The AI Cost Wix Won't Tell You About Isn't Compute. It's the Margin.

Wix.com told investors its AI products were competitive and that AI-related costs were trending down. Then the numbers came out. Quarter after quarter, the gap between Wix's revenue growth and its profitability widened into something the spreadsheet couldn't explain away.

The company's adjusted operating margin collapsed from 21% in early 2025 to single digits by the first quarter of 2026. Then the company flipped from a GAAP net profit of $57.7 million in Q2 2025 to a GAAP net loss of $76.4 million in Q2 2026. Revenue grew 15% each time. The profit disappeared somewhere between the headline and the bottom line.

The stock tracked the discovery. From a 52-week high near $191, Wix fell to $40 before recovering to around $88 today. A year ago the company was worth roughly $8 billion. Today it trades at about $3.7 billion.

A securities class action, filed in July and captioned Yappi v. Wix.com Ltd., alleges that management overstated the competitiveness of its AI products and understated the costs of building and promoting them. The class period runs from February 19, 2025 through May 12, 2026, with a September 22, 2026 deadline for investors to seek lead plaintiff status.

The lawsuit is the easy part of this story. The harder question is the one lawyers won't answer: is Wix's AI investment a bridge to higher margins, or a trap that permanently rewired its cost structure?

Start with the number that doesn't add up.

Wix reported Q2 2026 revenue of $563 million, up 15% from the prior year. That beat Wall Street estimates and grew at the pace investors signed up for. Bookings rose 12%. Annual recurring revenue hit $1.96 billion. On the surface, the growth engine is intact.

Now follow the same quarter into the income statement.

Non-GAAP gross margin fell 3 percentage points year-over-year to 67%. Sales and marketing expenses surged 67% to $173 million. Total operating expenses jumped 46% year-over-year to $423 million in Q1 and remained elevated through Q2. The GAAP operating margin for Q2 sat at minus 10.4% — a number that erases the "profitable growth story" Wix built its premium on.

This is not a one-quarter accounting blip. In Q1 2026, the company posted a GAAP operating loss of $69.7 million, down from $37.4 million of operating income the year before. Non-GAAP operating margin, the number management asks investors to focus on, collapsed from 21% to roughly 5%.

The filing calls the cost surge temporary. It has appeared for four consecutive quarters now.

The cost is not one line. It's three.

AI compute. Wix runs two major AI products: Wix Harmony, its AI-powered website builder, and Base44, a no-code application platform the company acquired in June 2025 for an initial $80 million. Both are inference-heavy. Both are expensive to run.

Base44's gross margin was near break-even at the start of 2026. Management projects it will reach approximately 60% by the second half of the year, helped by the launch of Base1, a proprietary large language model designed to reduce inference costs. Management said Base1 is already outperforming frontier models for this use case at "dramatically less" cost and will improve consolidated gross margin by about 2 percentage points in the second half versus the first half.

That sounds like a turnaround until you read the next sentence.

Marketing reinvestment. Management confirmed that the entirety of the gross margin savings from Base1 will be reinvested into Base44 sales and marketing rather than flowing to the bottom line. The company raised its customer acquisition cost payback period target, signaling a willingness to extend the payback clock to capture market share in the AI app-creation market.

This is the margin trap: the cost savings that were supposed to restore profitability are pre-committed to funding more growth. The savings never reach the operating margin line they were supposed to fix.

The professional developer exodus. Wix has two customer segments: self-creators who build their own sites, and partners — agencies and professional developers — who build sites for clients. The partner business was a high-margin, high-retention growth engine.

During the May 13, 2026 earnings call, management acknowledged that the Wix Harmony platform had "holes" and "missing capabilities". Professional developer customers were defecting to competing AI tools. Wix admitted it had fallen behind "the workflow and the needs of" professional developers, with delays in product updates and innovation pushing them elsewhere.

Partners revenue grew 17% in Q2, but the sequential momentum had shifted. The "softer start to the year" management flagged in Q1 was not a billing-cycle anomaly — it was customers voting with their feet.

The lawsuit alleges that Wix had told investors Harmony's early performance was "better than expected" with "improved conversion and monetization". The earnings call admissions — holes, missing capabilities, defection to competitors — read as the correction.

