NextFin News - OpenAI is trying to move advertising from the link to the conversation. On September 16, 2026, the company announced Sponsored Agents - ads in ChatGPT that let a user open a clearly labeled, one-on-one conversation with a business-sponsored agent instead of clicking through to a website - alongside AI ad-building tools in ChatGPT Work and native integrations with HubSpot and Shopify. The launch upgrades ChatGPT Ads from a static placement into an end-to-end advertising platform, and it arrives just as the ad business crossed a $1 billion annualized revenue run rate on the way to a $2.5 billion target for 2026.
The stakes run well beyond one product line. ChatGPT has 900 million weekly active users - more than double the 400 million it had a year earlier - and OpenAI is burning cash fast enough that advertising has become one of its few scalable monetization levers. But the deeper story is structural: if AI agents become the front door for product discovery, the company that owns the agent owns the ad slot, and the decades-old duopoly of Google and Meta finally has a credible third competitor. The question is not whether OpenAI can sell ads. It is whether it can move the auction itself.
What OpenAI Actually Launched
The announcement has four parts, and the combination matters more than any single feature. Sponsored Agents let a user who sees a relevant ad choose to start a conversation with a business-sponsored agent inside ChatGPT. The user can ask follow-up questions and follow a link to the merchant's site when ready. OpenAI is careful to note that the sponsored conversation is distinct from ChatGPT's independent answers and separate from the user's original thread. The feature is being tested with select advertisers in the United States.
"Today, we're introducing new AI-powered experiences to make ads more useful for people and advertising easier for businesses."
On the marketer side, advertisers can now create, update, and analyze campaigns in ChatGPT Work using natural-language prompts, and Ads Manager suggests copy and imagery based on the advertiser's landing page and campaign objective. A new text-customization option adapts headlines and descriptions to the context of a conversation and automatically translates ad copy into the user's preferred language.
The distribution piece is the integrations. HubSpot is OpenAI's first CRM partner: businesses can connect a ChatGPT Ads account, create ads, track performance, and follow up on leads directly inside HubSpot using their existing customer context. Shopify is the first ecommerce partner, with a ChatGPT Ads app for US merchants that pulls products from Shopify Catalog so campaigns can launch without rebuilding product data. The app becomes available internationally in ChatGPT Ads markets on September 23, 2026.
OpenAI frames the bundle as infrastructure, not a feature set: "Together, these updates represent the next step in building our AI-based advertising platform."
Why Advertising Is No Longer Optional for OpenAI
The timing is not accidental. OpenAI reported $6.7 billion in revenue for the second quarter of 2026, up from $5.7 billion the quarter before, but it posted a net loss of $38.5 billion in 2025 on $13.07 billion of revenue. Its annualized revenue run rate reached $40 billion by August 2026, yet gross margin sits at 33% and inference costs are projected to rise from $8.4 billion in 2025 to $14.1 billion in 2026. Cash burn is projected at roughly $27 billion in 2026 and $63 billion in 2027, with cash-flow positivity not expected until 2030.
Against that backdrop, advertising is one of the few revenue streams with software-like scalability. The company has told investors it expects $2.5 billion in ad revenue this year, $11 billion in 2027, $25 billion in 2028, $53 billion in 2029, and $100 billion by 2030. The ad business reached $100 million in annualized revenue within six weeks of its US pilot launch on February 9, 2026, and it crossed the $1 billion run-rate mark roughly 200 days later. Those are run rates, not recognized revenue. Hitting a $2.5 billion calendar-year target from a mid-year $1 billion run rate would require averaging more than $540 million a month through the fourth quarter - a steep acceleration. But the trajectory is what matters to investors ahead of a planned 2027 IPO.
There is also a strategic reason to prefer ads over subscriptions. ChatGPT's free tier and its $8-per-month Go plan carry the vast majority of the 900 million weekly users. Raising prices on that base would slow growth; ads monetize free users without putting the product behind a paywall. A company spokesperson said in March that ads were reaching fewer than 20% of eligible US free-tier and Go users on any given day, which means the current revenue represents only a fraction of the addressable inventory.
The Real Prize: Owning the Agent, Not the Ad Slot
The first-order read of this launch is that OpenAI is building a better ad product. The second-order story is that it is trying to own the agent through which consumers discover products - and that is a threat to the economics of search advertising itself.
Google's model sells placement against user intent expressed in keywords. OpenAI's Sponsored Agents sell placement inside a conversation that the advertiser's own agent then steers. A shopper curious about a dining table can ask the sponsored agent whether it fits their space, how many people it seats, and how to care for the finish - all before ever visiting the merchant's site. The advertiser that answers best wins the sale, not necessarily the advertiser that bid highest on a keyword. The value of the ad slot shifts from capturing intent to shaping it.
That shift matters because the market for AI-mediated discovery is already forming without OpenAI. Google launched shopping ads with Direct Offers inside its AI Mode conversational search in February 2026, and its AI Max campaigns eliminate keyword targeting entirely, using Gemini to match advertiser landing pages to user intent. AI Overviews now appear on roughly 48% of Google search queries, up from about 6% in early 2025, and organic click-through rates fall between 34% and 61% when an Overview is present. AI search advertising is on track for roughly $2.08 billion in 2026 - about 1.3% of US search spending - and is projected to reach around $26 billion by 2029, or 13.6% of the search ad market.
OpenAI's advantage is not that it has AI. Google and Meta both have that. Its advantage is that ChatGPT is where users already go for open-ended advice, and its agent is not tethered to a search index or a social graph. Its disadvantage is that it lacks the conversion data - the purchases, returns, and lifetime value - that Google and Meta have spent twenty years accumulating. The HubSpot and Shopify integrations are the down payment on closing that gap. They are not convenience features; they are data pipes.
