I built and run Brainlabs, one of the largest independent advertising agencies in the world. I used to work on the ads team at Google. Fourteen years of putting ads in front of people and measuring whether those billions of $ of ads made money for our clients. I am, by any definition, an advertising man. So when I tell you that ChatGPT ads are going to fail, understand that this is not the opinion of someone who is betting against advertising. It is the opinion of someone who has spent his career in the arena, figuring out what makes it work. On February 9, 2026, OpenAI started showing ads inside ChatGPT. The ads appear at the bottom of responses on the free and lower-cost tiers, initially served through partnerships with Criteo and StackAdapt. A few days ago, OpenAI opened a self-serve Ads Manager to US advertisers, added conversion tracking via a Conversions API and pixel, and eliminated the $50,000 minimum spend from the pilot phase. More than a thousand brands are now live. OpenAI projects around $2.5 billion in ad revenue this year, scaling to what it hopes will be nearly $100 billion annually by 2030. Those numbers will not happen. Here is why.

## 1. Wrong Context

People don't like ads. But they accept them, and sometimes they even find them useful. The ads worked because the context was right. You searched for "plumber near me" and an ad from a local plumber appeared at the top of the page alongside ten other results. That ad was not just tolerable. It was often better than the organic listings, because the advertiser whose money was on the line tried harder to be relevant than the websites gaming SEO. Google's genius was that the ad improved the experience of choosing. You were already evaluating options. The ad gave you one more, and frequently the best one. Meta ads worked for a similar reason, though the context was different. You were scrolling to discover things. An ad for a product you didn't know existed could help you discover something. It was another card in a feed of cards, and the best ones earned their place by being interesting. Nobody loved every ad in their feed, but the format made sense. You were browsing. The ad helped you browse. That's how these companies built multi-hundred-billion-dollar advertising businesses. Not by tricking people into looking at ads, but by placing ads in a context where they could add value. ChatGPT is a different context entirely. You didn't search for something and get a page of options. You didn't scroll a feed looking for things to catch your eye. You asked a question and got one answer. That's it. One answer, and then an ad sitting next to it. The ad doesn't improve the answer. It doesn't expand your choices. It doesn't help you discover anything. It just sits there, doing nothing useful, in a context that has no room for it. Nobody ever said they were chatting with Google. They'd say they were researching something, or looking something up. People chat with ChatGPT. It's in the name. And the expectations that come with a chat are fundamentally different from the expectations that come with a search or a feed. Google ads improved the experience of choosing. Meta ads improved the experience of browsing. ChatGPT ads just sit there.

## 2. The Alternatives Are Better

Even if people didn't like ads on Google, they stayed because there was nowhere else to go. Google built a product so insanely good that it earned the right to show ads. That quality gave Google the room to run ads without losing its audience, and the company tested obsessively to make sure ads did not degrade the experience. No competitor came close for over a decade, so users stayed. People accepted ads on Meta because their friends were already there. A social network is only useful if the people you know are on it. That lock-in gave Meta the same room Google had. You might not love the ads, but you weren't going to leave, because leaving meant losing your entire social life. ChatGPT has no monopoly and no lock-in. There are at least three credible alternatives, and switching between them takes about thirty seconds. Grok, built by xAI, went from 1.9% US market share in January 2025 to 17.8% in January 2026. The models are very good, and for anything that relates to live events or social conversation, Grok's access to X data gives it an edge that produces phenomenal answers. Moreover, if your main priority is truth, Grok outperforms all the other LLMs. Claude, made by Anthropic, is also a better tool when you use the wrappers. I have tried all the tools extensively, and Claude is better on every metric that matters. Anthropic spent millions on Super Bowl ads specifically promising that Claude will remain ad-free, with no sponsored links, no product placements, no third-party influence on responses. More than 80% of their revenue comes from enterprise contracts and paid subscriptions. They do not need advertising revenue and they have made a public commitment not to pursue it. Gemini is also probably better than ChatGPT, with the added advantage of being deeply connected to the Google ecosystem and access to enormous compute. Alphabet is worth $4.8 trillion. Google does not need Gemini to make money. It needs Gemini to protect everything else. If subsidizing an ad-free AI assistant for the next decade is what it takes to keep users inside Google's ecosystem, Google can write that check without noticing. This is the structural problem. Google could afford to put ads in its product because users had nowhere to go. Meta could afford it because your friends were locked in. ChatGPT has neither advantage. If the ads annoy even a fraction of users enough to try something else, they will find that something else is already good, in many cases better, and free, and ad-free. For Google ads to become the industry standard, every other search engine had to adopt ads too. That will not happen with AI assistants. Anthropic has publicly ruled it out. Alphabet does not need it. If ads never become the standard, ChatGPT is not a new ad platform. It is just the one that feels worse to use.

