Advertising on AI Platforms for B2B SaaS 2026: ChatGPT, Google AI Mode & Copilot (and Why Perplexity Pulled Out)

Advertising on AI Platforms for B2B SaaS 2026: ChatGPT, Google AI Mode & Copilot (and Why Perplexity Pulled Out)

Quick answer: A new paid-media category arrived in 2026: ads inside and beside AI assistants. ChatGPT moved fastest: after a February pilot it launched a self-serve Ads Manager on May 5, 2026 with the old six-figure minimum eliminated, CPC bidding around $3 to $5 (CPMs as low as $25), a conversion-optimized objective (oCPC), custom audiences from email and phone lists, and a Conversions API and pixel. Google now shows ads in about a quarter of AI Mode and AI Overview results using your existing Google Ads, and Microsoft Copilot runs ads at reportedly higher CTR than traditional search. Perplexity pulled all its ads and went subscription-only, and Claude will not run ads by design. For B2B SaaS the honest 2026 read is that the buying and measurement tooling has caught up fast, but the audience has not: ChatGPT ads still only reach free and Go-tier users, so if your buyers have paid to opt out, the ad may miss your ICP. The pragmatic stance: about 80% of AI ad value sits beside AI content in the Google and Microsoft ecosystems you already use, with ChatGPT as the one native chatbot placement worth learning early.

Key takeaways

• ChatGPT ads are live (Feb 2026): contextual “Sponsored” text, ~$3 to $5 CPC, now self-serve.
• Google AI Mode shows ads in ~25.5% of results using your existing Google Ads (near-zero setup).
• Copilot runs ads at higher CTR; Perplexity pulled all ads (subscription-only); Claude runs none.
• The free-tier catch: ChatGPT ads miss paid-tier users, who are often your B2B ICP.
• ~80% of AI ad value is beside AI content (Google, Microsoft), not inside chatbots.

For the first time in roughly a decade, a genuinely new paid-media surface has opened: advertising in AI assistants. It matters for B2B SaaS because buyers increasingly build vendor shortlists inside AI tools before they ever visit a website (one survey of 400 senior marketers found AI-native platforms are now the second-most-common source of qualified leads, ahead of organic search and email). But the landscape is fragmenting fast, each platform is on its own timeline, and the hype outpaces the reality. This is the honest 2026 map of what is actually buyable, what it costs, and where it fits for B2B SaaS. (For the how-to decision and test playbook, see the companion piece; this is the landscape.)

The AI-platform ad landscape in 2026

The platforms diverged sharply in 2026, so “advertising on AI” is really several different products at different stages:

The single most important pattern: this is not one channel you buy, it is a fragmenting set of surfaces opening on their own timelines, and two of the most-discussed (Perplexity and Claude) are not buyable at all. So the strategic question is not “which AI assistant do I advertise on,” it is “which surfaces are live, which fit my buyer, and in what order do I approach them.”

ChatGPT ads: the one native chatbot placement

ChatGPT is the one true native, self-serve chatbot ad product, and it moved astonishingly fast. OpenAI announced testing on January 16, 2026, ran an invite-only pilot (originally a ~$200,000 minimum, CPM-only at ~$60), and then on May 5, 2026 launched a self-serve Ads Manager at ads.openai.com with the minimum eliminated and CPC bidding (recommended starting bid $3 to $5, on a relevance-weighted auction where a strong ad on a smaller bid can beat a weak ad on a larger one). Since then the tooling has caught up quickly: three objectives now exist (CPM for reach, CPC for clicks, and oCPC, a conversion-optimized objective that optimizes toward a downstream conversion), plus custom audiences from email and phone lists, US state/DMA/ZIP geo targeting, platform targeting across iOS, Android, and web, automatic advanced matching, view-through reporting, and, crucially, a Conversions API and pixel that let you tie ChatGPT impressions to closed revenue (the user’s chat prompt stays private, so what you get is post-click signal, not pre-click intent). CPMs have fallen as low as $25. The format stays deliberately restrained: contextual “Sponsored” text at the bottom of the response, separated from the answer, targeted by conversational context (“context hints”) rather than keywords, with OpenAI firm on “answer independence” and not selling user data. The scale is staggering: ChatGPT ads reached roughly 51% of US replies by mid-2026, hit a $1 billion annualized revenue run rate in under 200 days, and ChatGPT accounts for around 82.6% of all generative-AI traffic, with the platform rolling out across Canada, Australia, New Zealand, Japan, and (through mid-to-late 2026) the UK, Mexico, Brazil, Korea, India, Europe, and MENA. (Note the EU designated ChatGPT a Very Large Online Search Engine under the DSA in August 2026, bringing ad-repository and targeting-disclosure duties.) What is still missing versus Google and Meta: granular demographic and lookalike targeting and placement controls, and full audience-syncing for retargeting (expected later in 2026), so context-hint writing and creative relevance do more of the heavy lifting here than bidding levers do.

