Airwallex

The Numbers at a Glance

1 · Business Overview

A decade of licenses, compounding into a platform

The founding story is now fintech canon: four founders running a Melbourne coffee shop, bleeding margin on international transfers for imported beans, decided to build the rails themselves. Eleven years later Airwallex clears about 93% of transactions across its own network rather than SWIFT — the structural fact that separates it from every FX brokerage that resells someone else's correspondent banking.

Named customers include McLaren, Qantas, Canva and Shein. The revenue base splits roughly 60% cross-border payments and embedded API products and 40% business-account products, and the mix has been migrating toward the higher-margin end: by mid-2025, more than half of gross profit came from domestic payments and card issuing rather than the original cross-border FX spread.

2 · Product Suite & Moat

From FX arbitrage to an AI-native financial operating system

The moat thesis rests on four pillars, and only one of them is software:

• Regulatory surface area — 85+ licenses across four regions is a decade-long, capital-intensive, non-parallelizable asset. A competitor with better code still cannot legally move money in Japan, South Korea, the UAE and Brazil next quarter.
• Own-network settlement — routing ~93% of transactions internally rather than through SWIFT is what funds the take rate. Volume growth deepens internal liquidity, which lets Airwallex serve progressively larger enterprises.
• Multi-product attach — more than 90% of revenue now comes from customers using two or more products. That is the single most important disclosure in the Series H release: it converts a transactional FX relationship into a platform relationship with switching costs.
• Geographic diversification — Americas revenue grew 171% YoY and EMEA 116% YoY through 2025, outpacing the APAC home base. The company is no longer an Australia–China corridor story, which matters for both the multiple and the political risk.

3 · Financials & Unit Economics

$1.3 billion of revenue on a 0.45% take rate

Airwallex crossed $1 billion in annualized run-rate revenue in October 2025, up 90% year-over-year at the time. Five months later, in March 2026, the company reported $1.3 billion in annualized revenue, up 74% YoY. Third-party researcher Sacra estimates ARR reached $1.5 billion by May 2026, and Jack Zhang has said the company is tracking toward $2 billion by the end of 2026.

Volume compounded faster than revenue. Annualized transaction volume reached $287 billion in March 2026, up more than 120% year-over-year — which implies a prior-year base near $130 billion.

The take-rate math: $1.3B of annualized revenue against $287B of annualized volume implies a blended take rate of roughly 0.45%. That is materially thinner than dLocal's emerging-markets gross take rate and materially thicker than a pure interbank FX spread — it sits almost exactly where Stripe's net take rate lands. The important read-through is directional: revenue grew 74% while volume grew 120%, so the blended take rate compressed. Larger enterprise customers pay less per dollar moved. That is the normal cost of moving upmarket, and it is the number to watch in any eventual prospectus.

Two more derived figures worth holding. At 2,300 employees and $1.3B of annualized revenue, Airwallex runs at roughly $565,000 of revenue per employee — high-end infrastructure economics, not neobank economics. And the mix shift is real: gross profit growth accelerated to 78% YoY in the first half of 2025 from 40% the prior year, as cards and domestic payments displaced FX spread.

The profitability question is where sources diverge, and the divergence is the story. Zhang told the Australian Financial Review the company has been profitable since 2023. Airwallex's APAC general manager describes the business as cash-flow-positive with deep reserves. Sacra dates EBITDA profitability to Q4 2025. Zhang simultaneously says AI investment has made margins too volatile for public markets. All three can be true at once — the company is profitable, but not yet profitable on a line that would survive quarterly guidance.

4 · IPO Status & Timeline

Every listing signal except the listing

Airwallex has assembled the full pre-IPO apparatus in eighteen months. In April 2026 it appointed Elana Rubin, a Reserve Bank of Australia director, to chair a new governance board. In June 2026 it hired Pranav Sood as its first Chief Financial Officer, returning from Bain Capital's growth equity fund. Its APAC general manager said the company is "spending this year getting IPO ready."

And then, in the same week, the CEO explained why it will not list. Zhang told the AFR that Airwallex had pushed back plans to float in 2026 or 2027, because roughly 90% of capital is going into customer-facing agentic-finance product and the resulting swings in EBITDA and gross margin would create volatility public shareholders would punish.

Three details in that table matter more than the headline. First, the Series F included roughly $150 million of secondary share transfers — early liquidity that reduces the pressure any listing would otherwise relieve. Second, Addition led three consecutive rounds, marking its own position up 38% in six months; concentrated conviction and a conflicted mark are the same fact viewed from two angles. Third, the flat stretch from 2021 to 2022 is not a rendering artifact — Airwallex's mark barely moved for three years while the 2021 vintage repriced across the sector. The company grew into its multiple rather than defending it.

