Hydrogen Is Finally Ready and Most Plus Stores Still Shouldn't

Hydrogen is finally a credible default. The 2026.1 release on React Router v7 stabilised the developer experience, the April 9 release locked the Storefront API proxy and consent-mode contracts, and Allbirds is running production storefronts across 35 countries on it (Hydrogen, April 2026 release notes; Allbirds case via Shopify's own Hydrogen marketing). Gymshark, SKIMS, Good American, and Denim Tears are all live. The framework is no longer the early-adopter wager it was in 2022.

And most Shopify Plus merchants in 2026 still should not use it.

That sentence will read as backwards to anyone who has been pitched a headless rebuild in the last six months, so it is worth being precise about why. The argument has three parts: the total cost of ownership most agencies do not surface in their pitch decks, the things OS 2.0 themes have quietly closed since 2023 that used to be the headless argument, and the direction Shopify itself is investing in (AI commerce, Storefront MCP, Sidekick, agentic surfaces) which mostly does not require Hydrogen to access.

The cost number nobody puts on slide three

Weaverse compiled a 2026 cost audit across multiple Shopify Plus Hydrogen projects in the $5M to $20M GMV range and landed on a year-one all-in of $300,000 to $600,000 (Weaverse, "Shopify Headless Pricing 2026"). That is not the build. That is the build plus the things that follow the build.

The professional services for the build sit at $120,000 to $300,000. App retrofitting (because half your existing Shopify apps do not run on a custom storefront) adds $20,000 to $80,000. CMS integration if you need real content tooling adds $10,000 to $40,000. Then ongoing engineering settles at $15,000 to $40,000 per month, and infrastructure adds $500 to $3,000 per month if you use Vercel rather than Oxygen.

The same data set puts the median timeline at 8 to 14 months from contract to feature parity. A focused Hydrogen build on an existing Plus store runs 2 to 4 months. A full Liquid-to-Hydrogen migration with the custom integrations most Plus merchants actually have runs 4 to 7 months.

For a $10M GMV store, that year-one number is between three and six percent of revenue. The follow-on years are not free either; engineering retainers and infrastructure for Hydrogen tend to settle at $200,000 to $500,000 annually once you account for security patches, version upgrades (Hydrogen ships on a calendar cadence now), and the in-house developer time pulled into framework work that used to go to features.

None of this is hidden. It is just rarely on the slide deck.

What OS 2.0 closed

The original Hydrogen pitch leaned on three weaknesses of liquid themes: speed, design flexibility, and the inability to compose modern frontend tooling. By 2026, two of those three have largely been fixed at the OS 2.0 layer and the third is narrower than it used to be.

Theme speed is the clearest example. Tobi Lütke himself shipped a coding-agent-driven optimisation that made Liquid 53 percent faster on common rendering paths in 2026 (covered in trade press; the underlying commit is in the Shopify theme store reference repos). The Horizon family shipped in 2025 with eight levels of nested blocks against Dawn's two, AI block generation through Sidekick, and component-driven layouts that close most of the design-flexibility gap that drove DTC brands toward React in 2022 and 2023.

The composability gap is real and remains. If you genuinely need a custom configurator, a quiz funnel that holds dozens of state transitions, multi-region storefronts with per-locale content models, or content-commerce at editorial scale, OS 2.0 is still the wrong tool. But the merchants who actually have those needs are a small fraction of the merchants who get pitched headless.

What Shopify is signaling about the next two years

Read Shopify's own platform investment instead of agency pitch decks. The most material storefront work in the last twelve months has been Agentic Storefronts, Storefront MCP, AI block generation in Horizon themes, the Knowledge Base app, Sidekick, and the Customer Account API expansions. Almost none of it is gated behind Hydrogen. Most of it is delivered to OS 2.0 stores by default.

The exception worth flagging is Storefront MCP, which lets brands wire custom AI shopping agents directly into a storefront on their own domain (Hydrogen update, December 2025 / Special Winter '26 Edition). MCP integration on Hydrogen is genuinely cleaner than retrofitting it into a Liquid theme. For a brand whose 2026 strategy is built around a domain-native AI agent, that is a real headless argument. For everyone else, the AI commerce story is happening on the standard storefront and showing up in ChatGPT, Perplexity, Copilot, Google AI Mode, and Gemini without any Hydrogen work at all.

Harley Finkelstein has been consistent on this in public commentary throughout 2026. The framing in the Spring Editions material and his Shoptalk remarks was not "go headless." It was that AI-referred traffic to Shopify is up 7x since January 2025, that orders flow back into the standard admin with full attribution, and that the platform's direction is to make the default storefront better rather than to push merchants off it. That is a strong signal about where Shopify is allocating engineering and where the path of least resistance for merchants will be.

