Google and Microsoft Both Call It AI Max. Where They Differ

September 2026 is the month Google begins automatically upgrading Search campaigns that use campaign-level broad match or automatically created assets to AI Max, according to the timeline Google published in April and revised in June. It is also the first full month in which Microsoft Advertising’s product of the same name, AI Max for Search, is generally available after an open pilot that began in May. The two products share a name, a broad purpose and a marketing vocabulary. They differ in the one thing that matters to an account owner this month: Google’s is happening to you, and Microsoft’s is waiting for you.

This post is not a walkthrough of either interface. It sets the two out side by side on the questions that decide a budget: what each does to a campaign, where the ads run, what controls you keep, what each vendor claims, and what to do in the next four weeks. Where a figure comes from a vendor it is labelled as the vendor’s; where no primary figure exists we say so rather than quote one from a benchmark page.

On the Google side the facts are in a single announcement, “Dynamic Search Ads are upgrading to AI Max,” published April 15 and updated June 11. Campaigns that use automatically created assets or the campaign-level broad match setting are auto-upgraded starting September 2026. Dynamic Search Ads, the original target, were pushed to February 2027 after advertiser feedback. The same post declares AI Max “moving out of beta, with improved performance quality across targeting and creative capabilities,” and adds brand restrictions, location targeting and text guidelines as the controls advertisers get. The post offers no opt-out for the September cohort; the choice it gives is to upgrade voluntarily first, so that the configuration is yours rather than the default. Our June playbook on the delayed DSA migration covers the DSA cohort; this month is about the other two.

On the Microsoft side the equivalent document is the August 2026 product news post, which moves AI Max for Search to general availability and describes it as “an optional suite of Search campaign features” with a toggle to enable it. Nothing in an existing Microsoft account changes unless someone turns it on. The same post carries two items that will change accounts regardless: Smart Shopping campaigns are being deprecated and auto-upgraded to Performance Max from August, and a dual-identifier requirement for account linking became mandatory for API calls on August 26.

The cleanest way to see that these are different products is to ask where an impression happens. Google’s AI Max is a way of matching and rendering Search ads more flexibly on Google Search. Microsoft’s is that too, on Bing, but its April announcement leads with the other two surfaces: “Copilot Search and Copilot Answers, where people’s questions are more nuanced and complex.” Turning on AI Max for Search in Microsoft is, in practice, the decision to let your Search campaigns appear inside an AI assistant’s answers, with the matching model deciding which nuanced questions your ad fits.

That is a materially different placement from a results page. The query is longer, the intent is often mid-decision rather than navigational, and the ad sits next to a generated answer rather than a list of links. Microsoft’s companion formats point the same way: Offer Highlights surface details like free shipping “within the context of Copilot conversations,” and Copilot Checkout and Brand Agents, launched in January, complete purchases inside the assistant. An advertiser who wants broader matching on Bing but has not decided about assistant placements does not yet have a way to separate the two in Microsoft’s product as described; the surfaces come together. Google’s equivalent question, ads inside AI Mode, is a separate programme and is covered in our AI Mode ads study.

Both vendors answer the same objection, that broader matching takes decisions away from the advertiser, with the same three words: brand, location, text. Google names brand restrictions, location targeting and text guidelines as shipped controls and frames them as letting advertisers “steer AI with greater precision.” Microsoft’s April post promises brand inclusions and exclusions, term exclusions and messaging constraints, but in the future tense, “soon,” and its August GA post does not update the list. If those controls matter to you, confirm in the account which have arrived before enabling the suite. Our guide to testing Google’s brand and location controls in experiments describes the test design; the same design works on Microsoft once the controls exist.

On reporting the two platforms took different paths that reflect their broader strategies. Google added migration-date fields to the Ads API in v25.1, which is the practical tool for finding out which campaigns are scheduled for upgrade and when; we covered the fields when they shipped in August. Microsoft shipped a Model Context Protocol server, an interface that lets AI assistants call a service’s tools, so that Microsoft 365 Copilot, ChatGPT and Claude can “run reports, campaign audits, keyword research” against live account data. It is read-only and available as an open flight with public setup guidance. Microsoft cites Stagwell cutting audit timelines “from hours to minutes” and Realtor.com wiring it into Claude-based workflows. For an agency running both platforms, this is the more interesting release of the two: it is the first time a major ads platform has handed its account data to whichever assistant the buyer already uses.

Most search budgets are split heavily toward Google, and the usual justification is reach. What September changes is not reach but the matching model. Once both platforms match queries the same way, with an AI system deciding which long, specific questions an ad fits, the historical argument that Microsoft needed separate keyword work to be worth the effort weakens. Enabling AI Max for Search on an existing Microsoft import is close to free in labour, and it opens surfaces Google does not sell in the same product. The LinkedIn Job Seniority attribute Microsoft added in August, ten standardised levels, is the other half of that argument for business-to-business advertisers, and we set out the targeting case in our LinkedIn seniority targeting guide.

The honest counterweight is measurement. Google publishes a lift figure for its suite, with its comparison basis stated; Microsoft does not publish one for AI Max at all. A budget decision should rest on an experiment in your own account on each platform, run for long enough to see conversion volume, and read with the vendor’s claims set aside. That is slower than reading a percentage off a blog, and it is the only method that survives contact with a real account.