It’s easy to think of AI, especially generative AI, as just a fancy chatbot. And for many, ChatGPT is still the undisputed king of that particular hill. A recent report from a16z, a venture capital firm, paints a fascinating picture of how we’re actually using AI, and it’s far more nuanced than just asking questions.
Globally, over 10% of us are now weekly ChatGPT users – that’s a staggering 900 million people. And while its lead is significant, with Gemini and Claude rapidly gaining ground in paid subscriptions, the competition is heating up. What’s really interesting is that many users aren't sticking to just one platform. We’re seeing this “multi-tenancy” behavior, where people hop between AI tools depending on what they need to do. It’s less about brand loyalty and more about utility.
This shift is partly driven by how quickly these companies are innovating. Google’s creative models, like Nano Banana, are churning out millions of images, while Anthropic is pushing into “prosumer” territory with plugins for everything from Excel to code. The report highlights a crucial concept: “context compounds.” The more an AI understands you, the better it gets, and the harder it is to leave. This is where the idea of app stores for AI, like ChatGPT’s GPTs and Claude’s Connectors, becomes a game-changer. Once your calendar, emails, and workflows are integrated, switching costs skyrocket.
We're seeing two distinct paths emerge: the “super app” approach, where ChatGPT aims to be the default interface for the internet, integrating everything from travel bookings to grocery shopping. Then there’s the “professional tool” route, exemplified by Claude, which is doubling down on specialized integrations for finance, development, and scientific research. The report suggests this could lead to a mobile OS-like battle, with two fundamentally different ecosystems vying for dominance, rather than a single search-engine-like winner.
But the AI story isn't just about the West. The global AI market is fragmenting into three distinct worlds. While Western AI tools share users in markets like the US, India, and Brazil, they’re largely absent in China and Russia due to policy and sanctions. DeepSeek is a notable exception, bridging these divides. Russia, in particular, has emerged as a surprising third pole in AI adoption, with local products filling the void left by sanctions. And in a truly counter-intuitive finding, Singapore, the UAE, and Hong Kong are leading in per capita AI adoption, with the US surprisingly ranking only 20th.
The creative tools landscape is also undergoing a seismic shift. Image generation, once the gateway for many into AI, is seeing its dominance wane. While tools like Midjourney were early pioneers, they’re now being outpaced by integrated solutions within larger platforms like ChatGPT and Gemini. The real action is moving towards video and music generation. Chinese models are leading in video quality, and while OpenAI’s Sora saw a massive initial download surge, its inability to foster a social component meant it didn't stick as a mobile app. Music and voice generation, however, remain more defensible niches, less susceptible to being bundled into core AI offerings.
Finally, the concept of AI “Agents” is moving from theory to reality. Tools like Vibe Coding and OpenClaw are no longer just answering questions; they’re actively doing things for users. While Vibe Coding is gaining traction among technical users, the real promise lies in agents that can perform complex tasks, making AI feel less like a conversational partner and more like a capable assistant. This evolution suggests AI is becoming less about what it can say, and more about what it can accomplish.