The Mechanics: Pricing, Self-Serve, and the Long Tail
How the platform is priced reveals how OpenAI plans to scale it. The pilot began with cost-per-thousand-impressions buying at roughly $25 to $60 CPM and minimum spend commitments in the hundreds of thousands of dollars. In April 2026, the company added cost-per-click pricing with starting bids around $3 to $5 per click, and it removed the $50,000 minimum spending requirement when it opened Ads Manager to all US businesses. That sequence - high-touch pilot, then self-serve with a low floor - is the same playbook Google and Meta used to turn advertising from an enterprise sales business into a small-business utility.
The India launch makes the playbook explicit. More than 50 brands are going live with agency support from WPP and Omnicom, and self-serve access opens on September 4 with daily budgets starting at 725 rupees - about $8.50. For a market with hundreds of millions of ChatGPT users, that is a deliberately low barrier designed to pull in the long tail of small advertisers, the same base that built the incumbents' empires.
The risk in that strategy is efficiency. Practitioners running live campaigns report click-through rates between roughly 0.6% and 1%, and an analysis by search-intelligence firm Adthena found a 0.91% CTR for one brand, about seven times below the 6.4% benchmark for traditional Google search in the same sector. Some advertisers also report discrepancies between the clicks OpenAI reports and the sessions that appear in their analytics. OpenAI's own case studies point to stronger outcomes, but the early data suggest the channel is still searching for a repeatable performance formula.
This Is Structural, Not Cyclical
The question investors should ask is whether this is a cyclical revenue boost or a structural change in how advertising is bought. The answer is both, and they need to be separated.
The cyclical leg is the revenue ramp. Any new ad channel grows fast from a small base, and a $1 billion run rate reached in 200 days looks impressive until it is measured against Google's roughly $204 billion in 2026 search revenue and Meta's far larger ad empire. If Sponsored Agents prove annoying to users, or if advertisers do not see a return, the growth will mean-revert - as it did for many of the early metaverse and NFT ad experiments of the 2021-2022 cycle. The arithmetic on the $2.5 billion target is a fair reminder that run rates are not revenue.
The structural leg is the interface change. Advertising follows attention, and attention is migrating from search boxes and social feeds to conversational agents. Once a critical mass of product discovery happens inside agents, the auction that determines which brand gets heard moves with it. That is a regime shift in the plumbing of advertising, and it will not revert on its own. The evidence is that the incumbents are already reacting on OpenAI's turf: Google embedded ads in AI Mode, and Meta announced its own Meta Business Agent in June 2026. When competitors copy the format rather than defend the old one, the shift is real.
The cleanest way to state the call: the revenue number is cyclical and will fluctuate with advertiser demand; the agent-as-gatekeeper model is structural and will persist regardless of OpenAI's share of it.
The Strongest Case Against the Thesis
The bear case is straightforward, and it comes from the economics of the channel itself. Early advertisers report click-through rates well below search benchmarks and, in some accounts, clicks that do not reconcile with their own analytics. A platform that cannot prove return on ad spend will not keep budgets when the novelty fades. Users did not open ChatGPT to be sold to, and a sponsored agent that feels like a salesperson may erode trust in the very product that makes the ad inventory valuable. OpenAI also faces a data disadvantage: Google and Meta can afford to lose money on a new ad format while they tune it, because their core businesses print cash. OpenAI, burning tens of billions of dollars a year, has less room to subsidize a learning curve.
That argument is strong on the revenue line but misses the strategic point. OpenAI does not need ChatGPT Ads to outperform Google Search ads tomorrow. It needs the agent to become the default starting point for product discovery, because once it is, the ad revenue follows whether or not the first wave of campaigns is efficient. The falsifying signal is specific: if AI Overviews and AI Mode ads stall below 10% of Google's search ad revenue by 2028 - roughly $20 billion against the $26 billion projection for AI search - then the agent-mediated discovery thesis is wrong, and conversational advertising is a niche rather than a new layer of the market.
What Comes Next
In the short term, watch the run rate. The $2.5 billion 2026 target requires the ad business to roughly quintuple from its mid-year pace, and the September 23 international rollout of the Shopify app, plus the India launch with more than 50 brands, are the first tests of whether the channel scales outside the US. The India self-serve floor of 725 rupees a day is a deliberate move to open the long tail of small advertisers.
In the medium term, watch the data integrations. HubSpot and Shopify are the first partners, but the platform's real test is whether it can build closed-loop attribution - connecting an agent conversation to a purchase - without the two decades of conversion history its rivals enjoy. Measurement partnerships and conversion APIs will matter more than creative tools.
In the long term, watch who owns the agent. If OpenAI's Sponsored Agents become the template, advertising budgets will follow the agent layer wherever it forms - including inside Google's AI Mode, Meta's Business Agent, and whatever Apple eventually ships. The winner will not be the company with the best ad server. It will be the company whose agent users trust enough to ask for advice.
Base case: ChatGPT Ads becomes a meaningful third channel alongside search and social, capturing a single-digit share of global digital ad spend by 2030 as the agent interface normalizes. Upside case: if sponsored agents prove measurably more efficient than keyword ads, the agent layer captures a disproportionate share of discovery budgets and OpenAI's $100 billion projection starts to look conservative. Downside case: if click-through and conversion rates stay below search benchmarks and users resist commercial agents, the format plateaus as a brand-awareness niche and the revenue ramp stalls well short of target.
OpenAI is not just selling ads inside ChatGPT. It is betting that the future of advertising is not a link on a page but a conversation with an agent - and if it is right, every other ad platform is now playing catch-up.
Explore more exclusive insights at nextfin.ai.
