Tax Wasn't the Risk You Bought

Around 30,000 Britons have left the UAE since the Iran war kicked off in late February. Most of them can't go home. The FT's calling it Regrexit. But the label is wrong. They're not regretting the move, they just can't afford to undo it. So they're parked somewhere else — Switzerland, Portugal, Spain — watching two clocks tick down at once and hoping one of them stops before the other does.

Why most of them can't go home

The UAE has two residency certificates. The one HMRC and most double-tax treaties actually accept needs 183 days inside the country. There's a 90-day version too, but it gets sold a lot more than it works — EmaraTax cross-checks treaty rules at the application stage now, and if anyone pitched you the short route as good enough, it isn't.

The UAE tax year runs January to December. Anyone who left after 28 February has roughly ten months left in the year to clock six of them back in country. The arithmetic gets tight fast. By July, for a family with school-age kids, you're either coming back and disrupting school terms a second time, or accepting that this year's certificate isn't going to survive.

Coming back to the UK doesn't help either. The UK killed the remittance basis on 6 April 2025. The new four-year FIG regime is for people who haven't been UK tax resident in the prior ten years, which rules out almost everyone who emigrated to Dubai recently. Inheritance tax went residence-based the same day. Worldwide assets in scope after ten years. Returning Britons walk straight back into the regime they left to escape.

So they're stuck somewhere else. Beauchamp's reporting a 10% jump in inquiries from Gulf-based UK nationals, and most of them aren't asking about return moves. Switzerland, Portugal, Spain. Parking lots. The clearest signal is what families are doing with their kids' schools. They've enrolled them in UK private schools mid-year. You don't do that if you're planning to be back in Dubai by September.

The underlying flow was already running before any of this. Henley's 2025 migration data put UK net HNWI outflow at 16,500 last year, the steepest single-year figure on record for any country it tracks. About £66bn in associated wealth went with them. The UAE took around 9,800 on net. The UK tax reset drove the departures and the UAE's zero-rate model pulled the destination. The Iran war is the stress test those drivers were never built for.

Dominic Volek at Henley pushed back on the Regrexit framing in CNBC. Internationally mobile families have options across the Americas, Europe, the Middle East, and Asia, he said. They rotate strategically rather than reacting to events. He's right, but only about a specific kind of family — the kind whose structure was never single-jurisdiction to start with.

Lakshmi Mittal moved to Dubai in November 2025, ahead of the Reeves Autumn Budget and four months before any missiles fell. He kept his tax residence in Switzerland, primary residence in Dubai, UK presence wound down. Three jurisdictions, none of them carrying the full structural weight. When the UAE day-count clock got tight in February, that structure didn't break, because it was never single-point. Nick Storonsky, founder of Revolut, changed his address at Companies House in October 2024. Petra Ecclestone. Michael Platt. John Fredriksen. The press list cycles through the same handful of names for a reason. Tax residence one place, primary residence another, operating presence somewhere third, no fallback that depends on the country you just left keeping its door open.

What the Britons sitting in Lisbon hotels haven't priced is that concentration risk doesn't change shape when the asset is a jurisdiction. It looks defensible right up to the moment a second variable moves.

I've been on the wrong side of one variable moving. In 2014 I was four years into a public-private partnership in Crimea. Airport, seaport, roads, real estate, agriculture. Over $1bn in total commitments, with a $200m term sheet from a Chinese bank in hand. Then Russia annexed Crimea in February. Government counterparties vanished overnight. Sanctions made everything untouchable. The structure that had looked solid for four years went to zero in a single day, and I spent the rest of that year paying back personal debts to the people who'd backed me.

That's what one variable moving looks like when you haven't priced for it. The whole setup assumed Crimea would stay Crimea. It didn't, and there was nothing to do about it. Same thing happens any time you build a structure that depends on the conditions staying the way they were. If those conditions move, the structure stops working, and usually there's no time to fix it before the cost shows up.

Tax wasn't the risk most of these Dubai families bought. They weren't stupid. The UK was making the old non-dom regime untenable, the UAE was offering a clean alternative, and on the headline numbers the math worked. What they were actually buying was a structure that worked only as long as Dubai stayed quiet, their certificate stayed clean, and the UK kept its old non-dom door open as a fallback. All three held until 6 April 2025. The UK fallback closed first. The Dubai-quiet condition broke ten months later, on 28 February 2026. Two of the three conditions are gone now for anyone trying to assemble this from cold in 2026. I wrote about the structural side of this in January's Signal, Jurisdictions Are Competing Like Products. What 2026 has done is put a price on the variables nobody bothered to model in the first place.

The UAE's response is competitive, not structural

The UAE Federal Tax Authority is reportedly thinking about case-by-case force-majeure treatment for British nationals who can't make 183 days this year. The FT broke it in late April. There's no circular yet, no ministerial decision, nothing actually from the FTA — just lawyers telling reporters what they think might be coming.

Case-by-case is a worse signal than it sounds. A rule is something you can plan against. Discretion means someone at the FTA looks at your file and decides yes or no, exactly when you most need predictability. The 90-day certificate already exists and most foreign tax authorities won't accept it for treaty purposes. A case-by-case carve-out doesn't change that. It just stacks one layer of discretion on top of a route that didn't work properly to begin with.