The free-tier problem most coverage skips

Here is the structural catch that decides whether ChatGPT ads even reach your buyers: ads only appear for users on the free and Go tiers. Subscribers on Plus, Pro, Team, and Enterprise plans see no ads at all. Globally, a large majority of ChatGPT users (roughly 80%+) are on the free tier, which sounds like broad reach, but for B2B SaaS the relevant question is where your ICP sits. If your buyers are the kind of technical or senior professionals who have already paid for ChatGPT Plus, Team, or Enterprise, they have effectively opted out of your ads, and the free-tier audience you can reach may skew away from your best prospects. This is the most important and least-discussed B2B nuance of ChatGPT ads: the ad audience is not “everyone on ChatGPT,” it is “the free-tier subset,” and whether that subset contains your buyers is a company-specific question you must answer before spending. For some B2B SaaS the free tier is full of relevant early-stage and individual users; for others, the ICP has largely paid to remove ads. Check your own buyer’s likely plan tier before assuming reach. This is not hypothetical: one agency’s June 2026 B2B test on ChatGPT produced 8,940 impressions, 58 clicks, a 0.65% click-through rate, and zero leads, and attributed the result directly to the free-and-Go-only constraint. As one field guide put it, the buying and measurement stack has caught up with what planners need, but the audience has not: until paid-plan users are reachable, ChatGPT ads are effectively a consumer channel with B2B-shaped tooling.

Google AI Mode and Microsoft Copilot: where the near-term value is

The less flashy but more practical opportunity is the AI advertising bolted onto systems you already use. Google now shows ads in roughly 25.5% of AI Mode and AI Overview responses (up from about 5.17% in early 2025), including shopping ads and Direct Offers, monetizing AI search at a similar rate to traditional search, and crucially these run on your existing Google Ads campaigns and assets, so the setup cost is near zero. Microsoft Copilot likewise serves ads within its responses (reported at materially higher CTR than traditional search) through its existing bidding. This is why an estimated 80% of 2026 AI ad spending is expected to appear beside AI content (like ads next to Google AI Overviews) rather than inside standalone chatbots. For B2B SaaS the implication is clear: most of the near-term, accessible AI ad value lives in the Google and Microsoft ecosystems you already operate, so making sure your Performance Max feeds, assets, and Microsoft Advertising campaigns are healthy is the first and highest-return “AI advertising” move, well before you build a native ChatGPT program.

Why Perplexity pulled out (and what it signals)

The most instructive story of 2026 is the platform that walked away. After testing sponsored follow-up questions and answers through 2024 and 2025, Perplexity abandoned advertising entirely in February 2026, went subscription-only, and repositioned as the ad-free alternative, targeting substantial subscription revenue instead. Its reasoning is the key signal for the whole category: sponsored placements near AI answers risked making users doubt the neutrality of the product, and Perplexity judged answer confidence more valuable than the ad slot. This is the tension at the heart of AI-platform advertising. When someone asks an AI assistant for a recommendation, they are borrowing judgment, so an ad that feels like it tilts the answer risks “trust leakage” that damages the platform (and the advertiser). It is why OpenAI is proceeding cautiously with answer independence and clear labeling. For advertisers, the practical takeaway is that Perplexity is not a paid channel in 2026 (it is a citation engine you reach through AEO, not ads), and that the platforms which do sell ads are engineering hard to keep ads separate from answers, which shapes what these ads can and cannot do.

