, and workflows that span multiple applications. This poses a methodological challenge for the Top 100 report, which can track web and mobile usage but not desktop activity. Moving forward, revenue may become a more important ranking signal than visits, as high-value proumer products like Cursor generate massive ARR with minimal web presence.

Voice is crossing into mainstream adoption faster than expected. What began as an early-adopter behavior among engineers — voice dictation, meeting transcription, ambient AI assistants — is now normalizing within tech companies and will likely reach general consumers within six to nine months. Voice is the «most information-dense, high-quality source of media we have», with much of daily work and communication happening downstream of spoken language. The infrastructure is ready; the cultural shift is underway.