Now step back and look at the capital structure, because this is where the shareholder cost gets real.

Wix has $263 million in cash and $3.59 billion in total debt. Total equity sits at minus $1.74 billion. The negative equity isn't an operating problem — it's the afterimage of a $1.6 billion modified tender offer completed in April 2026, funded by a combination of cash reserves and a $260 million private debt placement. The company bought back roughly 17.5 million shares at $92 each, reducing the share count by about 30%.

That tender offer is the most important number in this file.

Wix repurchased 30% of its shares at $92 per share — a price roughly three times what the stock trades at today. The buyback was funded by debt. The debt carries interest. The company is now reporting GAAP losses while servicing $3.6 billion in obligations it took on to return capital at a valuation that no longer exists.

The balance sheet tells a story the income statement is still catching up to. Cash flow from operations is $421 million trailing twelve months, and free cash flow margin sits at 24.5%. Wix generates real cash. But the debt load has tripled from roughly $1.9 billion at the start of 2024 to $3.6 billion today, and net debt swung from a net cash position of roughly $430 million in late 2024 to $1.02 billion in the most recent quarter.

The shareholder invoice here isn't legal damages from a class action — those are speculative and small relative to the operating reality. The invoice is the gap between $92 and $88 multiplied across 17.5 million shares, layered atop a debt service burden that didn't exist two years ago, while the company's own products cannibalize each other and eat margin.

Here is the benign case. Management says Base44 ARR grew from near zero to roughly $150 million in about six months. New user cohort bookings are up 46% to 50% year-over-year. Base1 is cutting inference costs. The core website business still grows in the mid-teens. Revenue is rising, cash flow is positive, and the company has simply front-loaded investment in a new growth cycle. Software companies burn through an AI investment phase before reaping scale — Amazon, Google, Microsoft. The argument is that Wix will too.

Here is the stress test for that case.

First, the reinvestment commitment. If gross margin savings flow to marketing, not to operating income, the path back to profitability depends on Base44 generating enough revenue to justify the marketing spend it's eating. That depends on renewal rates. Management acknowledged Base44 is in "the early stages of its customer lifecycle" and that renewal trends are improving monthly but "are not yet fully proven". Unproven renewals are the difference between growth and churn.

Second, the partner business. Wix's partners were the stickiest, highest-value customers. Professional developers leaving for competing AI tools isn't just a Q1 timing issue — it's a product-positioning question. If Harmony has genuine capability gaps, and competitors are filling them, then the core business that funded the AI bet is leaking revenue.

Third, the capital structure. The buyback at $92 was rational when the stock was near $180. At $88, it was already a mistake. At $40 — where the stock briefly traded — it was a catastrophic misallocation of capital that locked the company into debt service during a margin contraction. The company can't undo the buyback, and the debt doesn't disappear because the stock recovers.

Fourth, the competitive moat. Wix competes with frontier model providers — OpenAI, Anthropic, Google — who are building app-creation features directly into their platforms. A proprietary model like Base1 helps, but the company described its relationship with competitors like Lovable as "more collaborative than competitive in educating the new market". In a market being educated by the people who also build the infrastructure, being "collaborative" is a polite way of saying the moat is shallow.

The class action will sort out whether management's earlier statements about AI cost trends and Harmony's competitiveness were materially misleading. That is an evidentiary question for a court, and the lawsuit sits at Level Four on the evidence ladder: filed allegations, not adjudicated findings.

The investment question is separate and more urgent. Wix is growing revenue while spending itself into negative GAAP margins, and the mechanism that was supposed to restore profitability — lower AI inference costs — is contractually diverted into more marketing. The company is also carrying $3.6 billion in debt, roughly half of it accumulated to fund a share repurchase at a price point the market has since rejected.

For an investor deciding whether Wix belongs on a watch list or in a portfolio, the decision point is clear. The next two earnings reports will show whether Base44's gross margin actually reaches the projected 60%, whether Base1 delivers the promised 2 percentage-point improvement to consolidated margins, and whether that improvement stays in marketing or finally flows to operating income. They will also show whether partner revenue momentum recovers or whether the professional developer exodus accelerates.

The number is not proof. It is the door. The question now is what Wix walks through.