01
48 / 52
Construction starts · 2017–2024
Of 52 nuclear reactor construction starts worldwide, 48 used Chinese or Russian designs.
IEA · published Jan 2025
Civilian nuclear power · Industrial deployment · Evidence cut 28 July 2026
Siderian
How to Win the Second Nuclear Race
There is a nuclear race, and the West is losing.
What the race is
The race is to build civilian nuclear power plants as repeat fleets: one proven design, supplied and constructed again and again.
What Siderian does
Siderian is building the decision and evidence system that identifies which Western AP1000 projects are ready, verifies the suppliers behind them, protects one standard design, and assigns scarce industrial capacity by evidence.
One fleet. One reference design. One evidence trail. Positions earned by readiness.
The case · public evidence
The race is industrial, immediate, and economic.
The current score shows who is building now. Three further numbers explain why the gap matters: who captured the last construction cycle, how fast electricity demand is rising, and the cost repeat Western nuclear must target.
Why starts matter
Repeated construction preserves factories, skilled teams, quality records, financing confidence, and control of the standards that shape the next fleet.
Evidence cut · 28 July 2026
02
415 → 945
Data-centre electricity · TWh
Data-centre electricity, 2024 → 2030, IEA Base Case.
IEA · published Apr 2025
03
$50–60
Economic objective · USD/MWh
The target range for repeat Western nuclear deployment. An objective, not a forecast or guarantee.
IEA/NEA + INL · assumptions disclosed
Every material figure is dated and linked to its source. Model outputs, policy tariffs, Siderian analysis, and economic objectives are identified as different kinds of evidence.
Open evidence ledger →Founder's letter
There is a nuclear race, and the West is losing.
Demian Reidel
Founder, Siderian
The AI buildout is accelerating electricity demand. The IEA estimates that data centres used about 415 terawatt-hours in 2024 and could use about 945 by 2030. The world needs far more power that is clean, firm, and scalable. Nuclear is one of the few proven sources capable of delivering all three at system scale.
Look at the construction record. Since January 2025, twenty nuclear reactors have begun construction. Twelve use Chinese designs. Three use Russian designs. Two are South Korean, two are Indian, and one is Western: Canada's Darlington BWRX-300.
15:1
Against China and Russia, the score is now fifteen to one.
One Western start is not a fleet.
The countries that build repeat nuclear programs will shape the standards, supply chains, skills, and financing architecture of the next industrial era. That position is being formed now.
The West is not losing for lack of reactor technology. Western institutions helped pioneer it. The West is losing because it allowed serial deployment to disappear.
China builds a fleet, the West makes collectors' items.
No durable nuclear fleet built itself. France made a national program. South Korea built a standardized export system. China is sustaining the world's largest active construction program. Fleets are designed, financed, governed, and repeated.
My own record crosses physics, capital markets, central banking, public strategy, and board-level nuclear governance. It does not include operating a nuclear plant or delivering an AP1000, and Siderian will not pretend otherwise. That boundary is why independent technical judgment and assurance governance must be designed into the institution.
To win the second race, the West must turn credible reactor technology into a repeat fleet. That requires a system that qualifies demand, verifies capacity, protects the reference design, governs scarce production slots, and gives capital evidence it can use.
The economic objective is equally explicit: move repeat Western nuclear toward $50–60 per megawatt-hour. That is a target, not a promise. It is the standard against which the industrial system should be judged.
Part I · Industrial diagnosis
How the West lost the first industrial nuclear race
The West did not lose nuclear because nuclear failed. It lost because it stopped building long enough for the industrial base to forget how.
The loss was not operating capability — the United States still runs the world's largest nuclear fleet. The loss was serial new-build capability: the institutional habit of turning the same design, the same suppliers, the same records, and the same construction discipline into repeat deployment.
France proved that scale works when continuity is protected. China and Russia then preserved the deployment machine while Western demand fragmented. The West still has vendors, utilities, regulators, and capital. It has pieces, not a system.
Figure 1 · The West built first. China built cadence.
Sourced public-data figure
Operating reactors, five-year intervals plus 2026 year-to-date. The United States built early, then plateaued. Germany built, then exited. China started late, then compounded. Source: World Nuclear Association reactor register, cut 28 July 2026; historical year-end counts reconstructed from grid-connection and closure records. The evidence ledger discloses the current one-unit timing difference with IAEA PRIS.
Sources: World Nuclear Association reactor register and country profiles. Method and cut disclosed in the evidence ledger.
Figure 2 · Construction starts by design origin
Sourced public-data figure
Construction starts · 2017 – 2024
48 / 52
Chinese or Russian designs captured 92.3% of global construction starts.
