SEBE Revenue Model

Jason Huxley Version 2.0 February 2026

1. Purpose

This document derives SEBE revenue estimates from real data: UK energy consumption statistics, data centre capacity figures, wholesale and retail bandwidth pricing, and LLM token economics. It replaces the placeholder £200-500B estimates used in earlier documents, which were never calculated from the stated rates.

The earlier estimates implied energy consumption of 1,000-1,600 TWh at the stated SEE rates. UK total electricity generation is approximately 300 TWh. The error is acknowledged and corrected here.


2. Data Sources

Source Data Date
DESNZ (formerly BEIS) UK electricity consumption by sector 2023
Ofgem Commercial and domestic electricity prices 2025/26
IEA Global data centre electricity demand projections 2024
LINX London Internet Exchange peak throughput (~7 Tbps) 2024
Ofcom Connected Nations UK broadband and mobile traffic 2024
Cloudflare/drPeering Wholesale IP transit pricing trends 2024
Google Cloud Egress pricing (europe-west2, London) Feb 2026
AWS EC2 data transfer pricing Feb 2026
OpenAI API token pricing (GPT-5.2, GPT-5 mini) Feb 2026
Anthropic API token pricing (Opus 4.6, Sonnet 4.5, Haiku 4.5) Feb 2026
ONS UK population mid-2023 estimate 2024

3. SEE Revenue Derivation

3.1 UK Electricity Consumption

Total UK electricity consumption: approximately 300 TWh/year.

Sector TWh Notes
Households ~105 Always exempt from SEE
Commercial (>500kW facilities) ~45 Subject to SEE
Industrial (>500kW facilities) ~25 Subject to SEE
Commercial (<500kW) ~65 Below current threshold
Industrial (<500kW) ~15 Below current threshold
Transport and other ~45 Not applicable
Total ~300  

Current SEE tax base: ~70 TWh (facilities exceeding 500kW IT/operational load).

Not all 150 TWh of commercial/industrial electricity is consumed in facilities above the 500kW threshold. Most commercial electricity is used by small and medium enterprises (shops, offices, restaurants) that fall below 500kW. The 500kW threshold targets data centres, large industrial operations, automated warehouses, and similar facilities.

3.2 SEE Rate Structure

Consumption Bracket Rate (£/kWh)
0-500kW Exempt (small business protection)
500kW-5MW £0.05
5MW-50MW £0.15
>50MW £0.30

3.3 Distribution Across Tiers

Estimated distribution of the 70 TWh tax base:

Tier Estimated TWh Rate Revenue
500kW-5MW 25 £0.05/kWh £1.2B
5MW-50MW 25 £0.15/kWh £3.8B
>50MW 20 £0.30/kWh £6.0B
Total SEE 70   £11.0B

Weighted average SEE rate: ~£0.16/kWh (across the 70 TWh base).

3.4 Rate Adjustment

The rates above are conservative. To provide a meaningful revenue base while remaining within the range commercial operators can absorb, a moderate rate adjustment is proposed:

Tier Current Rate Adjusted Rate Rationale
500kW-5MW £0.05 £0.08 Still below commercial electricity price
5MW-50MW £0.15 £0.20 Approximately doubles energy cost for mid-tier
>50MW £0.30 £0.45 Significant but absorbable for hyperscalers

Impact on data centre operating costs:

Energy represents 30-40% of total data centre operating costs. Current large DC electricity price (long-term PPA): £0.06-0.12/kWh.

SEE Rate Total energy cost Energy increase Total cost impact
£0.05/kWh £0.15/kWh 1.5x +18% total costs
£0.20/kWh £0.30/kWh 3.0x +70% total costs
£0.45/kWh £0.55/kWh 5.5x +158% total costs

At £0.45/kWh (top tier), a hyperscaler’s total operating costs increase by approximately 160%. This is severe but not fatal: all domestic competitors face the same rate (no competitive distortion within the UK), and the DCD ensures offshoring is more expensive (see Section 4).

Adjusted SEE revenue (current base):

Tier TWh Adjusted Rate Revenue
500kW-5MW 25 £0.08/kWh £2.0B
5MW-50MW 25 £0.20/kWh £5.0B
>50MW 20 £0.45/kWh £9.0B
Total SEE 70   £16.0B

Non-compute commercial energy (manufacturing, logistics, automated warehouses) adds a further ~60 TWh at lower weighted rates, contributing approximately £8-12B. Total SEE at launch: approximately £24-28B (2026 prices, see Section 3.6).

