Three of the most significant advertising platforms right now are all AI-powered, all competing for the same pot of money, and all asking advertisers to learn something new before they can use them well. Google is running ads inside AI Mode and AI Overviews. ChatGPT opened self-serve advertising on May 5. Perplexity, which had been quietly testing ads, walked the whole thing back and pulled advertising from its platform entirely.
For a US advertiser trying to figure out where to put money in the second half of 2026, the noise around all three of these is genuinely hard to cut through. This article is an attempt to do that.
Google Is Still Where Most of Your Budget Should Sit
This is not a controversial statement, but it is worth saying directly before talking about anything else. Google processes somewhere around 14 billion searches per day. Its ad infrastructure has been refined over two decades. The targeting controls, the attribution tooling, the audience data, and the sheer volume of commercial intent flowing through it are not things ChatGPT or any other platform comes close to matching in 2026.
The more interesting question is not whether Google deserves most of your budget. It is whether the Google you knew how to run twelve months ago is the same Google you are running today.
It is not, for two reasons. The first is AI Mode and AI Overviews. Ads now appear in 25.5% of AI Mode results, up from just over 5% in early 2025. The users in AI Mode are behaving differently from users in standard search. They are asking longer, more exploratory questions. They are reading full paragraphs before they see your ad. The copy and landing page strategies that work in standard search do not automatically translate. We covered what that means for campaign structure and ad copy in detail in our post on why your Google Ads are not showing in AI Overviews.
The second reason is the AI Max transition. Google is retiring Dynamic Search Ads in September 2026 and auto-upgrading campaigns to AI Max for Search, whether advertisers are ready or not. Accounts that are not prepared for that migration are going to see performance disruption in Q3, heading into what is the most important quarter of the year for most US businesses. If you have not audited your DSA campaigns yet, that work belongs at the top of your list. We broke down the full timeline and what to do about it in our AI Max transition guide.
Google is still the anchor of a paid search budget in 2026. But anchoring to the wrong version of Google, the one from 2024, is how accounts lose ground without understanding why.
ChatGPT Ads: Real, but Early
ChatGPT Ads launched self-serve on May 5 with no minimum spend, CPC and CPM bidding, a measurement pixel, and a Conversions API. The platform hit $100 million in annualized revenue within weeks of launching. Those are real numbers, and they reflect genuine advertiser interest in reaching the hundreds of millions of people using ChatGPT every week.
The platform is genuinely different from anything most advertisers have run before. There are no keywords to bid on. OpenAI’s system reads the conversation context and decides when your ad is relevant. You do not control the matching. You control your bid, your creative, and your daily budget, and you receive aggregated performance data in weekly CSV files. Retargeting individual users is not possible. Audience segmentation does not exist yet.
What works on ChatGPT Ads is a specific, intent-matched creative for users who are already deep in a research conversation. The standard Google Ads headline, written to be scanned in under two seconds, underperforms in a context where the user has just read three paragraphs. Offers with clear conversion actions outperform brand awareness campaigns, partly because the measurement infrastructure for brand lift on the platform does not exist yet.
For B2B software, professional services, education, and home services, the case for an early ChatGPT Ads test is reasonably strong. For retail ecommerce with large catalogs and tight ROAS requirements, waiting another six months makes more sense.
The allocation most US advertisers should consider is somewhere between 10% and 15% of their total paid search budget, structured explicitly as a test, with the pixel and Conversions API wired up from day one, and a minimum of four to six weeks of run time before drawing any conclusions. We covered the full setup in our ChatGPT Ads guide.
What Perplexity Dropping Ads Tells You
Perplexity walked away from advertising and announced it would pursue a subscription and API model instead. The reason given publicly was that ads created tension with the user experience of a platform built around direct, sourced answers.
That decision is worth understanding on its own terms rather than as a negative signal about AI advertising overall. Perplexity’s core value proposition is accuracy and citation transparency. Ads introduce a commercial layer that its user base appears to be particularly resistant to. That is a product-specific problem, not a platform-wide one.
What it does confirm is that not every AI platform will become an advertising channel, and budget allocation decisions should be based on platforms that have committed to advertising as part of their model, not ones that are still deciding. Google and ChatGPT are both committed. Everything else is speculation.
Organic Search Still Matters, and the Rules Changed There Too
Budget allocation across paid channels is only half the picture. The March 2026 core update shifted how Google evaluates organic content in ways that affect how much work your paid ads have to do.
Sites that lost organic traffic after March 2026 tend to be the ones with broad, shallow content written across too many topics. The sites that held or gained ground are the ones with genuine depth in a specific subject area. That shift in organic is directly relevant to paid search because your landing pages, which Google evaluates as part of ad relevance in both standard search and AI surfaces, need to reflect the same depth. A paid ad pointing to a thin landing page is fighting the same battle on two fronts.
Building topical authority on your site does not just recover organic rankings. It strengthens the content that your AI Max campaigns draw from, improves the conversations that Business Agent for Leads generates, and makes your pages more likely to be surfaced in AI Overviews alongside your ads. We covered how to build that authority in the context of the March update in our post on topical authority after the Google core update.
Paid and organic are being evaluated by the same AI systems now. Treating them as separate workstreams is increasingly the wrong frame.
A Practical Budget Framework for the Rest of 2026
There is no universal allocation that works for every business, but there is a logic that most US advertisers in performance-driven categories can follow.
The majority of paid search budget belongs in Google, but it needs to be the right version of Google. That means having a migration plan for the AI Max transition before September, having campaigns that are structured to appear in AI Overviews, and being set up for Business Agent for Leads if you are in automotive, education, or real estate.
A testing allocation of 10% to 15% in ChatGPT Ads is defensible for most categories right now, with specific attention to copy that fits a conversational context and a conversion goal that can be tracked through the pixel. Businesses in categories that skew toward research-heavy buying decisions will get more out of this test than those in impulse-purchase categories.
Organic investment should run in parallel with paid, not separately. The content you build for topical authority is the same content that improves your AI Max campaign performance, your Business Agent conversations, and your AI Overview ad relevance.
The AI advertising market is projected to grow from $35 billion in 2025 to $142 billion by 2030. The platforms and formats that will capture that growth are being built right now. The advertisers who understand how each platform works, what it can actually do today, and what it cannot yet do, will make better decisions about where to put money in the meantime.