ChatGPT will add ads in 2026. We all expected that. It was a matter of when. Now everyone is speaking about it as a new marketing channel. But ads inside a conversational AI change something deeper: the channel itself. It stops being an AI assistant and becomes the most powerful editor-in-chief in human history. An assistant is at your service. An editor decides what you get to see, with an interest. Search gives you options. Chat gives you ONE answer. Once that answer can be influenced, even slightly, the assistant stops being neutral. Ads get laundered into “advice.” If someone going through a divorce asks for emotional guidance, they will still get some support — but also an option to download the latest meditation app, or even a dating app. Not necessarily a top-rated psychologist nearby. Because digital businesses with the biggest ability to influence the model (and measure the redirects) will naturally invest the most. Meanwhile, Google has the opposite problem. They sit on a $180Billion+ search business — one of the most profitable business models in modern history. Every single search generates ad revenue. Even “Weather in Singapore today” pays. But AI answers don’t generate clicks. For now, they generate cost. The general assumption is that Google will lower Gemini’s price. But if Google makes Gemini too good and too cheap, they would end their own monopoly. So instead, they are giving developers massive TPU credits (around USD 300k or more) to rewrite their code — because once your model is no longer written in CUDA, you don’t need Nvidia at all. So for now, they need to keep AI priced high enough to fund this transition, supporting developers as they move away from Nvidia ecosystem. But the bigger picture is more concerning. Ads won’t be about “selling” products — they will shape judgment. Businesses with the greatest ability to drive online sales (and measure it) will overinvest. In this process, trust will become a finite resource… most likely platforms will spend it all. So this new step puts at risk the main thing that made these models valuable: neutrality.
Very cool insights , lots to ponder on.
Totally agree and I think you articulated it in a way many people haven’t fully internalised yet. A lot of marketers have started dabbling in genAI, but there’s still no tested, reliable framework to anchor against. The ripple effects, like you mentioned, run deep. It shifts how we think about top-funnel visibility, brand perception, and even our basic search habits. What makes this even trickier is that we still don’t have clear visibility into how different chat models rank, retrieve, or prioritise information and those aren’t the same as SEO. They evolve quickly, they’re influenced by a mix of training data, live retrieval, and system-level guardrails… and very few teams are truly set up to navigate that at scale. It’s why conversations like this feel so important. Anyone looking to stay ahead will probably need to start speaking with people who’ve been deep in the weeds of this shift, because the learning curve is only going to get steeper from here. Gonzalo Olivera by the way, if you’ve been exploring any early frameworks or thinking models on this, would love to learn from what you’re seeing too — always helpful to hear how others are approaching it. 🙏
Great perspectives as always, Gonz! It'll be interesting to see how ads on ChatGPT will look like, and how users and marketers adapt to it.