3.5 Context: Existing Energy Taxes

Tax Rate Revenue
Climate Change Levy (CCL) £0.00775/kWh ~£2B
UK ETS (carbon trading) Variable (~£50/tonne) ~£5B
Fuel duty £0.5298/litre ~£25B
Total energy/environment taxes   ~£50B

SEE at £24-28B would be the largest single energy tax, approximately 3x the CCL but comparable to fuel duty. It is a step change, not without precedent in UK fiscal policy.

3.6 CPI-Indexation

All SEBE rates (SEE and DCD) are CPI-indexed annually, rising automatically with inflation. This follows the precedent of alcohol duty, tobacco duty, and air passenger duty escalators.

Without indexation, SEBE rates erode in real value over time: at 3% annual inflation, the £0.45/kWh top rate would be worth £0.33/kWh in 2040 real terms. CPI-indexation prevents this.

All revenue projections in this document are stated in 2026 real prices. Nominal revenue will be higher due to CPI adjustment of rates. For example, at 3% annual CPI, the nominal £93B mid-scenario for 2040 would be approximately £136B in nominal terms. The real purchasing power is what matters for policy analysis.


4. DCD Revenue Derivation

4.1 DCD Design: Digital Customs Duty

DCD is a border tariff on commercial data crossing the UK digital border, in either direction. It is not a general bandwidth tax.

Who pays:

Who does not pay:

4.2 DCD Rate Derivation

The DCD rate must make offshoring compute more expensive than operating domestically (where the operator pays SEE). If DCD is cheaper than SEE for the same workload, rational actors move offshore, which is the opposite of the intended behaviour.

Matching DCD to SEE-equivalent:

For a given workload, the offshore operator avoids SEE by not being in the UK. DCD must capture at least the same revenue from the data that workload sends and receives across the border.

The ratio of energy consumed to data transferred varies by workload type:

Workload Type Energy per TB Bandwidth per MW per year
AI inference (LLM) Very high ~200 TB/MW/year
Web/SaaS/database Moderate ~2,000 TB/MW/year
CDN/streaming Low ~20,000 TB/MW/year

AI inference is energy-heavy and bandwidth-light. A 50K token input plus 2K token output (typical code review request) transfers approximately 1.6 MB but consumes significant GPU compute. This asymmetry is critical: LLM workloads use far more energy per byte transferred than streaming or web serving.

Breakeven DCD rates by SEE tier:

SEE Tier SEE Cost (per MW/yr) Border Data (per MW/yr) DCD Breakeven
500kW-5MW (£0.08/kWh) £0.8M ~7,000 TB £110/TB
5MW-50MW (£0.20/kWh) £2.0M ~7,000 TB £280/TB
>50MW (£0.45/kWh) £4.3M ~7,000 TB £620/TB

The bandwidth per MW figure uses a balanced workload mix (20% AI, 50% web/SaaS, 30% CDN). This gives approximately 7,000 TB per MW per year of sustained operation (at 85% utilisation, PUE 1.3).

Proposed DCD rate: £300/TB

At £300/TB:

SEE Tier Domestic cost (SEE) Offshore cost (DCD) Ratio
500kW-5MW £0.8M/MW/yr £2.1M/MW/yr Offshore 2.6x more expensive
5MW-50MW £2.0M/MW/yr £2.1M/MW/yr Roughly equal
>50MW £4.3M/MW/yr £2.1M/MW/yr Domestic 2.0x more expensive

This reveals a problem at the top tier: at £300/TB, the largest facilities (>50MW at £0.45/kWh) would find offshoring cheaper than domestic operation. The DCD rate must be higher, or the top SEE tier lower, or both. Two options:

Option A: Higher DCD rate (£500/TB)

SEE Tier Domestic (SEE) Offshore (DCD @£500) Ratio
500kW-5MW £0.8M/MW/yr £3.5M/MW/yr Offshore 4.4x more expensive
5MW-50MW £2.0M/MW/yr £3.5M/MW/yr Offshore 1.8x more expensive
>50MW £4.3M/MW/yr £3.5M/MW/yr Domestic 1.2x more expensive

Still problematic at the top tier.

Option B: Adjusted SEE top rate (£0.35/kWh) with DCD at £400/TB

SEE Tier Domestic (SEE) Offshore (DCD @£400) Ratio
500kW-5MW (£0.08) £0.8M/MW/yr £2.8M/MW/yr Offshore 3.5x more
5MW-50MW (£0.20) £2.0M/MW/yr £2.8M/MW/yr Offshore 1.4x more
>50MW (£0.35) £3.4M/MW/yr £2.8M/MW/yr Domestic 1.2x more

The top tier remains marginal. This is inherent to the problem: the largest facilities have the highest SEE rates, and DCD is a flat per-TB rate that cannot be tier-matched without knowing the operator’s energy profile.

