Two weeks ago, OpenAI started showing ads inside ChatGPT conversations. Target, Adobe, Williams-Sonoma, and Albertsons are the first advertisers. The ads appear below answers for users on the Free and Go tiers -- anyone paying $20/month or more for Plus, Enterprise, or Education sees nothing.
Last week, according to the Financial Times, Perplexity quietly decided to wind down its own advertising program -- barely a year after launching it.
Meanwhile, Anthropic has been running ads of its own -- not inside Claude, but about Claude, emphasizing that its answers are never influenced by advertisers.
Three companies. Three opposite strategies. And each one is making a bet about what AI platforms actually are.
The OpenAI Play: Attention Is the Product
OpenAI's logic is straightforward and familiar: hundreds of millions of people use ChatGPT for free. Free users cost money to serve. Ads offset that cost while keeping the product accessible.
The structure mirrors the Google Search model. Sponsored results appear alongside organic answers, labeled clearly, excluded from sensitive categories. Advertisers cannot access private conversations. OpenAI has hired Kevin Weil's team from Instagram to build the ad product, and early partners include some of the largest retailers in the US.
The minimum buy-in for the pilot program is reportedly $200,000. This is not a self-serve ad platform for small businesses -- at least not yet. It is a premium placement for enterprise brands willing to experiment.
The risk is subtle but real. When you ask ChatGPT to recommend a product, compare services, or evaluate options, the answer now exists in a context where someone paid to be adjacent to it. Even with clear labeling, the line between "answer" and "ad" is thinner in a conversational interface than it ever was on a search results page.
The Perplexity Reversal: Trust Cannot Coexist With Ads
Perplexity tried ads first. It launched sponsored placements in early 2025, built a publisher revenue-sharing model, and positioned advertising as essential to a sustainable AI search business.
Then it changed its mind.
Fewer than 0.5 percent of brands that applied were ever admitted to the program. The ad sales lead, Taz Patel, departed after nine months. And internally, executives concluded that the mere perception of ad influence was enough to undermine what users valued most.
"A user needs to believe this is the best possible answer, to keep using the product and be willing to pay for it," one Perplexity executive told the FT. "The challenge with ads is that a user would just start doubting everything."
Another was blunter: "We are in the accuracy business, and the business is giving the truth. Ads are misaligned with what the users want."
Perplexity is now doubling down on subscriptions and device partnerships -- including a deep system-level integration with Samsung's Galaxy S26 that lets users summon it by saying "Hey Plex." The bet is that distribution deals and Pro subscriptions can replace what advertising would have provided.
Anthropic's Counter-Positioning: Selling the Absence
Anthropic has never run ads inside Claude, and it is making sure everyone knows it. The company has purchased prominent ad placements -- including during the Super Bowl -- specifically to contrast itself with competitors that do.
The strategy is elegant: use traditional advertising to sell the idea that your AI product is free from advertising influence. It positions Claude as the "clean room" option -- the platform where an answer is an answer, not a surface for sponsored content.
Anthropic's revenue instead depends on Claude Pro subscriptions, enterprise API contracts, and partnerships like its recent deal with Goldman Sachs for AI-powered accounting workflows. The model works as long as willingness to pay scales with capability -- and as long as Claude remains competitive enough that businesses choose it over free, ad-supported alternatives.
Google's Fourth Path: AI Enhances the Ad Machine
Google occupies a unique position in this split because it already runs the world's largest advertising business. For Google, the question was never whether AI would touch ads. It was how to keep AI from cannibalizing the search ads that generate most of its revenue.
The answer has been to embed generative AI into its advertising infrastructure. AI-driven creative tools, automated campaign generation, and enhanced targeting through AI Overviews all serve the same goal: make the existing ad business more efficient rather than replacing it with something new.
For businesses that already spend on Google Ads, this changes the tooling, not the model. For businesses trying to decide where to invest attention, it means Google's AI products will always be entangled with its advertising incentives in ways that Anthropic's and Perplexity's are not.
Which Camp Matters for Your Platform Choices
This split is not academic. If your business relies on AI-generated recommendations, research, or customer-facing answers, the monetization model of your chosen platform directly affects output quality and trustworthiness.
If you are evaluating AI platforms for internal use -- research, analysis, decision support -- the ad-free platforms remove a variable. You do not have to wonder whether a recommendation was influenced by who paid for placement. Claude and Perplexity Pro both guarantee this. ChatGPT Plus does too, but only because you are paying enough to skip the ads.
If you are building customer-facing tools on top of AI APIs, the distinction matters less today. API access for all major providers is currently ad-free and pay-per-token. But OpenAI's willingness to introduce ads at the consumer layer signals where things could go at the API layer if unit economics tighten.
If you are a small business considering advertising on these platforms, ChatGPT is the only option right now, and the $200,000 minimum puts it firmly in enterprise territory. Google's AI-enhanced ad tools remain the most accessible path for smaller budgets. Perplexity has exited the game entirely.
The Real Question Is Lock-In
The deeper issue is strategic. Businesses are building workflows, integrations, and institutional knowledge around specific AI platforms right now. The monetization model a platform chooses today will shape its incentives for years.
Platforms funded by advertising will optimize for engagement and attention. Platforms funded by subscriptions will optimize for utility and retention. Platforms funded by enterprise contracts will optimize for reliability and compliance. These are not just business model differences -- they are product design philosophies that compound over time.
We saw this play out with social media. Facebook's ad model shaped its algorithm. Google's ad model shaped search results. The AI platforms making their monetization choices right now are laying the same kind of foundation.
The companies that chose their AI stack based on today's capabilities without considering tomorrow's incentive structures may find themselves rebuilding sooner than expected.