Purpose
This proposal introduces the Sovereign Energy and Bandwidth Excise (SEBE): a tax on the physical infrastructure of automated production. SEBE taxes commercial energy consumption (kWh) and cross-border commercial data (TB). It generates £34-46 billion at launch in 2030 (2026 prices), growing automatically with automation to £93 billion by 2040 and £159 billion by 2045.
The case for SEBE does not depend on mass displacement occurring. It is justified by the asymmetric consequences of preparing versus not preparing, and by a structural fiscal gap that is widening regardless of whether anyone loses their job.
The proposal follows the three-framing structure: what is happening now, where things are heading and where we want them to go. Each section identifies the harms and positives, then sets out what SEBE does about them.
1. Now
1.1 Harms That Need Addressing
The tax base is structurally vulnerable.
Income Tax (£329B) and National Insurance (£205B) together provide 42% of UK government revenue (OBR November 2025 EFO, updated February 2026). Both depend on people being in paid work. No existing UK tax grows with automation. If automation displaces workers at scale, this revenue shrinks while welfare costs grow. Whether that happens quickly or slowly is uncertain. That the tax base is exposed to it is not.
Early signals are visible.
AI systems already match or exceed human performance in diagnostics, translation, legal research, code review, financial analysis and customer service across a range of benchmarks (OBR, Economic and Fiscal Outlook, March 2026). Block (the company behind Square, Cash App and Afterpay) cut 40% of staff citing AI (CNN, 26 February 2026). Salesforce eliminated 4,000 support roles after AI deployment (CNBC, 2 September 2025). AI was cited in 12,304 job cut announcements in the first two months of 2026 alone (Challenger, Gray & Christmas, March 2026). Youth unemployment is 16.1%, the highest in over a decade (ONS, February 2026). 946,000 young people are NEET.
A randomised controlled trial of 5,172 customer service agents found AI tools raised novice productivity by 35% while experienced workers gained almost nothing, demonstrating the mechanism by which the expertise premium erodes (Brynjolfsson et al., Quarterly Journal of Economics, May 2025).
These are early signals, not mass displacement. Whether they are the beginning of a structural shift or normal cyclical churn is an open question. But the fiscal consequences of each answer are not symmetric.
The consequences of being wrong are asymmetric.
| Displacement is modest | Displacement is severe | |
|---|---|---|
| We prepare (SEBE) | Automation-linked tax infrastructure built. Revenue exists but is not critical. Small fiscal cost. Reversible. | SEBE revenue scales with the problem. Fiscal architecture intact. Transition is painful but the state functions. |
| We do not prepare | No harm done. The optimists were right. | £534B in employment-linked revenue erodes. No replacement mechanism. Benefits system overwhelmed. No time to build the alternative. |
Three quadrants are acceptable. One is a fiscal crisis. The entire case for SEBE reduces to: which outcome can you least afford to be wrong about?
“Do nothing” is not free. Even without mass displacement, a growing share of economic value is generated by automated systems that pay no employment tax. A data centre generating £100M in revenue employs approximately 50 people. A professional services firm generating £100M employs approximately 500 (illustrative; figures vary by sector and facility type). The tax system captures 42% of the second but barely touches the first. This gap widens every year as the automated share of the economy grows, regardless of whether anyone loses their job.
The gains from automation are concentrating in capital owners.
Returns to automation accrue to those who own the infrastructure (hyperscalers, tech monopolies, shareholders), not to the broader population. UK Corporation Tax receipts from the tech sector are well below headline rates because profits are routed through Ireland, Luxembourg and the Netherlands. The productive gains of automation are real. Their distribution is not.
UK compute infrastructure depends on foreign operators.
A significant share of UK commercial compute runs offshore (AWS Ireland, Azure Netherlands, GCP Belgium). Revenue from this economic activity escapes UK taxation entirely. The UK has no fiscal mechanism for taxing the productive value of offshore compute serving the UK market.
1.2 Positives Worth Protecting and Building On
Automation is increasing productive capacity.
Manufacturing output per worker is rising. AI inference outperforms humans in pattern recognition, diagnostics and materials science. Robotic process automation eliminates repetitive tasks. Unit costs are falling across automated sectors. This is not a harm. It is a structural gain that should be welcomed and harnessed.
The green transition is accelerating.
AI optimises energy grids, accelerates materials science research and improves building efficiency. Renewable energy costs have fallen ~90% in a decade. Computing costs fall ~30% annually. These trends, properly directed, can decarbonise the economy faster than any human-managed programme.
