The Fair Question: Won’t OpenAI Just Build This?
Any thesis about advertising inside AI has to survive one blunt question: OpenAI owns the surface, the model, and the auction, so why would it not simply absorb the creative layer and squeeze out anyone building on top? It is the right question to ask, and dismissing it with hand-waving about “focus” is not good enough. The better answer starts by conceding the strongest version of the objection, then showing why it does not lead where people assume.
So here is the concession up front, stated as plainly as possible: OpenAI is not going to leave creative untouched. It has both the capability and the incentive to help advertisers make ads, and it has already started. The question is not whether OpenAI touches creative. It is whether OpenAI will build the neutral, cross-channel, advertiser-aligned creative and campaign layer that a serious advertiser actually needs. Those are very different things, and the gap between them is where an independent platform belongs.
Yes, OpenAI Is Adding Creative Tools
Let us be specific rather than defensive. Around June 2026, OpenAI published its Ad Tools Terms, defining two optional features for advertisers: Audience Tools, which let an advertiser upload first-party customer data to build custom audiences, and Creative Tools, described as AI-powered features that “generate, modify, transform, optimize, localize, or translate advertising creatives” from a brand’s own materials (Digiday, 2026). Trade press correctly read this as OpenAI moving beyond selling inventory toward the fuller toolkit advertisers expect from a mature platform.
There is a clean logic to it. As one analysis put it, every major platform has spent years automating the buying process, which made creative the last real variable, and platforms have a structural incentive to close that gap because more variants mean more auction signals, more liquidity, and ultimately more revenue (Digiday, 2026). Read that incentive carefully, because it is the whole story: OpenAI is automating creative to feed its own auction. That is a legitimate goal. It is also a fundamentally different goal from producing the best creative for your business across every channel you run. The rest of this article is about that difference.
“More variants, more liquidity, more revenue”
the platform’s reason for automating creative: to feed its own auction, which is not the same as optimizing your results across every channel
Reason 1: The Seller Cannot Be Your Neutral Optimizer
The deepest reason is a conflict of interest that no feature roadmap can resolve. OpenAI runs the auction and sells the inventory. If it also grades, generates, and “optimizes” the creative that competes in that auction, it is on both sides of the table. When the same party that profits from your spend also decides what “good” looks like, the optimization target quietly becomes its revenue, not your return. That is not a claim about anyone’s intentions; it is the structure. A tool built by the media owner is engineered to increase spend on that media, because that is how the media owner makes money.
Advertisers have always understood this, which is why independent measurement, verification, and creative shops exist alongside every walled garden. You do not ask Meta whether you should spend less on Meta. You do not expect Google’s asset generator to tell you the best move is to shift budget to Reddit this quarter. A creative and campaign layer that is not the seller can make exactly those calls, because its only incentive is your outcome. Neutrality is not a nice-to-have in advertising; it is the entire reason a category of independent tools has existed for two decades.
Reason 2: A Single Surface, By Design
The second reason is simpler and harder to argue with: OpenAI’s creative tools only make ads for OpenAI’s surface. By construction, they will never produce your Meta feed and Stories set, your Google responsive assets, your Reddit unit, or your LinkedIn variant. A media owner has no reason to build you great creative for a competitor’s inventory, and every reason not to.
But real advertising is cross-channel, and it has to be, because a single young surface is volatile. ChatGPT ads launched at a $60 CPM that fell toward $25 within ten weeks as inventory grew (industry reporting, 2026). A business that can only make ads for one surface is hostage to that surface’s economics. The job an advertiser needs done is “make my whole campaign, everywhere, from one brief,” and that job is intrinsically off-limits to any single platform’s in-house tools. It is native territory for a neutral layer.
1 surface vs every surface
an ad platform’s built-in creative tools serve only its own inventory; advertisers need on-brand creative for ChatGPT, Meta, Google, Reddit, and LinkedIn from one brief
This is where Lapis lives. From one prompt it generates production-ready, on-brand creative for ChatGPT plus Meta, Google, Reddit, and LinkedIn in under three minutes, each unit sized natively for its placement. No single ad seller will ever offer that, because it would mean building creative for channels it does not monetize. For the fuller architecture of how these layers fit, see LLM ad infrastructure explained.
Reason 3: Ads Are a Means, Not the Mission
The third reason is about where OpenAI’s attention and capital actually go. OpenAI’s stated purpose is building advanced AI, and advertising is a way to fund that. Its own leaders have framed it exactly so: at its first Cannes Lions, OpenAI described ads as a way to “subsidize” access to its products and tie revenue to the mission of reaching more people (AdExchanger, 2026). Advertising is a monetization lever for a company reportedly facing multibillion-dollar losses as it spends enormously on compute, with ad revenue projected to scale from roughly $2.5 billion in 2026 toward $100 billion by 2030 (Axios, 2026).