5 · Competitive Landscape

Growing twice as fast as the listed comp set

Airwallex is the only name in that set growing above 70%. Wise is the closest structural analogue and the most instructive comparison: $2.5 billion of FY2026 net revenue on $243.5 billion of cross-border volume, growing 19%, with 18.9 million active customers. Airwallex moves more volume than Wise — $287 billion annualized — on roughly half the revenue, because it charges a business take rate on larger tickets rather than a consumer spread on small ones.

Wise also just handed Airwallex a template and a warning at the same time. Wise listed on Nasdaq in May 2026, moving its primary listing off the LSE, and now trades at a market capitalization near $12.4 billion — barely above Airwallex's private mark, on nearly twice the revenue. That is either evidence that Airwallex's private mark is aggressive, or evidence that growth deserves the spread. Section 7 argues the latter, with a caveat.

The competitive dynamic that has changed most is with Stripe. Stripe attempted to acquire Airwallex for more than $1 billion in 2018; Zhang, then 34, declined — a decision he has described as a lifestyle call rather than a financial one, and one that split the founding team. In 2026 the two compete directly across platform payments, global accounts and now in-store terminals.

6 · The Agentic Bet

T:0 and Airi — the reason the IPO slipped

Distribution is being wired in parallel with the product. Airwallex signed Visa in July 2026 to build embedded finance for freight and shipping platforms, added Affirm pay-over-time at US checkout in August 2026, and runs BNPL through Tabby in the UAE and Saudi Arabia. Geographic buildout continued through the same window: a Type 2 Funds Transfer registration in Japan, the Paynuri acquisition in South Korea in January 2026, in-principle UAE licenses, a new Saudi entity, and a $1.1 billion EMEA commitment including roughly 100 senior engineering hires in the UK and the Netherlands.

7 · Valuation Framework

8.5× revenue — the cheapest growth-adjusted multiple in the set

Read that chart carefully, because the honest version is less flattering than the headline. Airwallex at ~8.5× revenue against Wise at ~5.0× looks like a modest premium for four times the growth rate — attractive. But dLocal, the fastest-growing listed comp, trades at roughly the same ~8.3× on gross profit, its only line comparable to Airwallex's net-style revenue. Against dLocal, Airwallex is priced in line, not cheap. The genuine dispersion is at the top: Stripe carries ~24.5× at half Airwallex's growth rate, and nothing in the operating data explains a spread that wide.

A caution on dLocal that cuts the other way: its revenue grew 56% in Q2 2026 while gross profit growth was guided to only 25–30% for the full year. Growth bought at the cost of margin is worth a lower multiple. Airwallex's own take rate is compressing for the same structural reason, which is precisely why the $2 billion year-end target matters more than the growth rate.

Fair-value triangulation:

• Bull case ($18–22B): The $2B ARR target lands, CFIUS closes without remedy, and T:0 or Airi shows attach in the platform base. Airwallex lists into a receptive window at 9–11× forward revenue — the multiple Wise commands plus a growth premium that is actually earned.
• Base case ($11–14B): Growth decelerates toward 50% as the enterprise mix compresses take rate. The mark holds and grinds higher through 2027 on revenue growth rather than multiple expansion. No listing before 2027.
• Bear case ($5–7B): A CFIUS remedy forces a US structural separation, or the AI investment cycle pushes margins negative into a risk-off tape. The multiple reverts toward Wise's ~5× on decelerating growth — which on $1.3B of revenue is roughly a 45% markdown from the Series H.

8 · Key Risks

What public-market diligence will price in

The China question is not a rumor and it is not resolved. In December 2025 investor Keith Rabois — a board member of rival US fintech Ramp — publicly called Airwallex a "Chinese backdoor into sensitive American data." In June 2026, Senator Tom Cotton wrote to Treasury Secretary Scott Bessent urging a CFIUS review of Tencent's and HongShan's minority stakes, following an earlier request for a Department of Justice review.

Airwallex's rebuttal is specific rather than rhetorical: Zhang says the two Chinese backers hold less than 20% combined and have no board seats, that US customer data is stored in the US and inaccessible to staff in China or Hong Kong, and that the data-security program has been independently verified beyond minimum requirements by third-party assessors. He has called the allegations "wild and totally unfounded conspiracy theories," and Square Peg co-founder Paul Bassat has suggested the scrutiny tracks Zhang's ethnicity rather than the cap table.

For an investor the merits are almost beside the point. What matters is that a live CFIUS referral is unresolvable on the company's own timetable, and that no US-listing underwriter will price into an open federal review. This is the binding constraint on the IPO, more than the margin volatility management cites.

9 · Pre-IPO Exposure Routes Today

Thin direct access, credible listed proxies

Airwallex is one of the harder large private fintechs to own. There is no ARK-style listed fund with a headline position, no dominant public strategic holder, and — unlike Stripe — no visible secondary premium to arbitrage.