Shopify's Summer '26 Edition has 150+ updates. Here are the five worth your time.

Twice a year Shopify gathers up everything it's shipped, wraps it in a launch video, and calls it an Edition. The Summer '26 one streamed on June 17 with more than 150 changes attached to it. That number is the point of the format and also the problem with it. A hundred and fifty updates isn't a product release you read. It's a firehose you're meant to feel excited by.

Most of those 150 things will never touch your store. Five of them might: a new theme system, checkout extensibility going fully live for Plus, a set of AI merchandising tools, native A/B testing, and built-in B2B payment terms. So the useful question isn't "what's new in Shopify." It's "what's new that changes a decision I'm about to make." Here's what each one actually changes, and which are worth your attention now versus later.

Horizon and the slow death of theme code

The headline feature is Horizon, Shopify's block-based theme system, and the thing it changes is who gets to touch your storefront layout. Older themes like Dawn let you nest blocks two levels deep. Horizon goes up to eight. That sounds like a developer detail, but in practice it's the difference between "we need to edit the theme code to do that" and "someone on the team can build it in the editor on a Tuesday afternoon."

Shopify also wired its AI into the theme editor. You can describe a section you want, something like a testimonial block with star ratings, and it builds the block for you. There are ten free Horizon presets aimed at specific industries, each with its own layouts and type choices, so a new store can start from something closer to its category than a generic template.

Here's the part the launch video skips. Re-theming a store that already converts is a real risk, not a free upgrade. The new layout might look better and convert worse, and you won't know until traffic tells you. If you're launching a new store or your current theme is genuinely holding you back, Horizon is the right place to build. If your existing storefront is doing its job, the move is a project to plan and measure, not a button to press because Shopify made a nice video about it.

Tied to this: checkout extensibility is now generally available to every Plus merchant. The post-cart flow you used to customize by editing checkout.liquid is now built from blocks, the same way the rest of the store is heading. If you're a Plus store still sitting on the old checkout, this release is the official nudge. The old way is on borrowed time, and the new way is finally stable enough to commit to.

The AI features, and the one that lets you check them

Shopify built three merchandising tools into the admin: AI Collection Sort, which orders products in a collection automatically; predictive cross-sell blocks, which pick the "goes well with this" items for you; and a merchandising insights panel that tells you how your collections are performing. All three take work off your plate, which is the pitch and also the catch.

Automated sorting optimizes for whatever metric it's told to, and that isn't always the one you'd choose. An algorithm chasing click-through can bury a high-margin product that sells slowly but matters to your business. It can also surface whatever's trending this week and quietly retire something you were building a reputation on. The tools are probably good. "Probably" is the operative word, and it's your store, so probably isn't enough.

Which is why the genuinely useful AI-adjacent feature in this Edition is the boring one: native A/B testing, branded Rollouts. You can now split-test themes, checkout, and customer-account changes inside Shopify, without bolting on a third-party tool. That matters because it removes the last excuse for turning AI merchandising on blind. Run the AI sort against your current order on half your traffic, watch revenue per visitor for two weeks, and let the numbers decide. The AI features are the flashy headline. The A/B testing is the thing that makes them safe to use.

Native B2B terms, which is a bigger deal than it sounds

Quieter than Horizon, and more interesting if you sell wholesale: net terms with automated invoicing are now built into Shopify B2B. Letting a business customer pay on net-30 or net-60, with invoices that go out on their own, used to mean a separate app or a manual finance process bolted onto the side. Now it's native.

The pattern worth noticing is what it means when Shopify absorbs a feature category. Every time the platform makes something native, a class of apps that charged a monthly fee to do exactly that becomes harder to justify. If you're paying for a B2B payment-terms app, this is the release to check whether you still need it. If you've been avoiding wholesale because the billing side looked like a hassle, the hassle just got smaller. Either way it's the kind of unglamorous change that shifts what's worth building, which is usually more durable than whatever the AI demo showed.

How to actually decide what to turn on

The trap with any Edition is treating it as a shopping list where adopting more equals doing better. It doesn't. Each new feature you switch on is something to configure, monitor, and eventually debug. The merchants who get the most out of these releases are the ones who adopt the least, slowly.

A reasonable order of operations from this list: start with whatever removes a cost you're already carrying, like a B2B app you can drop now that terms are native, or a checkout customization you were about to pay a developer for. Then, and only then, experiment with the growth features, with A/B testing switched on so an experiment stays an experiment. Re-theming with Horizon sits in the middle, worth doing if you're rebuilding anyway and easy to defer if your store already converts.

Everything else in the 150 can wait until you have a reason to care about it. A feature you adopt because it solves a problem you actually have tends to stick. A feature you adopt because it was in the launch video tends to become the thing nobody remembers configuring when it starts behaving strangely six months later. Read the Edition the way you'd read a tool catalog, not a news feed: most of it isn't for you, and the discipline is in knowing which parts are.