The frontier and production programmes in this manifesto are costed separately in their own papers. This note brings them together into a single budget for the first five years, sets out where the money will come from and distinguishes four kinds of money: existing spending that will be redirected, loans and equity through the public investment banks, contributions from the defence budget and genuinely additional spending. It covers the British Frontier AI programme, Power from Orbit and the missions in Building British Industry.
The central proposition is that a large part of the cost can be met from money the state already spends, because support for business, innovation, computing and defence technology is spread across many schemes with no common purpose. That proposition is a planning assumption rather than a settled fact. It will be tested line by line by the review of business support published early in the Parliament, which will name each reallocation before the first block begins.
1Principles
Research is protected. Funding for basic research will not be cut to pay for these programmes. Broad scientific discovery is the foundation on which every programme draws, and its returns cannot be predicted in advance.
Redirect before adding. Money already spent on business support, sector programmes and existing artificial intelligence and computing commitments will be concentrated on the programmes before any new money is sought.
Invest where there is a return. Where the state takes a stake that is expected to earn a return, it will invest through the British Business Bank, the National Wealth Fund or the Sovereign AI Unit, and the expected losses on that investment will be budgeted as a cost.
Defence pays for defence capability. The defence budget will fund elements that strengthen Britain's defence industrial base and military capability. It will not be used to relabel civil spending.
One budget, released in blocks. Each programme is funded in five-year blocks with published gates, so that money is released only as results are achieved.
2What the programmes cost
All figures are public costs in the prices used by each paper, and all programmes are assumed to begin in year one, the first full year after taking office. Each mission figure is the full cost of its measures, including the equity and loans it contains. Private investment, which the artificial intelligence and space papers expect to be substantial, is excluded, as are repeat defence orders for autonomous systems, which are funded from existing defence equipment budgets and counted in neither the costs nor the sources.
| Programme | Years 1 to 5 | Years 6 to 10 | Years 11 to 15 | Basis |
|---|---|---|---|---|
| British Frontier AI | £30bn (£6bn a year) | £25bn to £40bn | Not yet costed | Central scenario in the AI paper. Later years from its ten-year estimate of £55bn to £70bn, including hardware refresh |
| Power from Orbit, all branches | £2.1bn | £7.5bn | £26.2bn | Cost model in the space paper, 2026 prices with Treasury optimism bias, including the defence and people-in-space branches |
| Advanced manufacturing | About £4.1bn | About £4.1bn | Not yet costed | Half of the ten-year estimate of about £8.25bn, of which around £1.75bn is equity and loans |
| Advanced materials | About £1.75bn | About £1.75bn | Not yet costed | Half of the ten-year estimate of about £3.5bn, of which around £1bn is equity and loans |
| Life sciences and biomanufacturing | About £3.75bn | About £3.75bn | Not yet costed | Half of the ten-year estimate of about £7.5bn, of which around £2bn is equity and loans |
| Autonomy and robotics | About £1.65bn | About £1.65bn | Not yet costed | Half of the ten-year estimate of about £3.3bn, of which around £750m is equity |
| Conditions for production | About £0.2bn | About £0.2bn | Not yet costed | Half of a gross ten-year cost of about £375m. Around £100m is expected back in fees, leaving about £275m net |
| Expected loss on equity and loans | About £1.4bn | About £0.7bn | Not yet costed | A quarter of the equity and loans in each block, set out in section 4 |
| Total | About £45bn (about £9bn a year) | About £45bn to £60bn | At least £26bn plus the AI programme and industry missions |
The budget makes a deliberate choice. Around £30 billion of the £45 billion in the first block goes to sovereign artificial intelligence, because the AI paper judges an independent ability to train and operate frontier models to be a strategic necessity and because the other missions depend on it. If that judgement were revised, the AI paper's sovereign floor scenario, with public spending of about £18.5 billion to £19.5 billion over five years, would reduce the first block by around £11 billion while keeping a sovereign capability below the frontier. The space programme is small at first and rises sharply in its third block, when the first full-scale station is built. Advanced manufacturing and life sciences carry the largest industry budgets, the first because it underpins every other mission and the second because it serves the largest market.
