The very assumptions we hold about what AI can and cannot do are being shattered every few months, revealing just how early we still are in this technological revolution. A16Z’s latest GenAI 100 list, which tracks the top consumer applications by web traffic and mobile active users, serves as a dynamic map of this rapidly shifting landscape. The conversation highlights that while foundational models like ChatGPT ignited widespread awareness, pivotal cultural moments—from the Balenciaga Pope to AI-generated songs—have continually broken down public skepticism and opened new creative avenues. The data shows a market still very much in the early adopter phase, with 17 new companies appearing on the web rankings in just six months and unexpected players like DeepSeek challenging the dominance of established leaders.

A significant portion of the discussion centers on the new trends driving user adoption. AI video is finally reaching a quality threshold that appeals to mainstream creators, evidenced by new entrants like Kling, Hilo, and Sora on the list. Simultaneously, “vibe coding” platforms such as Cursor and Bolt, which allow users to generate applications from simple prompts, have seen explosive adoption among technical and even non-technical users, indicating a massive, previously untapped desire for low-cost software creation. This aligns with the overarching theme of AI drastically reducing the cost and effort of creation across all modalities, from images and music to fully functional apps.

The analysis also delves into the nuanced difference between popularity and profitability. For the first time, the report examined the top apps by mobile revenue, finding only a 40% overlap with the list of top apps by usage. This reveals a market where serious prosumer tools—like specialized video editors or language learning apps—can generate substantial revenue from smaller, dedicated user bases willing to pay for high-value features. Meanwhile, many apps with massive user counts are often built on aggressive, low-cost acquisition strategies with thinner margins. The consistency of 16 companies appearing across all four lists underscores that real, durable consumer AI businesses are being built, even as the chaotic, inventive frontier continues to expand wildly.

Surprising Insights

• DeepSeek’s Meteoric Rise: Despite the settled wisdom that the horizontal LLM market was locked up, DeepSeek reached 10% of ChatGPT’s web scale in just over a week and matched the engagement metrics of established players like Claude within a month.
• The Enduring Power of Companion Apps: AI companion products, including several NSFW-oriented ones, consistently rank in the top 10, demonstrating that demand for conversation and connection is a primary driver of consumer adoption, often surpassing more utilitarian uses.
• Chinese Video Models Leading the Pack: In the emerging AI video space, Chinese models like Kling and Hilo are perceived by many users as exceeding the output quality of their better-funded Western counterparts, due in part to different approaches to training data and a concentrated research focus.
• Creators Outnumber Consumers on “Vibe Coding” Platforms: For tools like Bolt and Lovable, the vast majority of traffic goes to the sites where people build apps, not to the apps themselves, highlighting a massive demand for creation before a single breakout consumer app has even emerged from these platforms.

Practical Takeaways

• For founders, product delight and solving a clear pain point remain paramount over pure technical sophistication; sometimes a simpler, more stable AI feature integrated well beats a fully AI-native product built on unstable models.
• There is significant opportunity in building tailored applications around powerful new primitives like “deep research,” moving beyond generic market reports to niche use cases like tracing meme origins or analyzing creator career trajectories.
• When considering monetization, assess whether your product is a broad, viral tool or a deep, prosumer solution; high-revenue, niche products with passionate users can be more sustainable than chasing massive, low-engagement audiences.
• Explore integrating multimodality—especially voice and real-time screen interaction—into companion and productivity apps, as these features are still underexplored at the application layer but have clear user demand.
• Aggregator platforms that seamlessly stitch together multiple specialized AI models (e.g., for video, image generation, and upscaling) provide significant value as the ecosystem becomes more fragmented and multi-polar.