Where Hydrogen is the right answer

The honest list of cases where the math works in 2026 is narrow and consistent.

The first is annual revenue above $10M with a real in-house engineering team. Without internal capacity, the ongoing engineering retainer is not optional; you will spend it on an agency every month. With internal capacity, the marginal cost of going headless is significantly lower because the engineers are already on payroll.

The second is a concrete product reason that OS 2.0 cannot deliver. Bundle configurators with stateful logic. Multi-region storefronts with per-locale catalog and content models. Content-commerce at the depth of an editorial site. B2B checkout flows where the buying party model breaks the standard checkout assumptions. A configurator-driven category like furniture, custom apparel, or made-to-order. These are real and they exist. They are also a small slice of the Plus merchant population.

The third is brands building toward a domain-native AI commerce experience that goes beyond what Storefront MCP plus standard surfaces deliver. This is genuinely emerging. Most merchants will not hit the threshold for another twelve to eighteen months.

If a Plus merchant fits one of those three, Hydrogen 2026.1 plus Oxygen is a defensible build. The framework is mature, the deployment is clean, and the gains are real. Gymshark reported 30 percent faster load speeds and a 30 percent conversion lift after migration (Shopify case study via Hydrogen marketing materials). Allbirds is running 35 countries off a unified Hydrogen codebase. The wins are not theoretical.

If a merchant does not fit one of those three, the same $300,000 to $600,000 spent on conversion rate optimisation, paid acquisition, retention infrastructure, and OS 2.0 storefront polish returns three to five times more revenue. That is the math the Ask Phill audit landed on across multiple post-mortems, and it tracks with what we have seen in our own client work.

The composition pattern that is actually winning

Among the Plus merchants having a strong 2026, the dominant pattern is not full headless. It is selective composition. Keep the core storefront on a polished OS 2.0 theme (Horizon family is the increasingly common choice for its block flexibility). Move only the pieces that genuinely require frontend control to a Hydrogen-on-Oxygen subdomain or sub-experience: the configurator, the B2B portal, the regional storefront, the editorial commerce hub. Wire them into the same admin, the same inventory pool, the same checkout.

This is the architecture Shopify itself documents as a supported deployment path under "Bring Your Own Stack" alongside Hydrogen + Oxygen, and it is the architecture that lets a Plus merchant capture the genuine headless wins on the experiences that need them without paying the full cost of ownership across the whole storefront.

Most agencies do not pitch this because it is less revenue for them than a full rebuild. The merchants who insist on it are the ones who end up with a faster site, a less stressed engineering function, and a budget left over for the work that actually moves the business.

If you are evaluating a Hydrogen pitch right now

A few questions to ask in any meeting before signing a contract.

What specifically does Hydrogen unlock that a polished Horizon-family OS 2.0 theme will not deliver, and is that thing worth the year-one cost on its own. If the answer is "speed," ask for a specific Lighthouse delta on the agency's most recent two builds, not a vendor benchmark. If the answer is "design flexibility," ask which design moves are blocked on OS 2.0 today and walk through them.

What is the ongoing engineering commitment after launch and is it being quoted as a separate annual line item. A Hydrogen pitch that ends at the build is not a complete pitch. The maintenance, version upgrades, security patches, and feature work compound for the life of the storefront.

What apps in the current stack will not survive the migration, and what is the replacement plan. App retrofitting routinely adds 20 to 50 percent to the build cost and is the most commonly underquoted line in headless proposals.

What is the rollback plan if the migration delivers below expectation. Not a real rollback (you cannot un-build a custom frontend), but a plan for reducing scope, returning to a hybrid composition pattern, or accepting that some workflows stay on Liquid permanently.

If the agency cannot answer those four cleanly, the proposal is not ready and the contract should not be signed.

Hydrogen is a good framework now. The 2026 release line is mature. Oxygen is free with your Shopify subscription and the deployment story is clean. None of that changes the fact that for most Plus merchants in 2026, the right move is not to go headless. It is to make a fast, modern, AI-ready OS 2.0 storefront the centre of the operation, and to use Hydrogen surgically only where the budget genuinely earns its keep.

If your store fits one of the three cases above and you want a second opinion on the proposal in front of you, send it over. We have read enough Hydrogen pitches at this point to tell within an hour whether the math is real or whether it is going to quietly eat the next eighteen months of your roadmap.