UK data centre investment is growing.
Hyperscalers are investing billions in UK data centre capacity (London, Slough, Manchester). This is productive infrastructure. A policy framework that incentivises further domestic investment (rather than offshoring to Ireland or the Netherlands) strengthens the UK’s compute sovereignty and its tax base simultaneously.
1.3 What SEBE Does Now
SEBE addresses the harms while strengthening the positives:
-
Creates withdrawal capacity that grows with automation, closing the structural gap between where value is generated and where it is taxed. At launch, SEBE withdraws £34-46 billion from the corporate sector, creating fiscal space for the government to maintain aggregate demand.
-
Taxes physical resource consumption, not accounting constructs. A multinational can declare zero profit in the UK through transfer pricing. It cannot route its electricity consumption through Ireland. The meter reads what it reads.
-
Incentivises domestic compute investment. The Digital Customs Duty (DCD) sets offshore compute at a permanent price disadvantage. Companies that repatriate compute to the UK pay the Sovereign Energy Excise (SEE) instead, which is cheaper. The result is more UK data centres, more UK jobs in construction and operations, and a stronger domestic tax base.
-
Rewards energy efficiency. SEE taxes energy consumption per kWh. A facility that achieves the same output with less energy pays less tax. This is the correct incentive: automate well, not wastefully.
-
Exempts consumers and small businesses. The 500kW threshold means no household, shop, restaurant or office pays SEBE. This is a tax on commercial automation infrastructure, not on people.
2. Likely Future
2.1 Where Things Are Heading Without Intervention
Employment-linked revenue faces structural risk.
The Anthropic CEO has warned that AI could displace 50% of entry-level white-collar jobs within 1-5 years (Dario Amodei, The Adolescence of Technology, January 2026). The OBR worst case projects 500,000 additional unemployed with no growth uplift and £9 billion extra government borrowing (OBR, Economic and Fiscal Outlook, March 2026). The Bank of England is war-gaming a full AI shock scenario for inclusion in banking stress tests (Bloomberg, 5 March 2026).
These are not predictions. They are scenarios that central banks and fiscal authorities are actively modelling. The mechanism is straightforward: each displaced worker removes a salary from the Income Tax and NI base and adds a claim to Universal Credit. Revenue shrinks; costs grow.
Corporation Tax will not fill the gap.
Multinational firms structure their affairs to declare minimal UK profit regardless of how much economic activity they conduct here. The effective rate on large tech firms is a fraction of the headline rate. Even with OECD Pillar Two (15% minimum), the additional revenue is modest (~£2-3B) and easily absorbed by new avoidance structures.
Energy demand from compute will grow substantially.
The IEA projects UK data centre electricity demand to triple by 2030 (from ~10 TWh to ~30 TWh; IEA, Energy and AI, 2025). AI training runs are becoming larger, not smaller. Automated warehouses, robotic logistics and AI inference clusters all draw power at scale. This energy consumption is currently untaxed (beyond the Climate Change Levy at £0.00775/kWh, which raises only £2B nationally).
If displacement is severe and no fiscal preparation has been made:
- A structural deficit that grows with automation
- Austerity imposed on the population whose jobs were destroyed
- Concentration of automation gains in a small number of firms
- Increasing dependence on offshore compute infrastructure with no UK fiscal capture
- Political pressure for coercive responses (workfare, surveillance, benefit conditionality) as the conventional fiscal framework fails
If displacement is modest, the structural gap identified in Section 1.1 still widens. The automated share of the economy grows tax-free while employment taxation captures a shrinking proportion. The trajectory is the same, just slower. In either case, a fiscal mechanism linked to automation infrastructure is justified.
2.2 Is This a Direction We Want?
No. The Green Party has long recognised that GDP growth is an inadequate measure of progress (EC100, EC310) and that ecological sustainability requires structural economic change (EC200). The party is committed to Universal Basic Income (EC730), to abolishing National Insurance (EC727, EC764) and to ensuring that taxation incorporates the real costs of resource consumption (ST334, EC776).
None of these commitments can be met within a fiscal framework that depends on employment taxation. The party needs a mechanism that replaces the revenue (or, in MMT terms, the withdrawal capacity) that employment taxation currently provides.
