Xero Is Buying Payments Before Its AI Has Left the Lab

Xero has spent months telling investors it's an AI-native financial platform. In July it unveiled JAX, an agentic assistant that automates bookkeeping and cash-flow management, and celebrated 5 million customers. In May it launched XeroForce, a no-code builder that lets small businesses and accountants create custom AI workflows. Then, two weeks after the JAX announcement, it named Jonathan Meltzer — a former QuickBooks product and marketing leader — as Managing Director, US.

The headline framing makes it sound like Meltzer is there to execute the AI strategy. He isn't. The strategy he's inheriting is a payments play, and the payments play is burning margins while the AI plays are still in alpha.

Here's the number that matters. Xero's FY26 results, released in May, showed 31% revenue growth to NZD $2.8 billion. Impressive on the surface. But the growth came in two very different flavors. Subscription revenue — the recurring accounting-platform business Xero has been known for — grew 20% to $2.4 billion. Other operating revenue — almost entirely from Melio, the US bill-pay platform Xero acquired in June 2025 for $2.5 billion — grew 202% to $375 million. Including Melio, US revenue jumped 240%. Excluding it, organic US growth accelerated to 30%, which is good but not the headline number.

Then there's what happened to profitability. Net profit fell 27% to $167 million. Gross margins compressed 5.1 percentage points to 83.9%, entirely because Melio runs at a lower-margin payments economics. The pro forma Rule of 40 (revenue growth plus free-cash-flow margin, a proxy for whether a software business is growing fast enough to justify its cost base) dropped to 36%. Management expects Melio to reach EBITDA breakeven no sooner than the second half of FY28 — more than a year and a half away.

The stock fell 8.6% on the results. Growth was strong. Margins weren't.

This is the contradiction the AI announcements don't resolve. Xero is positioning itself as the company that will replace accountants with agents. But it just spent $2.5 billion on a low-margin payments engine that will drag down profitability for two more years, and its most ambitious AI product — XeroForce — is an invite-only alpha with a general release planned sometime later in 2026. Over 2.6 million customers used AI features in FY26, according to management, but the company admitted AI monetization will have a "small immediate impact" on FY27 guidance. It will figure out how to charge for these things later.

So what is Meltzer actually there to do?

His background is telling. Before Xero, he ran product management and business operations at LegalZoom and held leadership roles at Intuit's QuickBooks division. He isn't an AI person. He's a small-business product person who knows the US bookkeeping channel. His job description is about delivering "integrated accounting and payments solutions" alongside Matan Bar, CEO of Xero's US and Melio business. He reports to the newly appointed Chief Business Officer, whose remit covers sales, marketing, business development, and customer experience.

The hire signals that Xero's US strategy is still about distribution, not about waiting for the AI to land. Xero is spending up to NZD $55 million in incremental US brand investment for FY27. It's pushing the Melio integration, which went live as online bill payments in March 2026. It's building a North America product and technology hub in Vancouver. The company gave itself guidance of NZD $3.62–$3.73 billion in operating revenue for FY27 — 25–30% growth — and wants revenue to more than double FY25 levels by FY28.

None of that depends on XeroForce being good.

And that raises the question the AI announcements are designed to distract from. Is Xero building something that compounds, or is it stacking cost centers and hoping the stack eventually makes sense?

Doubling input and getting more than double output — that's the test. Xero's AI story is the kind of thing that could compound if the agent builder actually catches on with the partner network. Accountants who serve hundreds of clients could deploy a single workflow across their entire practice. That's a superlinear channel. But it's also something you have to earn by making the product good. Right now, XeroForce can automate month-end close, document requests, and receipt follow-ups in a controlled alpha. That's promising in concept. It's unproven at scale.

Meanwhile, the payments business is a linear cost. Every additional Melio customer adds revenue, but at a margin structure that drags the combined company down. Xero is counting on cross-sell — getting existing accounting customers to use Melio for bill pay — to eventually improve the blended economics. That's a reasonable theory. It's also a theory that requires Meltzer and his team to execute something harder than selling an AI demo: convincing US small businesses to move their payments rails to a platform they don't yet trust.

Most software companies that bolt on payments do it to improve retention, not to build a payments company. Stripe started as a payments company and built outward. Xero is doing the reverse. It's buying the payment rail and hoping the AI makes the whole stack sticky. The order of operations matters. Building intelligence on top of a payments problem is harder than building payments on top of intelligence, because payments is where the margin lives — and Xero just bought its way into that problem at a premium.

I suspect Xero's AI work is real. JAX can already answer financial questions against a company's ledger. Smart bank reconciliation learns coding patterns. The Anthropic partnership for Claude-powered financial intelligence is in delivery. But these are features, not moats. QuickBooks has its own AI automation — bank transactions are already categorized and matched in queues ready to post, with books arriving "half done" for US accountants. The gap between Xero's AI and Intuit's isn't yet visible enough to drive migration on its own.

What to watch:

The next quarter's gross margin trajectory will tell you whether Melio is starting to earn its keep or whether the integration is a drag that's getting worse before it gets better. Management's target of Melio breakeven by late FY28 is specific enough to hold them to. If the timeline slides, the margin story changes.

Second, watch what happens to XeroForce after general release. An invite-only alpha is a product that doesn't exist yet. When it opens, look at whether accountants actually build workflows with it, or whether it sits as another feature they tried once and forgot. The compounding channel only works if adoption spreads through the partner network.

Third, the US organic growth rate — 30% in FY26, up from 25% and 13% in prior years — is the cleanest signal of whether Xero's core accounting product is winning market share without Melio. If that number slows while brand spend increases, the distribution problem is harder than Meltzer's background suggests he can solve.

Xero is not a bad company. It's growing fast, has a strong balance sheet, and its AI work is further along than most enterprise software players. But the way to build a platform is not to buy a payments company while your intelligence layer is in alpha. It's to make the intelligence layer worth paying for first, then use that to pull money through.

The test is simple enough to run every quarter: does the AI make the payments business more valuable, or does the payments business just make the revenue number bigger?