Resolution: Tiered DCD

DCD should be tiered by the offshore operator’s scale (estimated from traffic volume, which correlates with facility size):

Annual border traffic DCD Rate Target
< 10 PB £200/TB Small/medium offshore use
10-100 PB £400/TB Large enterprise offshore
> 100 PB £800/TB Hyperscaler-equivalent

This ensures the DCD rate scales with the operator’s likely energy profile. A hyperscaler pushing >100 PB/year across the border is almost certainly running a large offshore facility that would be in the >50MW SEE tier if domestic.

4.3 Per-Unit Impact

At the proposed working rate of £300/TB (mid-tier):

Activity Data DCD Cost Notes
Single LLM request (50K+2K tokens) 1.6 MB 0.05p Invisible
1,000 LLM requests 1.6 GB 48p Negligible
ChatGPT query (1.5K tokens) 45 KB 0.001p Invisible
Web page load 2 MB 0.06p Invisible
Email + 5MB attachment 5 MB 0.15p Invisible
Video call (1 hour) 1.5 GB 45p Noticeable over time
Software update (500 MB) 500 MB 15p Negligible
Cloud backup (100 GB) 100 GB £30 Material
Enterprise cloud (10 TB/month) 10 TB £3,000/month Significant

DCD is invisible to consumers (who are exempt anyway) and to individual knowledge workers. It becomes material only at enterprise scale, where it creates a genuine incentive to use UK-based providers.

4.4 DCD Revenue Estimate

UK commercial cross-border data flows:

Category Sustained Tbps (est.)
Enterprise cloud (IaaS/PaaS from IE/NL/DE) 3-8
SaaS platforms (US-hosted) 1-4
AI inference APIs (US-hosted) 0.5-2
Financial services offshore compute 2-6
B2B media/conferencing (offshore servers) 1-3
Ad tech / analytics 0.5-2
DevOps / code repos / CI/CD 0.5-1.5
Cloud backup / disaster recovery 1-3
Total 9.5-29.5

1 Tbps sustained for 1 year = approximately 3,942,000 TB.

At the mid-estimate (19.5 Tbps) and weighted average DCD rate (~£300/TB):

However, this declines over time as the deterrent effect works and workloads repatriate to UK facilities (where they pay SEE instead).

Realistic DCD revenue:

The offshore share of compute serving the UK market is estimated at approximately 38% (2025). As DCD incentivises domestic location:

Year Offshore share DCD Revenue Notes
2030 (launch) 35% £8-12B Initial, before repatriation
2033 29% £7-10B Early movers repatriate
2035 23% £7-9B Significant shift domestic
2040 14% £8-10B Most compute domestic
2045 10% £10-19B Floor (some always offshore), offset by volume growth

DCD revenue is roughly stable in the early years (repatriation offsets volume growth) then slowly grows as total compute demand overwhelms the declining offshore share. But it is always secondary to SEE.


5. Combined SEBE Revenue

5.1 Year 1 (Launch, 2030)

Component Revenue (2026 prices)
SEE (~90 TWh taxable by 2030, adjusted rates) £26-34B
DCD (cross-border, £300/TB avg) £8-12B
Total SEBE £34-46B

The tax base is larger than the 2026 baseline (70 TWh) because 3-4 years of compute growth adds approximately 20 TWh of new taxable capacity by 2030 (IEA projections). The mid-scenario estimate is £38B.

5.2 Growth Trajectory

SEBE revenue grows because the tax base grows. Automation increases commercial energy consumption (more data centres, automated warehouses, robotic manufacturing). The IEA projects global data centre electricity demand growing from ~460 TWh (2022) to 1,500-2,000 TWh by 2030. UK data centre electricity alone is projected to grow from ~10 TWh to ~30+ TWh by 2030.

For context: a single world-class hedge fund (GResearch/NMS) is currently building a 234 MW data centre with its own power station in Virginia. That single facility, at 85% utilisation with PUE 1.3, consumes approximately 2.3 TWh/year. At the >50MW SEE rate of £0.30/kWh (current), it would pay £690M/year in SEE. Several such facilities will be built in the UK if the sovereign compute incentive works.