2.3 What SEBE Does to Reshape the Trajectory
SEBE creates a fiscal instrument whose capacity grows with automation rather than declining with employment:
| Instrument | Linked to | As automation increases |
|---|---|---|
| Income Tax | Employment | Withdrawal capacity falls |
| National Insurance | Employment | Withdrawal capacity falls |
| Corporation Tax | Accounting profit | Withdrawal capacity gameable |
| VAT | Consumer spending | Withdrawal capacity falls with demand |
| SEBE | Automation infrastructure | Withdrawal capacity grows |
SEBE is the only UK fiscal instrument whose capacity scales with the force that may create the need for greater public spending, and that is already generating untaxed economic value.
By 2040, SEBE withdrawal capacity reaches £93 billion. By 2045, £159 billion (approaching NI scale). This is not a one-off revenue boost. It is a structurally growing tax base tied to the fastest-expanding sector of the economy.
3. Wanted Future
3.1 What the World We Want Looks Like
A society where:
-
Automation serves everyone, not just capital owners. The productivity gains from AI, robotics and automated production are distributed through unconditional income and universal public services.
-
Nobody is compelled to work in order to survive. A Universal Living Income (ULI), funded by SEBE and complementary instruments, provides a floor high enough that employment is a choice, not a condition of survival.
-
The tax base tracks the economy as it actually is. Taxation is levied on the physical inputs to production (energy, data) rather than on the increasingly marginal human contribution. The fiscal architecture matches the productive architecture.
-
The UK is a compute sovereign. Domestic data centre investment is incentivised by policy. UK businesses run on UK infrastructure. The tax base, the skills, the supply chain and the strategic independence all reside within UK jurisdiction.
-
Energy efficiency is rewarded. SEBE taxes energy consumption, not output. Companies that produce more with less energy pay less tax. The incentive is to innovate, not to avoid automation.
3.2 What Policies Move Us There
SEBE is the fiscal mechanism. It provides the withdrawal capacity that creates space for the government to spend on redistribution, public services and infrastructure without inflationary pressure. The spending decisions (UBI, UBS, NHS, green investment) are political choices. SEBE provides the fiscal foundation.
The two-stage transition:
Stage 1 (from 2030 launch): SEBE revenue funds an initial UBI starting at ~£650 per adult per year (58.8 million adults, ONS 2030 projection). Not life-changing in year one, but universal, unconditional and self-scaling. As automation grows and SEBE revenue grows, the payment ratchets upward automatically toward a target of £2,500 per adult per year. Universal Basic Services (free transport, energy, broadband, mobile) phase in as revenue permits.
Stage 2 (as automation matures): UBI ratchets toward Universal Living Income: £29,000 per adult per year (matching median take-home pay of £31,627 minus £2,500 UBS value). This requires SEBE at maturity plus complementary instruments (wealth tax, land value tax, financial transaction tax). Stage 2 is achievable if, and only if, automation delivers the productivity gains that expand the economy’s real productive capacity. The two-stage design allows observation and adjustment: Stage 1 provides real-world data before Stage 2 is attempted.
SEBE operationalises existing Green Party policy:
| Policy | Commitment | SEBE Connection |
|---|---|---|
| EC661 | Money as common resource (MMT framework) | SEBE’s theoretical foundation |
| EC730 | Universal Basic Income | SEBE creates fiscal space for UBI |
| EC727/EC764 | Abolish employee and employer NI | SEBE replaces the ~£205B NI withdrawal capacity |
| ST334 | Taxation must incorporate real costs of resources | SEE taxes the energy externality of automation |
| EC776 | Eco-taxes on those responsible for environmental costs | SEE is an eco-tax on commercial energy consumption |
| EC779 | Import duties reflecting ecological/social impact | DCD is a border tariff on offshore commercial compute |
| EN110 | Border tariffs on high embodied energy imports | DCD captures the embodied energy of offshore compute |
| EC1050/EC1052 | Electronic transactions escape taxation in country of origin | DCD addresses the offshore compute problem these policies identified |
4. The Mechanism
4.1 Sovereign Energy Excise (SEE)
What it taxes: Commercial electricity consumption at facilities with IT load above 500kW.
Who pays: Data centres, automated warehouses, large-scale manufacturing, AI training facilities.
Who does not pay: Households, small businesses, shops, restaurants, offices below 500kW.
Rate structure (2026 prices, CPI-indexed annually):
| Consumption Bracket | Rate (£/kWh) |
|---|---|
| 0-500kW | Exempt |
| 500kW-5MW | £0.08 |
| 5MW-50MW | £0.20 |
| >50MW | £0.45 |
Revenue at launch (2030): £26-34 billion (2026 prices). For context: the Climate Change Levy raises £2 billion at £0.00775/kWh. SEE at a weighted average of £0.16/kWh is a step change, but comparable in scale to fuel duty (£25 billion).