What this means for B2B SaaS in 2026

Pulling the landscape together into a practical stance:

• The near-term value is in Google and Microsoft. Get your Google AI Mode exposure (via existing Search and Performance Max) and Microsoft Advertising healthy first; that is where roughly 80% of accessible AI ad value sits and it needs little new setup.
• ChatGPT is worth learning early, carefully. As the one native self-serve chatbot placement, it is worth a small, measured test to learn the format while it is cheap, but only after checking the free-tier ICP question and with realistic expectations about immature measurement.
• Perplexity and Meta AI are organic-only or watch-list. You reach Perplexity through AEO citation, not ads; keep Meta AI on a watch list as it may open.
• Claude is organic-only by design. Reach it, and the others where ads are limited, through answer-engine optimization.
• Paid and organic AI presence reinforce each other. Being cited organically in AI answers strengthens the resonance of paid placements in the same context, so the winning approach pairs AI ads with AEO rather than choosing between them.

The honest summary: AI-platform advertising is a legitimate, fast-growing new category (US AI-search ad spend is projected to grow from about $1B in 2025 to roughly $26B by 2029), but in 2026 it is early, fragmented, and unevenly buyable. Treat it as an additional layer to learn and test, weighted toward the Google and Microsoft surfaces you already run, not as a place to move your working budget.

Field note: The phrase “advertising on AI” collapses several very different realities into one hype cycle, and untangling them is most of the value for a B2B SaaS marketer right now. There is the flashy version (buying ads inside ChatGPT) and the boring, valuable version (the ads Google is already quietly serving in a quarter of AI Mode results through your existing campaigns), and roughly 80% of the actual money is in the boring one. There is the platform everyone assumes you can advertise on (Perplexity) that actually walked away from ads entirely, and the one nobody expected to move fast (ChatGPT) that went from a $250K enterprise pilot to self-serve in ten weeks. And there is the catch that decides whether any of the chatbot spend reaches your buyers at all: the ads only show to free-tier users, and a lot of B2B ICPs have already paid to make them disappear. The teams that will do well here are not the ones who rush to “advertise on ChatGPT” because it is new; they are the ones who make their Google and Microsoft AI exposure healthy first, run a small honest ChatGPT test after checking whether their buyer is even in the ad audience, reach the non-buyable platforms through AEO, and treat the whole category as an early layer to learn rather than a bet to make. New ad surfaces reward the prepared and punish the impulsive, and the preparation here is mostly clear thinking about which surface reaches whom.

Honest limitations

• The landscape is changing fast. Platform ad status, pricing, and features shifted repeatedly through 2026 and will keep moving; confirm current specifics before planning.
• Measurement is immature. AI-platform ad reporting and attribution are far younger than Google’s or Meta’s, so treat early performance data cautiously.
• Reach is uneven and tier-gated. ChatGPT ads miss paid-tier users; whether the reachable audience contains your ICP is company-specific.
• This is a landscape, not a benchmark. Costs are directional and early; validate with your own test rather than treating figures as stable benchmarks.
• Educational, not investment or financial advice. Validate against your own account and buyer.

Frequently Asked Questions

Q1. Can you advertise on ChatGPT in 2026?

Yes. OpenAI began testing ChatGPT ads in January 2026 and launched in the US in February 2026, moving from a roughly $250,000-minimum enterprise pilot to a self-serve Ads Manager with CPC bidding around $3 to $5 per click in about ten weeks, later opening to more markets. The format is a contextual “Sponsored” text ad at the bottom of the response, clearly labeled and separated from the answer, targeted by conversational context rather than keywords. Two caveats: ads only reach free and Go-tier users (not paid subscribers), and there is no website-visit retargeting yet, with measurement tooling still immature.

Q2. Can you advertise on Perplexity?

No, not in 2026. After testing sponsored follow-up questions and answers through 2024 and 2025, Perplexity abandoned advertising entirely in February 2026, went subscription-only, and repositioned as the ad-free alternative. Its reasoning is instructive for the whole category: sponsored placements near AI answers risked making users doubt the neutrality of the product, and Perplexity judged answer confidence more valuable than the ad slot. So there is no Perplexity ad product to buy. You reach Perplexity users through answer-engine optimization (earning organic citations), not through paid placements.

Q3. What is the biggest catch with ChatGPT ads for B2B SaaS?