## 3. The Company Itself May Not Survive

The ads play from ChatGPT looks like a survival reflex from a company that is spending more than it earns and running out of ways to close the gap. According to Reuters, OpenAI's CFO Sarah Friar has privately flagged concerns about $600 billion in infrastructure spending commitments against revenue that keeps missing internal targets. The Wall Street Journal reported that OpenAI failed to reach its internal goal of one billion weekly active users on ChatGPT and missed several monthly revenue targets in early 2026. The Information reported that Friar has told colleagues she does not believe the company is ready for an IPO, despite CEO Sam Altman pushing for a public listing as early as Q4 2026. When the CFO does not think you are ready to go public and the CEO is pushing anyway, that tells you something about the urgency of the cash position. On the BG2Pod podcast, investor Brad Gerstner asked Altman directly: "How can a company with $13 billion in revenues make $1.4 trillion of spend commitments?" Altman did not answer the question. He offered to find a buyer for Gerstner's shares. Meanwhile, a federal trial is underway in Oakland where Elon Musk is suing OpenAI, Altman, and President Greg Brockman, alleging they defrauded him when they converted the company from a nonprofit to a for-profit entity. Brockman's personal diary was entered into evidence. In November 2017, Brockman wrote: "If three months later we're doing b-corp then it was a lie." The company's president, in his own handwriting, questioning whether the nonprofit commitment was genuine, years before the for-profit conversion actually happened. None of this means OpenAI will definitely fail. The company has real technology, real users, and significant resources. But the ads product exists inside this context. A company burning cash at an extraordinary rate, with a CFO who questions the spending trajectory, a CEO whose response to financial scrutiny is to offer to buy out the questioner, and a federal trial picking apart the foundational decisions that created the company's current structure. These are not the conditions under which a company builds a hundred-billion-dollar advertising business. These are the conditions under which a company launches ads because it needs the revenue to survive. When the cofounder's personal diary is Exhibit A in a federal trial and the CFO is worried about the balance sheet, the ads look less like innovation and more like scrambling.

## A Note on Disagreeing Well

My colleague Adam Edwards wrote a piece on the Brainlabs blog taking a more bullish view of ChatGPT ads. He makes reasonable points, and he has noted that he has never seen an ad platform scale its features as quickly as OpenAI has. We disagree on the prediction. I think the structural problems are too deep. He thinks the product will improve fast enough to overcome them. That disagreement is great because either way Brainlabs is guaranteed to be right. Joking aside, I started Brainlabs with the belief that conjectures are cheap and refutation is what counts: that great advertising comes from testing lots of ideas, not backing one big idea and hoping to drive revenue. Brainlabs will be running ChatGPT ads regardless of my personal view. If the channel proves to be valuable, we will serve more ads on it, because all we care about is revenue maximization for our clients. So we'll find out.

ChatGPT ads aren't a new platform — they're a tax on a product that no longer has the room to charge it. Wrong context, better alternatives, and a company under pressure. The numbers won't get to $100bn. They probably won't get close.