The market hiding inside the market
By Micky Irons, founder of Mickai.
There is a headline number that everyone in artificial intelligence repeats, and a quieter number underneath it that almost nobody prices correctly. The headline is the total addressable market for enterprise artificial intelligence software, around £122.6bn by 2030, compounding at roughly 37.6 per cent a year. It is a genuine figure and it is enormous. Yet it describes a field, not a position. Treating it as the prize is like valuing a vineyard by the rainfall over the whole county.
The position that matters sits beneath it. Strip out everything that can only run where the operator holds its own keys, where every output must be auditable, where data is forbidden by law or contract from crossing into a third party's cloud, and you arrive at a serviceable addressable market of about £40bn. That is the regulated, private-deployment-eligible slice. Inside it sits a tighter, faster-growing core, the served market for governed, auditable artificial intelligence, around £4.6bn today and expanding at about 45 per cent a year. That core is where the Mickai Sovereign Intelligence Operating System, the SIOS, competes, and it is the part of the field the public cloud cannot enter by architecture rather than by effort.
The distinction is not pedantry. A vendor that wins 1 per cent of a £122.6bn field of general assistants has won a crowded commodity contest against the best-funded laboratories on earth. A vendor that wins a serious share of a £4.6bn governed core, growing at 45 per cent, has won a market with no incumbent, because the incumbents are structurally disqualified from entering it. The smaller number is the better business.
Why a sovereign perimeter exists at all
The reason this core is uncontested is the same reason it is hard to see. Governed artificial intelligence is defined by what it must never do, which is to leak, to drift, or to produce a decision it cannot later explain on demand. The frontier clouds are extraordinary tools for open work, and Mickai treats the leading cloud AI providers as partners in exactly that space. The difficulty is not capability. It is the data path.
"If you are a multibillion-dollar company running on Anthropic or OpenAI, and your direct competitor of comparable scale sits on the same vendor stack, what stops them paying a vendor insider to leak your data, your tactics, your leads, your sales strategy? Inside a third-party cloud, there is no safeguard you can verify from the outside. The only answer is a sovereign system where you hold the keys, with no third-party cloud data path."
Micky Irons, founder and CEO, Mickai LTD.
Once the requirement becomes a system the operator can verify from the outside, the field collapses to those who can deliver it. Fifty specialised brains run fully offline on hardware the customer owns. Data never leaves the building. Every action is sealed under a post-quantum signature, the Open Audit Record, the OAR, that anyone can verify offline without trusting the vendor at all. The operator holds its own keys. There is no third-party cloud data path to insure against, because there is no path. That is the wall around the £4.6bn core, and it is built of architecture, not marketing.
Two buyers, walking from opposite directions
What makes this market unusual is that demand arrives from two directions at once, and both arrive at the same door. Call them the forced-off and the never-started.
The forced-off segment is rescue revenue already in motion. These are organisations that adopted cloud artificial intelligence, discovered the perimeter problem the hard way, and pulled back. A major electronics manufacturer banned public AI tools internally after engineers leaked source code into them. Major global banks and several National Health Service Trusts restricted the same tools through 2023. A European data-protection regulator fined a major AI provider 15 million euros. A national privacy regulator in Asia issued its own penalty. A Cisco study found that 27 per cent of organisations had banned generative artificial intelligence outright, that 63 per cent restrict what data can be entered, and that 61 per cent restrict which tools may be used at all. Each of those restrictions is a buyer who has already concluded that rented intelligence will not clear their compliance bar. They are not prospects to be educated. They are budget holders looking for the thing they could not previously buy.
The never-started segment is net-new, unclaimed spend. These are the teams that never typed a sensitive prompt into a cloud model because their counsel, their regulator, or their classification told them not to. Magic Circle litigation teams handle privileged material. NHS clinical units are bound by the Data Security and Protection Toolkit, the DSPT. Ministry of Defence-cleared programmes operate under Joint Service Publications 440 and 604. Financial Conduct Authority-regulated wealth managers carry obligations no shared tenant can satisfy. United States federal workloads require the Federal Risk and Authorization Management Program, FedRAMP, and Impact Level 5 and above. International Traffic in Arms Regulations and Export Administration Regulations-controlled aerospace work treats a single cross-border token as an export violation. This spend has been sitting in a holding pattern, waiting for an artificial intelligence platform that arrives compliant rather than asking forgiveness. It has never clicked accept, and it never will, until the system is sovereign.
The elegance of the thesis is that one architecture serves both. The forced-off buyer needs somewhere to land after leaving the cloud. The never-started buyer needs somewhere to begin that was never the cloud. Both need the same thing, which is brains they own, on hardware they own, producing artefacts they can prove. Mickai does not build two products for two segments. It builds one perimeter that both walk into.
Sizing the prize in firms, not just pounds
Abstract market figures invite scepticism, so it helps to count the actual organisations. In the United Kingdom alone there are roughly 50,000 regulated firms in the categories where governed artificial intelligence is mandatory rather than optional. The Financial Conduct Authority oversees about 42,000. The Solicitors Regulation Authority covers about 8,900 law firms. The National Health Service runs about 200 to 215 trusts, alongside 6,277 general practice surgeries. Layered on top, about 8,250 large UK enterprises form the natural beachhead, each with the budget, the obligation, and the data sensitivity to justify a capital purchase.
