828 New Fintech Holding Shells, Zero Brands: The UK's AI Finance Incorporations Crisis

The Number That Doesn't Make Sense

1. That's how many new SIC 64.20 companies - the holding-company code that has become the default shell for aspiring UK AI fintech ventures - were incorporated in Q3 2026. Down 87.2% on the prior period. AIBD analysis of Companies House data.

Hold that thought, because here's the kicker: of all active SIC 64.20 companies on the register right now, 98.6% hold no Class 36 trademark whatsoever. No brand. No IP protection. No mark covering financial services, insurance or investment. Just a shell with a name and a registered office, frequently a serviced address in Shoreditch or Canary Wharf.

Charlie Munger called it seeing what everyone else sees but thinking what no one else thinks. What the incorporation data actually shows isn't an AI finance revolution. It's a holding-company graveyard.

Let's Do the Maths

Beauhurst counts more than 4,400 fintech companies currently active in the UK, but new fintech company incorporations had already contracted sharply, from a height of 397 new companies in 2019 to just 77 in 2025. The Q3 2026 SIC 64.20 crash is consistent with that secular contraction, but an 87.2% quarterly drop is a different order of magnitude. That is not a cooling market. That is a freeze.

Cross-reference the trademark side. AIBD's analysis of IPO filings shows only 430 Class 36 trademark applications in Q3 2026, down 82% on the prior period. Class 36 covers financial, insurance and real estate services, including investment services, fundraising services, valuation services, currency trading and pawnbrokerage. These are the mark categories any serious AI finance business would file the moment it had a brand worth protecting.

So: 828 companies incorporated. 430 Class 36 filings across the entire market. And 98.6% of the active SIC 64.20 universe holds nothing at all. The maths says most of these entities are not building a business. They are reserving optionality at £50 a pop.

The Regulatory Backdrop Is Getting Serious

The entity data is particularly stark against the regulatory environment now bearing down on anyone who actually wants to operate in AI finance.

The Treasury Select Committee concluded in January 2026 that the Bank of England, the FCA and the Treasury are exposing the public and the financial system to potentially serious harm due to their current positions on the use of AI in financial services, specifically their wait-and-see approach. That is parliamentarians, in print, saying the regulators are behind the curve.

More than 75% of UK financial services firms are now using AI, with the largest take-up among insurers and international banks, being used to automate administrative functions and to deliver core services such as processing insurance claims and credit assessments. The adoption is real. The governance isn't.

The FCA has been formally recommended to provide the financial services sector with greater clarity on the application of existing rules to AI use, and by the end of 2026, should publish practical guidance on consumer protection rules and accountability under the Senior Managers and Certification Regime for harm caused through AI. That guidance still hasn't landed as of today.

HM Treasury has failed to commit to bringing the major AI and cloud providers into the Critical Third Parties Regime before the end of 2026, despite criticism from the Committee that it is moving too slowly. Treasury Committee chair Dame Meg Hillier was direct: "The disruption which could be caused to our financial services system by an outage at a major provider could be extremely damaging. The powers offered by the Critical Third Parties Regime are sitting unused while we remain vulnerable. I simply cannot understand why this is taking so long."

What the Cambridge Data Actually Shows

Separate from the parliamentary theatre, Cambridge Judge Business School's Centre for Alternative Finance published its 2026 Global AI in Financial Services Report in late May. The numbers are striking.

Eighty-one per cent of surveyed financial services firms are adopting AI at some level, with 40% of industry respondents reporting advanced AI adoption at the 'Scaling' or 'Transforming' stages, more than double that of regulators. Fintechs are ahead of incumbents. Regulators are behind both.

Data privacy and protection (cited by 65% of AI vendors, 74% of industry and 80% of regulators), and model hallucinations and unreliable outputs (cited by 67% of AI vendors, 70% of surveyed industry firms and 70% of regulators) were rated as the top two risks by all stakeholder groups. The first time in years that regulators and vendors have broadly agreed on anything. It doesn't mean the risks are managed. It means everyone sees the cliff.

The Bank of England's most recent survey of AI in UK financial services found that 46% of respondent firms reported having only 'partial understanding' of the AI technologies they use, because of the adoption of third-party models, which account for a third of all use cases. Almost half of firms are deploying technology they don't fully understand. In financial services. Under the SM&CR. That is not a grey area.

The Trademark Gap Is a Valuation Problem

Here is why the 98.6% figure matters to investors, not just to lawyers.

A Class 36 trademark is not bureaucratic box-ticking. It is a legally defensible asset on the balance sheet. It separates a brand that can be licensed, acquired, and valued from a name that anyone can copy the moment your back is turned. When a potential investor searches for a company, their SIC code helps them understand what it does. If the code doesn't match the actual activity - for example, actively trading but still showing dormant status - it can create confusion and in some cases actively work against funding applications. Lenders and funding bodies use SIC codes to assess eligibility, and a mismatch can be enough to cause delays or complications.

No trademark in the services class that defines your business. No match between SIC code and actual activity. That is paying restaurant prices for a microwave meal, then wondering why no one will buy the franchise.

HM Treasury stated in April 2026 that UK fintech attracted over £2.6 billion in investment last year, second only to the United States. The investment flow is real. But capital concentrates in a handful of scaled businesses with genuine IP, genuine regulatory standing, and genuine brand moats. The 828 Q3 shells are not getting that capital. They are not even in the consideration set.

The FCA's Summer Deadline

Shortly after the Treasury Committee report, on 27 January 2026, the FCA announced a review into the long-term impact of AI on retail financial services, known as the Mills Review, seeking views across four main themes. That review is expected to deliver external findings and recommendations this summer.

Combine that with the guidance deadline on Consumer Duty application to AI the FCA is supposed to hit before year-end, and Q3 2026 is genuinely a pivot point for UK AI in finance. The regulatory framework is being written in real time.

For the 828 new shells with no brand, no IP and no regulatory standing, the window to get serious is closing. The Economic Crime and Corporate Transparency Act has introduced stronger requirements for all information on the Companies House register to be accurate, including SIC codes. Under new powers, Companies House can challenge, reject or remove information that appears incorrect, inconsistent or misleading. If a SIC code does not reflect a company's actual activity, filings may be rejected or delayed while discrepancies are investigated.

Nassim Taleb called it the silent risk: the one nobody prices because it hasn't happened yet. The UK AI finance holding company boom has been very loud. The bust is arriving quietly, one missed trademark application at a time.

Prediction: By Q4 2026, the FCA's Mills Review recommendations, combined with the first wave of Companies House SIC enforcement actions, will trigger a measurable spike in Class 36 trademark filings and a round of voluntary dissolutions among the SIC 64.20 shells. The serious AI finance businesses will be identifiable by their IP position. The rest will be Companies House statistics.