None of the other low-tax jurisdictions are signalling the same thing. Bahrain's quiet, so is Qatar, and Singapore hasn't even acknowledged the conversation. Take that for what it is. A competitive response to the visible British outflow, not structural reform of the residency framework. So if a wealth manager tells you Dubai is making the rules friendlier, what they're really saying is the FTA might be open to one-off discretion for someone like you. That's not the same as the certificate route fundamentally changing. Whether case-by-case actually counts depends entirely on how it gets applied in practice, and there isn't enough of a track record yet to know.

What else on the Radar

Family office holdings data is finally public, and the survey-vs-holdings gap is real. Knight Frank's April survey said family offices are growing real estate. Addepar and CNBC's new tracker, covering $1.4tn in aggregated FO holdings from $200m to $10bn+ AUM, says they're shrinking it. Public equities are growing fastest. So when the next adviser pitch tells you "family offices are doing X," ask one question: surveyed or held? The held position is the ground truth. Read more →

SpaceX is reportedly filing its S-1 this week. Two of the largest private listings in history may print inside ninety days. SpaceX's S-1 could land as early as 20 May for a Friday 12 June IPO at over $1.5tn. OpenAI is reportedly lining up a confidential filing for September at around $1tn. If you've got exposure to either through a fund or fund-of-funds, pull the LPA section on IPO-listed positions now. Some funds distribute shares in-kind at lock-up expiry, others sell on-platform. Better to know which before the window opens. Read more →

The supply problem inside late-stage AI tenders. OpenAI's October $10.3bn tender cleared only ~$6.6bn, Anthropic's April tender came in below the upper bound, and SpaceX saw similar dynamics in December. If a wealth manager pitches access to OpenAI, Anthropic, or SpaceX secondaries this year, what matters is what employees turned down at the last tender, not the headline price. A discount to that benchmark means the seller knows something. A premium means an intermediary's taking margin. Two questions before signing SPV docs: what did the last tender clear at, and what's the seller's relationship to the company. Read more →

Wealth-management AI middleware is consolidating. Addepar rolled out Addison at its annual conference this week. AI agents for portfolio data, document analysis, look-through into private markets. Anthropic's Claude CoWork wealth plug-ins are in active deployment at advisory firms. The AI layer for wealth tracking and reporting is consolidating around two or three primary stacks. If you're building a lean family-office function now, you have more choice than in 2024 and less than there will be in 2028. Worth a half-day comparing Addepar, Aleta, and Asseta against what you actually need to track, before a private bank picks the default for you. Read more →

Goldman BDC came in at exactly 4.999%, one basis point under the cap. Goldman Private Credit Corp reported Q1 redemption requests at 4.999% of outstanding shares, one basis point under the 5% gating threshold. BCRED hit 7.9% (Blackstone covered with a $400m employee top-up). Blue Owl's OBDC II saw 21.9% against the same cap. The difference between "we paid everyone" and "we paid out at the cap" is one basis point in places. For anyone holding non-traded BDCs, read the redemption-request percentage next to the pay-out percentage in the next quarterly letter. If they don't match, the published NAV and the implied market price aren't the same number. Read more →

Coller's private wealth secondaries platform is expanding into 2026. Coller's HNW arm has raised over $4bn since launching in 2023. The firm agreed in January to be acquired by EQT for up to $3.7bn. Combined entity will hit ~$50bn in secondaries AUM. The broader market hit $226bn in 2025, up 41%. The path from "secondaries was institutional-only" to "secondaries is accessible at $200k–$1m via interval funds and SICAVs" has compressed into about thirty months. Real diversification access for HNW allocators, but the product structure (interval fund vs SICAV vs RIC) determines the actual liquidity. That often looks tighter than the marketing copy implies. Read the prospectus pages on redemption mechanics before the marketing deck. Read more →

If you're still building the structure

If your wealth architecture runs through Dubai with no second jurisdiction underneath it, the question worth asking right now is whether anyone has actually stress-tested the setup. Most of the time the answer is no. It looked clean on the spreadsheet, and nobody ran the scenario where Dubai stopped being quiet for twelve straight months.

Two things worth watching from here. First, whether the UAE FTA actually publishes a formal force-majeure rule or lets case-by-case sit as unwritten norm. Rules and discretion behave differently when you need them in a hurry. Second, whether the UK exit-tax debate moves past the 20% proposal stage. If it does, the cost of structural mobility for anyone leaving the UK in future changes materially, and so does every wealth-architecture conversation happening this year.

If the conversation you're having with your wealth manager this quarter is purely about tax efficiency, the conversation is incomplete. April's Signal, Tax Certainty Arrived. Wealth Architecture Is Still Waiting., covered one half of this: the certainty arriving on the UK side. This is the other half. The mobility position depends on jurisdictions other than the one you just left.

New on the Site

Last Thursday's article looked at why most post-exit founders underestimate how expensive the sheer volume of decisions becomes, separate from whether any individual decision is good or bad. Jurisdiction choices are some of the heaviest decisions you make in that window. Making them on one variable is the kind of mistake that doesn't show up for months.

Read it: Decision Fatigue Costs More Than Bad Decisions