That ads only appear to free and Go-tier users, while Plus, Pro, Team, and Enterprise subscribers see none. Globally most ChatGPT users are on the free tier, but for B2B SaaS the question is where your ICP sits: if your buyers are technical or senior professionals who have already paid for ChatGPT, they have effectively opted out of your ads, and the reachable free-tier audience may skew away from your best prospects. The ad audience is not “everyone on ChatGPT,” it is the free-tier subset, so check whether that subset contains your buyers before you spend.

Q4. Where is the near-term value in AI advertising for B2B SaaS?

In the Google and Microsoft ecosystems you already use. Google now shows ads in roughly 25.5% of AI Mode and AI Overview results using your existing Google Ads campaigns and assets (near-zero setup), and Microsoft Copilot serves ads through its existing bidding at reportedly higher CTR than traditional search. An estimated 80% of 2026 AI ad spending appears beside AI content like this rather than inside standalone chatbots. So the first and highest-return “AI advertising” move for B2B SaaS is making your Performance Max feeds, assets, and Microsoft campaigns healthy, well before building a native ChatGPT program.

Q5. How much do ChatGPT ads cost?

As of 2026, ChatGPT ads run roughly $3 to $5 per click (around $60 CPM for some placements). The channel moved quickly from an enterprise-only pilot with a reported $250,000 minimum spend to a self-serve Ads Manager with CPC bidding accessible to smaller advertisers, so a modest pilot (around $5,000 over a defined window) is now feasible. Treat these as early, directional figures rather than stable benchmarks, because pricing and access changed repeatedly through 2026. Google AI Mode and Copilot ads run on their existing auction pricing since they use your current Google and Microsoft campaigns.

Q6. Is Claude going to have ads?

No. Claude will not run ads by design, so like Perplexity it is not a paid channel. The way to be present for Claude users (and for Perplexity, and within the answers of platforms that do run ads, since those ads sit beside rather than inside the answer) is answer-engine optimization: earning organic citations by having content the models trust and reference. This is why the durable AI-visibility strategy pairs paid placements where they exist with AEO across all platforms, rather than relying on ads alone, since several major AI surfaces cannot be bought at all.

Q7. Should B2B SaaS advertise on AI platforms yet?

Cautiously and selectively, yes, treating it as an early layer rather than a core bet. The category is legitimate and fast-growing (US AI-search ad spend is projected to rise from about $1B in 2025 to roughly $26B by 2029), but in 2026 it is early, fragmented, and unevenly buyable. The pragmatic stance: prioritize the Google and Microsoft AI surfaces you already access (where ~80% of the value sits), run a small measured ChatGPT test after checking the free-tier ICP question, reach non-buyable platforms through AEO, and pair paid with organic AI presence. Do not move working budget out of proven channels to chase it.

Sources & further reading

• DigitalApplied / Stackmatix (Google AI Mode ads in ~25.5% of results; Perplexity pulled ads Feb 2026; US AI-search ad spend ~$1B 2025 to ~$26B 2029); Lapis (ChatGPT the one native self-serve chatbot ad; ~80% of AI ad spend beside AI content; Claude none by design).
• Omni Lab (free-tier-only reach; ~80%+ of ChatGPT users on free tier; B2B ICPs often paid out; scale figures); Monks / SmartAcre / VertoDigital (ChatGPT ads timeline, format, ~$3 to $5 CPC, answer independence, no retargeting yet, $250K pilot to self-serve).
• Companion: Should B2B SaaS Advertise on ChatGPT? A Decision and Test Playbook; the AI Overviews paid-search benchmark and playbook (impact on existing Google search).

This guide is educational, not investment or financial advice; the AI-platform ad landscape is changing fast and measurement is immature, so treat costs as directional and validate against your own account and buyer.

Related guides: Should B2B SaaS Advertise on ChatGPT? A 2026 Decision and Test Playbook · AI Overviews & B2B SaaS Paid Search: The 2026 Benchmark Data · Paid + AEO for B2B SaaS: How AI-Overview Citations Lift Paid Clicks · The Cross-Platform Paid Waste Benchmark for B2B SaaS 2026 · Google Ads Benchmarks by Campaign Type: Brand vs Competitor vs Category.

Ishan Manchanda

Turning Clicks into Pipeline for B2B SaaS · Founder, GrowthSpree