The New AI Ad Channels in 2026: ChatGPT Ads, Demand Gen and Where B2B Budget Should (and Shouldn't) Go

Every couple of years a new ad channel arrives with enough noise that clients start asking whether they should be on it yesterday. In 2026 the noise is louder than usual, because the channels turning up are wearing the letters "AI" and the platforms behind them are very keen to sell inventory. Ads now appear inside ChatGPT. Google's Demand Gen has quietly become one of its most useful formats. And Microsoft has bolted LinkedIn's professional data ever more tightly onto its search and audience network. All three are real, all three are worth understanding, and none of them deserves your budget simply because they are new.

This is a straight look at what each channel actually is, the kind of buyer it reaches, and where B2B money should (and shouldn't) go. If you want the short version: for most B2B companies, one of these channels is a much worse fit than the hype suggests, and we'll make that case properly rather than just asserting it.

Ads inside ChatGPT: interesting, but read the tier print

OpenAI began testing ads inside ChatGPT in the US in February 2026, expanding to the UK and several other markets from May. The mechanics are worth getting right because the detail is where the B2B problem lives.

Ads appear below a ChatGPT response, clearly labelled as sponsored and visually separated from the answer itself. During testing only one ad unit shows at a time. Targeting is driven mainly by the topic of your current conversation, and — where a user has personalisation switched on — their past chats and previous interactions with ads. OpenAI does not share conversation content with advertisers; you get aggregated, non-identifying performance data such as views and clicks. Under-18s are excluded, and ads are kept away from sensitive topics like health and politics.

Here is the part everyone skips. Ads only show to users on the Free and Go tiers. Plus, Pro, Business, Enterprise and Edu accounts are ad-free. That single fact reshapes the entire opportunity, and we'll come back to it in detail because it matters more to B2B than any other feature of the platform.

What is ChatGPT-free advertising genuinely good for? Broad reach to a large, engaged consumer audience in a moment of curiosity. It behaves more like a discovery or brand channel than an intent channel — people are asking questions, not sitting in a structured buying journey. For high-volume, low-consideration, wide-audience offers, that is a legitimate and interesting place to show up early.

Google Demand Gen: the mid-funnel workhorse

Demand Gen is the campaign type that has replaced the old Discovery campaigns, and it has matured into something genuinely useful. A single campaign places visual and video ads across YouTube (including Shorts), Discover and Gmail — three of Google's most heavily-used surfaces — using its audience signals and lookalike (now "lookalike segments") tools to find people who resemble your best customers.

The buyer it reaches sits in the middle of the funnel. These are people who aren't searching for you yet but fit the profile of someone who should be. That makes Demand Gen a demand-creation channel rather than a demand-capture one, and it's why comparing its ROAS head-to-head with brand search is usually unfair to both.

For B2B it earns its place in specific roles:

• Warming a defined audience — upload your CRM or first-party lists and build lookalikes, then stay visible to that pool with video and image creative before they ever hit a search box.
• Retargeting with better creative — site visitors, webinar registrants and content downloaders can be re-engaged across YouTube and Gmail rather than only on the display network.
• Supporting a launch or category push where you need reach and narrative, not just a click.

The honest caveat: Demand Gen targeting is audience-based, not identity-based. You can get in front of "people like your customers", but you can't hand-pick job titles the way you can on LinkedIn. Treat it as a reach-and-nurture layer, measure it on assisted conversions and pipeline influence rather than last-click, and it does a real job. Expect it to directly close high-value B2B deals on its own and it will disappoint you. It's one of the services we lean on most for mid-funnel, but always in a supporting role.

Microsoft Advertising and LinkedIn: precise B2B targeting getting more precise

This is the quieter story of 2026 and, for B2B, the most consequential. In June 2026 Microsoft expanded its LinkedIn job seniority targeting to both Search and Audience campaigns, at campaign and ad-group level. There are ten standardised seniority levels — CXO, VP, Director, Manager, Senior, Entry, Owner, Partner, Training and Volunteer — which sit alongside the existing LinkedIn company and industry/function targeting.

Why does this matter? Because Microsoft is the one major search engine that can layer verified professional identity onto search intent. You can now reach someone who is both searching for your category and a director or above in a relevant industry. That combination — intent plus identity — is exactly what B2B has always struggled to buy at scale.

A practical tip: use these filters in observation mode first rather than restricting reach from day one. Let the data show you how conversion quality differs by seniority before you narrow, otherwise you can starve a campaign of volume on a hunch. Combined with Bing's older, higher-earning search demographic, Microsoft deserves a serious look from any B2B advertiser who has written it off as "just the smaller Google". For most of the B2B and B2B SaaS clients we work with, it's an underweighted channel, not an irrelevant one.