The wider context makes the slope clear. The United Kingdom's artificial intelligence sector reached £23.9bn in 2024, up 68 per cent in a year. The European Union Artificial Intelligence Act's high-risk obligations begin to bite from 2 December 2027, with fines reaching 35 million euros or 7 per cent of global turnover. Under the Financial Conduct Authority's Consumer Duty, in force since 2023, every consequential customer decision must be auditable and explainable. None of those rules can be satisfied by a system whose reasoning lives in someone else's multi-tenant storage. Each one is a demand-generation engine for the sovereign core.
What the operator actually buys
The eighteen enterprise studios are the concrete shape of that demand, mapped to the regulations that compel them. Nomos carries compliance and regulator reporting against FCA SYSC, the Digital Operational Resilience Act, and the Network and Information Security Directive. Astraea, the legal and contract-review studio, serves the SRA-regulated firms and Magic Circle litigation teams directly. Plutus handles finance, accounting, and financial planning and analysis under the Sarbanes-Oxley Act and the controls that follow it. Nemesis runs fraud and anti-money-laundering monitoring. Tyche covers underwriting, rating, and actuarial work under Solvency II for insurers. Panacea produces clinical documentation and electronic health record entries inside the NHS Data Security and Protection Toolkit and, in the United States, the Health Insurance Portability and Accountability Act. Aletheia delivers audit and continuous controls assurance, the live evidence trail that SR 11-7 model-risk reviewers and external auditors demand.
Around them sit the operational studios that make the same data usable without ever exposing it. Pythia drives business intelligence and analytics. Iris is the customer-service agent. Hermes runs procurement and source-to-pay. Demeter governs inventory and warehouse management. Prometheus handles demand forecasting and supply planning. Ergon covers human resources, Chiron training and learning, Hephaestus predictive maintenance, Triton after-sales and field service, Xenia customer relationship management, and Clio the meeting notes that never leave the room. Eighteen enterprise studios sit on top of thirty-eight base studios, fifty-six in total, all running on one operating system the operator owns outright.
The quality dividend nobody priced
There is a second reason the sovereign core grows faster than the field around it, and it is technical rather than legal. Owning the memory does not merely satisfy auditors. It produces better answers.
"When companies use the Mickai Sovereign Intelligence Operating System, the context-compression problem that plagues cloud LLMs is removed at the architectural level. Cloud systems hallucinate and drift off topic because shared multi-tenant storage forces aggressive context compression, summary-pass swaps, and lossy recall. Inside Mickai, the operator owns the memory. They expand it inside their own data centre or workstation, scale it on Poseidon rack-scale or local NVMe, and never compete with another tenant for context budget. The result is a measurable reduction in drift and hallucination."
Micky Irons, founder and CEO, Mickai LTD.
That is the part the total-addressable-market view misses entirely. The sovereign buyer is not accepting a compliance tax in exchange for a weaker tool. They are getting a tool that drifts less, recalls more, and seals what it does. The regulated perimeter, far from being a constraint, turns out to be where the better artificial intelligence lives. A field that prices governance as a cost has the equation upside down. Governance, delivered at the architectural level, is the feature.
The economics that close the deal
A sovereign system would be a curiosity if it were ruinously expensive, so the commercial model is built to invert the usual maths. Mickai is a capital purchase, not a subscription. Access for a fee, deployed free. The operator buys the SIOS, runs it on hardware it owns, and holds its own keys. Above roughly 50 million tokens a month on owned hardware, it runs 70 to 90 per cent cheaper than cloud application programming interfaces. Break-even commonly arrives inside eighteen months, and at high volume as fast as four to eight weeks. The ladder runs from Solo, at £4,500 to £6,500, through Team, Department, and Enterprise, to Sovereign deployments of £2m to £25m and above.
Run the arithmetic against the firm counts and the £4.6bn served market stops looking like a ceiling and starts looking like an early reading. Year-five global commercial revenue near £2.5bn, combined near £3.5bn, at about 59 per cent earnings before interest, taxes, depreciation, and amortisation, is not a claim on the whole field. It is a claim on a meaningful share of the one slice the field cannot reach. The patents matter here too, 101 filed United Kingdom patent applications and roughly 2,234 claims owned by Mickai LTD, named inventor Micky Irons. They are evidence that the perimeter was engineered deliberately, not improvised, but the position is what sells.
The prize, and why it stays uncontested
Most large markets are contested because anyone with capital can compete for them. The governed core is different. The barrier is not money or talent, both of which the frontier laboratories have in abundance. The barrier is the data path. A cloud provider cannot offer no third-party cloud data path without ceasing to be a cloud provider. The Open Audit Record cannot be retrofitted onto a system whose reasoning the operator cannot inspect. The fifty owned brains cannot be reduced to a managed service.
So the £40bn regulated serviceable market, and the £4.6bn governed core inside it growing at 45 per cent a year, sit waiting for a buyer who arrives from one of two directions. The forced-off leave the cloud with rescue budgets already approved. The never-started hold net-new spend they were never permitted to release. Both reach the same wall, and on the far side of it stands one operating system they can own, audit, and prove. That is the prize. It is large, it is compounding, and the only firms equipped to win it are the ones that built the perimeter rather than the platform.
Written by Micky Irons. Originally published at https://mickai.co.uk/articles/two-buyers-and-a-forty-billion-market. More from Micky Irons and Mickai at mickai.co.uk.