3Where the money comes from
Most existing commitments run only to 2029-30. The sources below therefore assume that spending in each line continues at broadly its current annual level into the next spending period, and that it can then be concentrated on the programmes. Each figure is an average a year for years one to five.
| Kind of money | Source | A year, years 1 to 5 | What it pays for |
|---|---|---|---|
| Existing spending, redirected | Business innovation and sector support: Innovate UK grants of £1.2bn in 2024-25, over £3bn for advanced manufacturing over four years, up to £520m for life sciences manufacturing, up to £600m for the Health Data Research Service and other sector programmes | £1.5bn to £2.5bn | Industry missions and sovereign research contracts |
| Existing spending, redirected | Existing computing commitments: over £2bn for computing in the current spending period, which includes up to £750m for the Edinburgh supercomputer, and the £500m Sovereign AI Unit | £0.5bn to £0.75bn | The sovereign compute operator |
| Loans and equity | Capacity allocated by the British Business Bank, the National Wealth Fund and the Sovereign AI Unit (section 4) | About £1.15bn | State equity in the AI laboratory and the equity and loans in the missions |
| Defence contributions | The defence budget, drawing on the commitment of a tenth of equipment procurement to novel technologies and the NATO allowance for defence-related spending on innovation and the industrial base | £1.5bn to £2.5bn | A share of sovereign computing, the AI laboratory's defence subsidiary and the space defence branch |
| Total identified | About £4.6bn to £6.9bn |
Against an average cost of about £9 billion a year, these sources leave about £2 billion to £4.5 billion a year of genuinely additional spending in the first block. Most of it is capital investment in computing hardware and facilities for the sovereign compute operator. Under the current fiscal rules, capital investment can be financed by borrowing without breaching the rule on day-to-day spending, but it adds in full to public sector net financial liabilities, because physical assets such as data centres and accelerators are not netted off against debt in that measure. Equity stakes are treated differently, because financial assets are netted off, although expected losses on them still count.
Two further distinctions matter. Defence contributions counted above pay for costs that are included in the budget, such as the space programme's defence branch and the AI laboratory's defence work. They are separate from the repeat defence orders for autonomous systems in the robotics mission, which are procurement commitments within existing defence budgets and appear in neither the costs nor the sources. The demand compacts in the industry missions are likewise procurement commitments by existing buyers rather than new spending.
4The public investment banks
The equity and loans in the first block draw on capacity that the public investment banks and the Sovereign AI Unit would allocate to the programmes. The amounts are small relative to the banks' capacity, but the note budgets an expected loss of a quarter of the amount invested as a cost, a prudent assumption for first-of-a-kind investments that each bank's own risk assessment will replace.
These allocations are separate from the growth fund programme in Capital for Britain, which commits up to £5 billion of British Business Bank capacity over ten years to matching private investment in British growth funds. Taken together, the two programmes remain well within the capacity of the public investment banks.
| Use | Years 1 to 5 | Provider |
|---|---|---|
| State equity in the AI laboratory | £3bn | Sovereign AI Unit or British Business Bank |
| Advanced manufacturing scale-up window | About £0.9bn | British Business Bank |
| Advanced materials and rare earth magnets | About £0.5bn | National Wealth Fund |
| Life sciences loans and growth capital | About £1bn | National Wealth Fund and British Patient Capital |
| Autonomy growth facility | About £0.4bn | British Patient Capital |
| Total allocated | About £5.75bn | Against combined capacity of £25.6bn at the British Business Bank and £27.8bn at the National Wealth Fund |
| Expected loss budgeted as a cost | About £1.4bn | A quarter of the amount allocated, included in the costs in section 2 |
5Later blocks
The second block, years six to ten, costs somewhat more, mainly because the space programme builds its 500 megawatt station. By then defence spending is committed to rise to 3.5% of GDP for core defence, with up to a further 1.5% for defence-related spending. A modest share of that increase, directed to sovereign computing, space and autonomy, would cover most of the growth in cost without new taxation.
The third block, years eleven to fifteen, is dominated by the construction of the first 2 gigawatt station, at an average public cost of about £5.2 billion a year. The space paper expects private finance to be raised against the orbital power contract from this point, and bill payers to contribute through that contract only for power actually delivered.
6Conditions
Money for each block will be released only when the previous block has met its published tests, and programmes that fail will be narrowed or closed under the rules in P6 of Building British Industry. The review of business support will publish the named reallocations that support the sources above, and where it finds less can be redirected without loss, the difference will be shown openly as additional spending or met by narrowing the programmes.
A country that already spends tens of billions a year supporting business and buying defence technology can afford to aim those resources at the frontier. What it cannot afford is to keep spreading them so thinly that they buy nothing that lasts.