UK's FCA Draws the Line at Algorithms: If Your App Picks the Trade Route, Get a License Before October 2027

I've read the FCA's proposed crypto guidance a few times now, and the line I can't shake isn't the dramatic one. It's the administrative one. Technical access sits on one side. Substantive transaction involvement sits on the other. And that's the whole ballgame.

Because most of the crypto app layer has spent five years building a business on the claim that it lives entirely on the first side of that line. Let's apply the standard the industry set for itself and see how many of them still stand there.

Where the Line Actually Sits

Here's the mechanic, and it's worth being precise about it because the details are where firms get caught. The FCA's proposal treats holding keys, running a node, or giving someone a connection to a chain as technical access. Infrastructure. Plumbing. If that's all you do, the regulator's position is that you probably aren't conducting a regulated activity in your own right.

Then there's the other category. Substantive transaction involvement means you're doing the deal, or meaningfully shaping it. And the trigger the FCA has zeroed in on is the moment your software, rather than the user, chooses the trade route.

That's it. That's the hinge. Not custody. Not whether you hold client money. Whether your code decides where the order goes.

A wallet where the user picks the venue is plumbing. A wallet with a shiny swap button that quietly evaluates three DEX aggregators, two market makers and a last-look liquidity provider, then routes your order to whichever comes out best, is making an execution decision on your behalf. And execution decisions are the oldest regulated activity in the book. UK markets have been arguing about best execution since before crypto existed. The FCA didn't invent a new principle here. It applied the precedent it already had.

The application window for the UK's regime opens in the run up to October 2027, when the authorisation gateway goes live. That sounds like a generous runway. It isn't. Authorisations of this complexity get done in months, not weeks, and every firm in the UK will be queuing at the same desk with the same understaffed supervisor. The teams that file in September 2027 will be the ones complaining about the FCA being slow. The teams that started in 2025 will be the ones with licenses.

Now, the obvious objection. "We're non-custodial, the user signs everything, we just provide software." Fine. Show me the audit. Show me the log that proves the user, and not your routing engine, picked the venue. Because that log is what the enforcement case will turn on, and I'd bet good money that a large share of today's apps don't keep it, don't keep it in a form that survives scrutiny, and can't reconstruct last January's routing decisions at all.

The marketing says decentralized. The multisig says otherwise.

Best Execution Doesn't Care About Your Branding

Pull the camera back, because this is bigger than one jurisdiction's paperwork.

Once you accept that routing is a regulated decision, a whole set of obligations follows it in through the door. Best execution is the big one. If your app is choosing venues on a user's behalf, you'd better be able to demonstrate you were choosing in the user's interest. And here's where the incentives get ugly, because a lot of routing in this industry isn't neutral at all. It's paid for.

Order flow rebates. Private market maker arrangements. Preferred aggregator integrations where the commercial deal came before the technical evaluation. None of that's automatically wrong, but every one of those arrangements is a conflict of interest, and conflicts of interest are exactly what a conduct regulator exists to interrogate. Ask yourself the uncomfortable question. When your wallet routed that swap last week, did it go to the venue with the best price, or the venue with the best revenue share?

Who wins under this framing? Firms that already run regulated brokerage infrastructure, because they've carried a best execution obligation for years and built the surveillance, the record-keeping and the conflicts register that goes with it. Boring firms. Compliance-heavy firms. Firms that crypto Twitter has spent a decade calling dinosaurs.

Who loses? The app layer that built a valuation on the fiction that routing is a neutral technical act. If you raised a Series B off the premise that your swap interface isn't a financial service, this guidance is a direct hit on your regulatory capital position, your legal budget, and quite possibly your business model. Token classification fights are loud, but this quiet routing question touches far more companies and far more users.

And there's a UK-specific reason it matters beyond the UK. Functional perimeters travel. If the FCA successfully defines the boundary by what the software does rather than what the company calls itself, other regulators get a workable template. That's the precedent nobody on the app side wants to see written down.

What I'd Actually Do With This

Skepticism isn't pessimism. It's due diligence, and the due diligence here's simple enough to start this week.

Open your product. Find the screen where a swap happens. Ask one question. Who picked the venue, the user or the software? If the answer is the software, even partly, even by default, you're in the substantive bucket, and the burden of proof sits with the team, not the community. Go get the legal opinion on your routing logic, not on your token, because the routing logic is the part that's now on the clock.

Then build the log. Every route, every venue, every reason, retained and queryable, because that's the artefact that separates a firm from a defendant. If your engineers can't explain the routing decision tree in plain English on a whiteboard, your lawyers can't defend it either.

And if the honest answer is that your app has never been neutral about where orders go, then stop saying it's. The FCA has already read the code. The only question left is what you tell them about it before October 2027.