- Chinese design25 · 48.1%
- Russian design23 · 44.2%
- UK / European designs2 · 3.8%
- South Korea / national2 · 3.8%
U.S.-origin reactor construction starts: 0
Of 52 nuclear reactor construction starts worldwide from 2017 through 2024, 48 used Chinese or Russian designs. A construction start is first structural concrete for the reactor building.
Source: IEA, The Path to a New Era for Nuclear Energy, January 2025.
Part II · The live race
The West is losing the second industrial nuclear race
The first race was lost when demand became too thin to preserve the build machine. The second can be lost for the opposite reason: demand is arriving before the system is ready.
The score establishes the industrial gap. The second-race problem begins after demand appears: announcements can claim scarce production attention before licensing, finance, reference discipline, and supplier capacity are ready.
15
China + Russia
1
West · Canada
4
South Korea + India
Total: 20. WNA records Darlington SMR-1 under construction from 26 June 2026. Design-origin methodology and caveats are disclosed in the evidence ledger.
First race
Lost by absence.
Demand became too thin to preserve suppliers, records, skills, and repetition.
Second race
Can be lost by congestion.
Demand is returning before qualified capacity is ready. Weak signals can occupy the production path before executable projects.
Why now
Three clocks are aligning.
Demand, finance, and supply policy are moving at once. Qualified capacity is not moving at the same speed.
01 · Demand
415 → 945 TWh
AI data-centre load, 2024 → 2030, on 2–3-year build cycles nuclear supply chains cannot match.
IEA, Energy and AI
02 · Finance
World Bank + IAEA
The World Bank's first nuclear re-engagement in decades opens the door for other development banks and private investors.
World Bank Group / IAEA, June 2025
03 · Supply
$17.5 billion
DOE conditional long-lead financing for 10 large-scale reactors, with utility partners still to be selected.
U.S. DOE, June 2026
The qualification problem
Named demand is not executable demand.
- Press releasereactor vessel
- Letter of intentqualified buyer
- Supplier nameverified capacity
- Scheduleevidence
- Conditional financefinal project readiness
The proposed standard asks for evidence of site basis, capital authority, reference-design discipline, critical capacity, and record continuity before a project can hold an industrial position.
See the gates →The economic objective
$50–60
USD / MWh
China has set the industrial benchmark. The West must match it.
The objective is Western nuclear power in the $50–60/MWh range. It is a design target for the industrial system, not a Siderian forecast or guarantee.
Siderian does not set tariffs, finance plants, design reactors, execute construction, or guarantee LCOE. It is being built to improve the industrial evidence, sequencing, repetition, and risk allocation on which lower costs depend.
China new-nuclear model · 3% / 7% / 10%
$49.92 / $66.01 / $82.08
LCOE per MWh in 2018 USD, 85% capacity factor. The spread isolates discount-rate sensitivity.
China national benchmark on-grid tariff
CNY 0.43 / kWh
A 2013 policy benchmark for qualifying new units. A tariff is a revenue rule, not an LCOE measure.
AP1000 · tenth series · moderate U.S. model
$49 / $65
2024 USD/MWh with a 40% investment tax credit / without the credit. Preliminary INL model.
AP1000 · second series · moderate U.S. model
$72 / $106
2024 USD/MWh with a 40% investment tax credit / without it. The target is not immediate.
What the evidence supports
The public basis is real but conditional. The joint IEA/NEA model reports Chinese new-nuclear LCOE at $49.92/MWh with a 3% discount rate, $66.01 at 7%, and $82.08 at 10%. Those are 2018 dollars, an 85% capacity factor, and a forward-looking 2025 commissioning assumption — not an audited current fleet average.
The same table shows why industrial confidence matters. For capital-intensive nuclear, the price of risk can move the result by more than $30/MWh before the reactor changes at all. Construction cost, duration, financing structure, capacity factor, fuel, and operating cost still matter.
A 2025 Idaho National Laboratory model gives AP1000 a pathway toward the range through repeat deployment. In its moderate U.S. scenario, the tenth two-unit series reaches $49/MWh with a 40% investment tax credit and $65 without it, in 2024 dollars. INL explicitly calls the projections preliminary, academic, and not investor-grade.
IEA/OECD NEA, Projected Costs of Generating Electricity 2020, Table 3.13a; China's National Development and Reform Commission benchmark tariff; Idaho National Laboratory, INL/RPT-25-84701. Full assumptions and caveats.
The cost stack
Four variables. Siderian governs three interfaces.
A durable real-dollar target range for repeat nuclear generation at plant level. Every public comparison states its dollar year, financing assumption, and measurement boundary.
- 01
Build cost
Repeat design, qualified suppliers, workforce continuity, and controlled rework determine overnight capital cost.
Qualified orderbook, capacity ledger, reference-package control, and evidence reuse.
- 02
Build time
Longer construction adds site overhead and financing cost before a unit produces electricity.