Growth assumptions:

Factor Rate Basis
Data centre electricity growth 15-25% pa IEA projections, AI scaling
Automated industry growth 8-12% pa Robotics, electrification
Traditional commercial 1% pa Efficiency offsets growth
Taxable share of total Rising 2pp/year More facilities cross 500kW threshold
Offshore share Declining 3pp/year DCD deterrent effect, floor 10%

Projected SEBE revenue (2026 real prices, CPI-indexed rates):

Year Total Compute (GW) SEE Revenue DCD Revenue Total SEBE
2030 (launch) 6.9 £30B £8B £38B
2033 9.5 £40B £8B £48B
2035 12.1 £50B £7B £57B
2040 21.3 £83B £10B £93B
2045 37.6 £140B £19B £159B

The key observation: SEE dominates and grows. DCD is the deterrent fence that keeps compute domestic, where SEE can tax it. DCD revenue is secondary and roughly stable. Total SEBE grows because the thing being taxed (automation infrastructure) is the fastest-growing sector of the economy.


6. Stage 1 Ramp Model

6.1 Principle

SEBE does not fund full Stage 1 (UBI at £2,500 + UBS at £2,500 per person) from day one. Instead, UBI and UBS ramp up together as SEBE revenue grows. The mechanism is self-scaling: as automation replaces workers, the tax on automation funds those workers’ replacement income.

6.2 UBS Phasing

UBS components phase in over 5-8 years, starting with the most impactful and cost-effective:

Phase Component Net New Cost Start Year
1 Free public transport £14B (fares replacement) Year 2 (2032)
2 Free energy (to threshold) £8-40B (phased over 5 years) Year 6 (2036)
3 Free broadband + mobile £5B Year 9 (2039)

Existing government spending on transport subsidy (~£12B rail), energy support (~£5B), and broadband programmes (~£1-2B) partially offsets these costs. Net new spending is lower than gross UBS cost.

6.3 Year-by-Year Projection

Revenue is allocated first to UBS commitments, then the remainder funds UBI. UBI grows automatically as revenue exceeds UBS costs.

All figures in 2026 real prices. Adult populations from ONS 2022-based projections (see SEBE Distribution Model Section 3.2).

Year Adults SEE DCD Feedback Total UBS Cost Available Adult UBI Monthly
2030 58.8M £30B £8B £0B £38B £0B £38B £650/yr £54/mo
2032 59.5M £37B £8B £8B £53B £14B £39B £660/yr £55/mo
2035 60.1M £50B £7B £5B £62B £14B £48B £800/yr £67/mo
2036 60.4M £55B £7B £5B £67B £22B £45B £740/yr £62/mo
2039 61.0M £75B £9B £7B £91B £51B £40B £660/yr £55/mo
2040 61.3M £83B £10B £8B £101B £59B £42B £690/yr £57/mo
2045 62.3M £140B £19B £27B £187B £59B £128B £2,050/yr £171/mo

“Feedback” is the additional conventional tax revenue generated by UBI spending (multiplier 0.8, marginal tax rate 33%).

Note: UBI dips in 2036 and 2039 as free energy and broadband phase in. The family is still better off because UBS has direct cash-equivalent value (free energy, transport, broadband worth ~£2,500/person/year at full rollout).

6.4 Family Impact

Combined UBI + UBS value for a family of 4 (2 adults + 2 children):

All figures in 2026 real prices. UBI per adult uses ONS-projected population for each year.

Year UBI (family) UBS Value Combined Monthly
2030 £2,600 £0 £2,600 £217
2032 £2,620 £836 £3,456 £288
2035 £3,100 £836 £3,936 £328
2040 £2,880 £3,522 £6,402 £534
2045 £8,600 £3,522 £12,122 £1,010

UBS value is the per-person cost of active UBS components multiplied by family size. It represents the actual services received (free transport, energy, broadband) which would otherwise cost the family that amount.

6.5 Comparison to Option 1 (Broader Tax Package)

SEBE alone reaches £34-46B at launch (2030). If combined with other progressive taxes that the Green Party already supports:

Source Estimated Revenue (2026 prices)
SEBE (SEE + DCD) £34-46B
Land Value Tax £40-70B
Wealth tax (2% on >£10M) £20-40B
Financial Transaction Tax £10-30B
Carbon tax (£100-200/tonne) £15-30B
Corporation tax reform £15-25B
Expanded Digital Services Tax £3-8B
Employer NI on automation £5-15B
Inheritance reform £5-10B
Total £134-263B

With the UBI economic feedback loop (£30-95B in additional conventional tax revenue), total available revenue reaches £164-358B. At the upper end, this approaches Stage 1 (net cost ~£370B at 2040 projected population) from year one.