Metering: Hardware Root of Trust at three points (generation, storage, load). Tamper-evident, cryptographically attested. The meter reads what it reads.
4.2 Digital Customs Duty (DCD)
What it taxes: Commercial data crossing the UK digital border (both directions).
Who pays: UK businesses using offshore cloud providers (AWS Ireland, Azure Netherlands, GCP Belgium), UK businesses calling offshore APIs (OpenAI US, Anthropic US), UK financial firms with offshore compute.
Who does not pay: UK consumers (exempt), UK businesses using UK-based data centres (they pay SEE instead), educational and research institutions, NHS and emergency services.
Rate structure:
| Annual Border Traffic | Rate |
|---|---|
| < 10 PB/year | £200/TB |
| 10-100 PB/year | £400/TB |
| > 100 PB/year | £800/TB |
Revenue at launch: £8-12 billion (2026 prices). Revenue remains roughly stable as the deterrent effect repatriates compute to UK facilities.
Enforcement: At Internet Exchange Point (IXP) level. Aggregate commercial throughput measured at the point where the cable meets UK jurisdiction. No individual traffic is inspected. Weighing the container, not opening it. One of three classification methods (SNI inspection) will become less effective as Encrypted Client Hello adoption increases; BGP routing and flow metadata remain functional regardless (see Appendix A, Section A7).
4.3 Why Energy, Not Compute
Proposals to tax FLOPS (floating-point operations), individual robots or AI models all fail on measurement:
- FLOPS is not a standardised unit (FP64, FP32, FP16, INT8 produce different counts for the same work)
- No physical meter exists for compute operations (hardware counters are trivially spoofable)
- Evasion is architectural (switch hardware type, change precision, use neuromorphic compute)
- A FLOPS tax punishes efficiency (newer hardware doing the same work in fewer operations pays less)
Energy is architecture-neutral, physically measurable, tamper-resistant and efficiency-rewarding. Every computation, regardless of hardware, consumes energy. It is the correct proxy.
4.4 Revenue and Growth
All figures in 2026 real prices. All rates CPI-indexed annually. Figures shown are mid-scenario central estimates. The full 2030 launch range is £34-46B (conservative: £34B, high: £44B). See SEBE Revenue Model Section 8.3 for scenario analysis.
| Year | SEE | DCD | Total SEBE |
|---|---|---|---|
| 2030 (launch) | £30B | £8B | £38B |
| 2033 | £40B | £8B | £48B |
| 2035 | £50B | £7B | £57B |
| 2040 | £83B | £10B | £93B |
| 2045 | £140B | £19B | £159B |
Full derivation from first principles (UK energy consumption data, data centre capacity, wholesale bandwidth pricing, LLM token economics) in the companion document SEBE Revenue Model.
5. The MMT Framework
For members familiar with Modern Monetary Theory (and consistent with EC661), SEBE fits naturally within the MMT fiscal framework.
The UK government does not need SEBE revenue to spend. A sovereign currency issuer creates money when it spends and destroys it when it taxes. The real constraint is inflation, not revenue.
SEBE’s function is inflation control. It withdraws purchasing power from the corporate sector (specifically, from automation-intensive operations) to create fiscal space for government spending without exceeding productive capacity.
The withdrawal instrument scales with automation. As automation grows, energy and bandwidth consumption grow (SEBE withdrawal capacity increases). If workers are displaced, the government must spend more (the tap runs harder). The drain and the tap are linked to the same variable. If displacement is modest, SEBE still closes the structural gap between where value is generated and where it is taxed. If displacement is severe, SEBE scales with the crisis.
Automation expands productive capacity. More goods and services can be produced with fewer inputs. This widens the fiscal space. SEBE taxes the infrastructure of the process that widens it. The tax and the capacity expansion are structurally linked.
SEBE is not hypothecated. Revenue is general-purpose. The government allocates spending through democratic processes, responding to economic conditions. This is consistent with both MMT principles and Green Party fiscal policy.
For readers who prefer the intuitive framing: SEBE taxes automation infrastructure and uses the revenue to fund public spending. Both framings describe the same mechanism. The MMT framing is more precise about why it works.
6. Addressing Objections
6.1 “This will drive businesses offshore”
DCD ensures offshore compute is always more expensive than domestic. A company that moves its data centre to Ireland pays DCD on every byte of cross-border traffic. A company that operates in the UK pays SEE, which is set below DCD. The incentive is to invest domestically.