A company in that position optimizes for inventory monetization at massive scale, not for the depth of an advertiser’s creative workflow. It wants enough creative flowing to keep its auction liquid, and it is happy to let the ecosystem handle the rest, which is why OpenAI partnered with the large agency holding companies as launch partners rather than trying to become the creative shop itself. Building Brand Intelligence, independent performance forecasting, competitor tracking, cross-channel analytics, a natural-language editing studio, catalog import, and multilingual generation into a polished product for millions of small advertisers is a full company’s worth of focus. For OpenAI it is a rounding error against AGI; for Lapis it is the entire product.
$2.5B → $100B
OpenAI’s projected ad revenue from 2026 to 2030, a monetization lever to fund its core mission, which is exactly why deep advertiser tooling is not its focus
Reason 4: The Google Precedent
If you want to know what happens when a dominant ad platform adds creative tools, you do not have to speculate. It already happened with Google and Meta. Both have offered automated asset generation, responsive creative, and increasingly aggressive AI creative features for years. And yet an enormous independent ecosystem of creative, testing, and optimization platforms not only survived but grew, precisely because advertisers wanted tools that were neutral, cross-channel, and aligned with their outcomes rather than the platform’s.
The pattern is consistent across every era of advertising: the platform owns the auction, and an independent layer owns the creative and the campaign workflow. Search advertising did not eliminate the SEO and creative industry; it created it. Social advertising did not remove the need for independent creative and analytics; it multiplied it. There is no reason the AI era reverses a twenty-year pattern. The platform automating creative to feed its auction and an independent layer building the advertiser’s creative advantage are complementary, not mutually exclusive. This is also why Anthropic keeping Claude ad-free and OpenAI leaning into ads both point to the same conclusion: the creative layer has to be independent of any one model provider. We cover that split in the AI advertising trust war.
What OpenAI Will Build, and What It Will Not
Putting it together, the line between the two is clear and durable. OpenAI will keep building whatever feeds its auction and monetizes its surface. It will not build the neutral, cross-channel, advertiser-aligned layer, because doing so would work against its own economics.
| Capability | OpenAI will build | Only an independent layer builds |
|---|---|---|
| Basic creative for its own surface | Yes | No |
| On-brand creative for every channel | No (competes with its inventory) | Yes |
| Neutral advice on where to spend | No (conflict of interest) | Yes |
| Independent forecasting and analytics | Unlikely (favors its own spend) | Yes |
| Competitor ad tracking | No | Yes |
What This Means for You
The practical takeaway is not “ignore OpenAI’s tools.” Use them where they help, especially quick localization or a fast variant to feed the auction. The takeaway is that you should not build your advertising operation on a single seller’s creative tools, any more than you would run your whole business on one channel. Your creative advantage, your cross-channel reach, your forecasting, your competitor intelligence, and your attribution are assets you want to own independently of whoever happens to be selling inventory this year.
That is what a neutral layer gives you. Lapis generates on-brand creative for ChatGPT and every other channel from one prompt, forecasts each variant before you spend, tracks competitors, and attributes results across your funnel, all optimized for your return rather than any platform’s auction. Creative quality is where roughly 56% of performance is decided (Meta for Business, 2025), so owning that layer independently is not a hedge; it is the highest-leverage decision in your stack. And because it is not tied to OpenAI, it keeps working as Gemini, Copilot, and Perplexity open their own inventory. For the practical case against relying on the ad platform alone, see why ChatGPT ads alone are not enough.
Getting Started
The way to test this argument is to hold both tools in your hands. Feed OpenAI’s creative helpers a brand and see what they produce for ChatGPT. Then generate the same campaign in Lapis and watch it produce on-brand variants for ChatGPT and Meta, Google, Reddit, and LinkedIn, with forecasts and competitor context attached. One is built to feed an auction; the other is built to grow your business.
Start with Lapis free (5 credits, no credit card). Paste your website, generate a cross-channel campaign, and see what a layer that answers to you rather than to a seller looks like. Lapis is one of the fastest-growing Y Combinator startups (F25), rated 5.0 on G2, with more than 10,000 campaigns generated across 30-plus industries, and it is built to stay independent of any single model provider by design.
Related guides
- Why ChatGPT Ads Alone Aren’t Enough: the case for a creative layer
- Lapis Is Building the AdSense for the AI Era: the bigger thesis
- LLM Ad Infrastructure Explained: who owns each layer of the stack
- The AI Advertising Trust War: why the layer must stay independent
- ChatGPT Ads Complete Guide: how the platform itself works