SEBE’s distinctive contribution is that it is the only component that grows automatically with automation. The other taxes are largely static or slow-growing. SEBE is the escalator; the other taxes are the foundation.


7. Sovereign Compute Incentive

7.1 The Arbitrage Problem

If DCD is too low relative to SEE, rational actors offshore their compute (avoiding SEE) and pay the lower DCD on cross-border data. The rate structure must ensure offshoring is always more expensive than domestic operation, at every tier.

7.2 Government as Compute Provider

The Amazon model: Amazon needed large-scale compute it could not buy, so it built its own infrastructure and sold off excess capacity as AWS. The UK government could follow the same approach:

This accelerates the repatriation of compute and increases SEE revenue. Government compute facilities would be subject to SEE like any other operator, but the revenue flows back to the exchequer.

7.3 Ramp-Up Period

New data centre capacity takes 2-4 years to build. During the ramp-up period (2030-2034), UK compute supply may not meet demand from repatriating workloads. The DCD should include a transitional allowance:

This gives the market time to build domestic capacity while signalling the direction of travel.


8. Sensitivity Analysis

8.1 SEE Revenue Sensitivity

Variable Low Mid High
Taxable energy (TWh) 60 70 85
Weighted average rate £0.16 £0.25 £0.35
Non-compute energy (TWh) 50 60 70
Year 1 SEE £16B £24B £35B

8.2 DCD Revenue Sensitivity

Variable Low Mid High
Cross-border throughput (Tbps) 9.5 19.5 29.5
Weighted DCD rate (£/TB) £200 £300 £500
Offshore share 30% 38% 45%
Year 1 DCD £4B £8B £15B

8.3 Growth Sensitivity

All figures in 2026 real prices.

Scenario 2030 (launch) 2035 2040 2045
Conservative (10% compute growth) £34B £47B £70B £108B
Mid (12% compute growth) £38B £57B £93B £159B
High (15% compute growth) £44B £74B £135B £260B

The high scenario reflects aggressive AI scaling (consistent with current IEA projections and industry investment patterns). The conservative scenario assumes AI growth decelerates after 2030.


9. Corrections to Prior Documents

The following documents contain revenue figures that are superseded by this analysis:

Document Section Old Figure Corrected Figure
academic_brief.md 3.3 SEE: £150-250B SEE: £16-35B (year 1)
academic_brief.md 3.3 DCD: £50-250B DCD: £4-15B (year 1)
academic_brief.md 3.3 Combined: £200-500B Combined: £20-50B (year 1)
green_party_submission.md 3.1 SEE: £150-250B SEE: £16-35B (year 1)
green_party_submission.md 3.2 DCD: £50-250B DCD: £4-15B (year 1)
cost_model.md 6.3 Stage 1 fully funded by SEBE Stage 1 ramps with SEBE growth
cost_model.md 6.4 £200-500B SEBE revenue £20-50B year 1, growing to £93-260B by 2040
glossary.md DCD entry £25-50/Mbps £200-800/TB tiered by volume
AGENTS.md Key Parameters £200-500B revenue £20-50B year 1, self-scaling

The DCD unit changes from £/Mbps (which was ambiguous and economically nonsensical) to £/TB (measurable, billable, comparable to cloud egress pricing).

The bandwidth rate of £25-50/Mbps had no defined billing period and implied costs of £625/second for a Netflix stream. The per-TB rate is grounded in data centre economics and produces sensible per-unit costs at every scale.


10. What This Means

SEBE at launch (2030) generates approximately £34-46B/year (2026 real prices, mid-scenario £38B). This is larger than Inheritance Tax (£7B) plus Stamp Duty (£15B) plus Climate Change Levy (£2B) plus Tobacco Duty (£10B) combined. It is a significant new revenue source, but it does not fund full Stage 1 from day one.

What SEBE does that no other tax does: it grows automatically with the thing that replaces human employment. Every new data centre, every automated warehouse, every AI inference cluster increases the SEE base. The revenue tracks the problem.

By 2040, SEBE is projected at £93-135B/year (mid-to-high scenario, 2026 prices). By 2045, £159-260B/year. Combined with the broader progressive tax package and the UBI economic feedback loop, Stage 1 is fully funded within 10-15 years of launch, with UBI growing every year. Nominal revenue will be higher due to CPI-indexed rates (see Section 3.6).

The honest pitch: SEBE starts modest and grows with automation. As machines replace workers, the tax on machines funds the workers’ income. The mechanism is self-scaling.


© 2026 Jason Huxley. Licensed under CC-BY 4.0.