6.2 “Energy taxes will harm the green transition”
SEBE taxes energy consumption, not energy source. A renewable-powered data centre pays the same SEE rates but benefits from lower energy input costs. The combination of SEE and cheap renewables creates a strong incentive to build green compute infrastructure. AI research that accelerates the green transition is not penalised.
The carbon levy (EC777, EN080) and SEE tax different externalities: emissions vs labour displacement. They are complementary.
6.3 “Consumers will pay through higher prices”
SEE applies only to facilities above 500kW. No household or small business pays. Some cost pass-through to consumers is possible (as with any tax on production), but the magnitude is limited:
- Energy is 15-25% of data centre operating costs
- Hyperscaler operating margins are 28-42% (Microsoft Azure 42%, AWS 37%, GCP 28%, FY2025 earnings)
- SEE at the lowest tier (£0.08/kWh) is modest relative to commercial electricity costs (£0.22-0.28/kWh)
Significant pass-through would require operators to sacrifice competitive position. In a market with multiple providers, the pressure is to absorb or efficiency-reduce, not to pass through.
6.4 “You cannot tax what you cannot measure”
SEBE taxes two of the most measurable things in the economy. Energy consumption is already metered at every commercial facility. Data throughput is already counted at every Internet Exchange Point. ISPs already meter aggregate commercial throughput for billing purposes.
Hardware Root of Trust metering (tamper-evident, cryptographically attested) closes evasion routes. One physical measurement replaces thousands of pages of corporate accounting.
6.5 “Inflation will eat UBI”
Stage 1 UBI starts at ~£650 per adult per year (~£54/month). This is a rounding error relative to median income. Demand-pull inflation from this injection is negligible.
SEBE simultaneously withdraws £34-46B from corporate balance sheets (anti-inflationary). The automation SEBE taxes also reduces production costs (deflationary at the unit level). The net inflation risk at Stage 1 is low.
Stage 2 (full ULI) depends on productivity gains from automation expanding the economy’s real productive capacity. The two-stage design allows observation and adjustment before any large fiscal expansion is attempted.
7. Implementation Approach
7.1 Phased Deployment
| Phase | Timeline | Action |
|---|---|---|
| 1 | 2027-2028 | Legislation: Energy Act and Finance Act amendments |
| 2 | 2028-2030 | Metering rollout: HRoT at facilities >500kW, IXP monitoring |
| 3 | 2030 | Launch: SEE and DCD live, monthly corporate invoicing |
| 4 | 2032 | Review: rates, thresholds and inflation impact assessed |
7.2 Administrative Simplicity
SEBE is simpler to administer than income tax. Meter readings replace corporate accounts. No deductions, allowances or exemptions to negotiate. Estimated administrative cost: less than 1% of revenue.
7.3 Expert Consultation Needed
The 500kW threshold, tiered rates and specific values in this proposal are worked examples based on current data. They are a starting point for consultation, not fixed policy. Expert input from the energy sector, network engineering and tax administration should inform the final design.
8. Recommended Next Steps
- Economy PWG review: Assess SEBE within the MMT fiscal framework and Green Party economic policy (EC661, EC730, EC727/EC764)
- Science & Technology PWG review: Assess technical feasibility of SEE metering and DCD enforcement
- Expert consultation: Commission independent review of rate structures, threshold calibration and revenue projections
- Formal modelling: CGE or SFC analysis of SEBE macroeconomic effects, in partnership with a research institution (IPPR, NEF or university)
- Pilot programme design: Voluntary trial with 10-20 large facilities to validate projections and compliance costs
- Conference motion: Prepare policy motion for Autumn 2026 conference, supported by modelling results and consultation feedback
Companion Documents
| Document | Contents |
|---|---|
| SEBE Revenue Model | SEBE revenue derivation from first principles |
| SEBE Cost Model | Revenue in context, scale comparison |
| SEBE Distribution Model | Two-stage UBI to ULI illustrative model |
| Infrastructure-Based Taxation for the Post-Employment Economy | Full technical paper (orthodox framing) |
| Inflation Control for the Post-Employment Economy | Full technical paper (MMT framing) |
| The Asymmetry of Automation Risk | Six automation waves, asymmetric risk analysis (MMT framing) |
| The Asymmetry of Automation Risk (Orthodox Framing) | Orthodox variant: revenue and tax base framing |
All documents are open-source, licensed CC-BY 4.0, and available at github.com/Djarid/SEBE.
Copyright 2026 Jason Huxley. Licensed under CC-BY 4.0.