The important bit: why ChatGPT-free ads are the wrong fit for most B2B

This deserves its own heading because it's the point most likely to save you money in 2026. For the majority of B2B companies, running ads aimed at ChatGPT's free tier is a poor fit — and it can actively work against reaching your ideal customer profile. The reasoning is straightforward once you line it up.

Your ICP has largely paid its way out of the audience. Ads show on Free and Go only. The senior decision-makers, specialists and budget-holders who make up a serious B2B ICP are disproportionately the people who have already upgraded to Plus, Pro, Business or Enterprise — where no ads show at all, or the experience simply isn't the same. The more valuable the buyer, the more likely their employer put them on a paid or enterprise tier. So the free-tier pool structurally skews away from the exact people you most want to reach.

The intent state is wrong. Someone chatting to a free consumer assistant is usually mid-task, not mid-purchase. They're drafting an email, checking a recipe, debugging code. That's a curiosity context, not a research-and-buy context. For a considered, committee-based B2B purchase, catching a stranger during an unrelated chat is a weak signal to build spend on.

Broad reach means paying for the unqualified. Wide free-tier reach sounds like a feature until you're a niche, high-ACV business. Then it means paying to be seen by a largely unqualified audience, with topic-level targeting that can't distinguish a VP of Operations from a curious student asking the same question. For consumer brands chasing volume, that breadth is the whole point. For most B2B, it's waste.

None of this makes ChatGPT ads a bad channel. It makes them the wrong shape for most B2B right now. Channel choice should follow where your ICP actually is and in what intent state they're in — not where the newest inventory happens to be. Today, ChatGPT-free ads are essentially a broad-reach, consumer-brand play.

When ChatGPT-free ads might make sense

• High-volume, low-consideration offers where a single impression can convert.
• Broad audiences with no tight seniority or company requirement.
• Brand-building at scale, where reach and recall are the goal rather than qualified pipeline.
• B2C, or "prosumer" products that individuals buy on their own card without a buying committee.

When it won't

• Niche categories with a small, specific ICP.
• High-ACV, considered purchases with long sales cycles.
• Committee-based buying where you need to reach several defined roles.
• Anything where the buyers you value are the ones sitting on ad-free paid tiers.

If you're a consumer brand or a high-volume prosumer product, keep an eye on it and consider an early test. If you're classic B2B, watch the space, but don't move budget here on novelty alone.

Where B2B budget should go first

Before any new channel earns a line in the plan, the proven intent-and-identity channels should be fully funded and properly optimised. In our experience that order rarely changes:

1. Google Search for demand capture. When someone is actively looking for what you sell, being there is still the highest-intent moment in the entire funnel. Fix this first — tight query control, strong landing pages, clean conversion tracking — before spending a penny on reach.
2. LinkedIn (and Microsoft's LinkedIn-powered targeting) for precise B2B reach. This is where you can actually name your ICP by seniority, company and function. For most B2B, it's the best combination of identity precision and buying context available.
3. Demand Gen and retargeting as the nurture layer. Once capture and precise targeting are working, use Demand Gen to warm lookalikes and stay visible to known audiences between search sessions.
4. New AI channels as considered tests, funded from experiment budget — not by robbing the channels that pay your pipeline.

A simple decision framework

Before adding any channel, ask three questions in order:

• Is my ICP actually here? If the people you need have opted out of the ad experience (as senior buyers largely have on ChatGPT paid tiers), stop there.
• Are they in a relevant intent state? Researching a purchase beats idly chatting. A curiosity context needs far more volume to justify itself.
• Can I target and measure precisely enough to protect budget? Broad topic targeting on a niche, high-ACV offer is how money leaks.

If a channel clears all three, test it properly. If it fails the first, no amount of novelty makes up for it.

The honest close

2026's new channels are genuinely worth understanding. Demand Gen has become a dependable mid-funnel tool. Microsoft's LinkedIn seniority targeting is a real step forward for B2B precision. And ads inside ChatGPT are an interesting arrival that will suit some advertisers well — just not most B2B ones, at least not yet. The discipline that wins isn't chasing every new surface; it's putting your budget where your ICP genuinely is and treating the rest as measured experiments.

If you'd like a clear view of which of these channels fit your buyers — and which are a distraction — we can map it against your actual pipeline. Start with a free growth plan and we'll tell you honestly where your next pound is best spent.