Slot sequencing, long-lead verification, readiness gates, and continuity rules.
- 03
Cost of capital
Nuclear LCOE is highly sensitive to the discount rate because most cost is committed before operation.
Decision-grade evidence and an explicit capital-recognition standard.
- 04
Fleet performance
Capacity factor, operating cost, fuel, and asset life determine how capital cost is spread over output.
Outside Siderian's operating authority; retained as an explicit dependency rather than a claimed result.
Why Siderian is necessary
A platform cannot win the race. A fleet can.
AP1000 is the nuclear power plant platform. Siderian is developing the rules and evidence layer that could turn separate projects into repeat deployment.
Without a governing layer, demand fragments, suppliers cannot plan against a credible orderbook, design deltas destroy evidence reuse, scarce slots go to the loudest buyer, and capital prices every unit as another exception. The market has no automatic mechanism that prevents that outcome.
What it proposes to do
Siderian proposes to qualify demand, verify capacity evidence, govern readiness gates, sequence scarce industrial slots, and preserve a decision record.
What it does not do
Siderian does not design, license, construct, finance, insure, own, or operate a nuclear power plant. It does not replace the NRC, another national regulator, the reactor vendor, the licensee, the EPC, the owner, or their legal duties.
Current status: proposed standard and founding-stage operating model. Independent assurance governance is in formation.
- 01
Qualify demand
Test whether a proposed buyer and site are ready to enter a fleet path.
- 02
Verify capacity
Replace supplier names and schedules with evidenced throughput windows.
- 03
Protect the reference
Control deviations so learning and records can compound across units.
- 04
Govern the queue
Advance, condition, re-sequence, or release demand by transparent rules.
Proposed standard
Six laws for repeat deployment.
The system is designed to make readiness visible, preserve the standard plant, and stop one weak project from damaging a fleet sequence.
- 01
Demand aggregates
Fragmented interest becomes a same-design orderbook suppliers can plan around.
- 02
Design freezes
The reference package is protected; deltas must earn their way into the fleet.
- 03
Capacity reserves
Verified throughput is reserved by readiness, collateralized by escrowed deposits.
- 04
Evidence compounds
Every claim is linked to its record, reusable across units and over time.
- 05
Capital recognizes
Output counts only when a counterparty that bears risk changes its terms.
- 06
Failure isolates
Weak units cure, re-sequence, transfer, or release before local failure becomes fleet failure.
Institutional boundary
Westinghouse remains the platform owner and design authority. Regulators retain their statutory authority. Owners, operators, EPCs, lenders, and insurers retain their legal duties. Siderian's proposed role is narrower: qualify, verify, sequence, and preserve the evidence trail.
Qualification Decision · demonstration
Three plausible projects. One scarce position.
Choose which project should receive the position. Then apply the published evidence gates. The system may overrule prominence, political weight, or requested date.
Fictional, anonymized demonstration. It is not a customer qualification or assurance decision.
Capacity positionSIS-01
Claimants03
Evidence gates05
StateUnallocated
| Published evidence gate | Dossier A | Dossier B | Dossier C |
|---|---|---|---|
| 01Site basisIs the site and licensing path evidenced? | Conditional | Verified | Verified |
| 02Capital authorityIs there decision authority and credible funding evidence? | Verified | Verified | Conditional |
| 03Reference disciplineDoes the project protect the standard plant and control deltas? | Conditional | Verified | Missing |
| 04Capacity basisAre critical supply windows verified rather than named? | Missing | Conditional | Conditional |
| 05Evidence threadCan the material records survive transfer, delay, or supplier change? | Missing | Verified | Conditional |
Verified means supported by a decision-grade record. Conditional means material evidence remains open. Missing means the dossier cannot support reliance on that claim.
Inspect all gates →Current status
Independence must be built, not claimed.
Siderian is designed for independent operation. That independence is not yet an accomplished fact: its assurance governance is being formed and no independent assurance authority has been appointed.
Until those controls exist, Siderian describes itself as designed for independent operation, not as an independent authority.
Assurance authority
in formation
Mandate and appointment criteria are being developed. No authority is appointed.
Independent technical review
in formation
Capability requirements are defined at a role level; appointments are pending.
Conflicts and recusal
proposed
Public policy outline published; operating procedure and register remain to be adopted.
Commercial separation
proposed
Target operating model published; legal and committee implementation remain in formation.
Founder · documented record
Demian Reidel
Demian Reidel is a physicist and economist whose career spans capital markets, central banking, public strategy, and nuclear governance. He founded Siderian to build the rules and evidence layer required for repeat nuclear deployment.
His role is systems architecture. Technical qualification, operating judgment, and assurance decisions must be independently governed.
- Scientific training
- Physics, financial mathematics, economics
- Institutional record
- Markets, central banking, public strategy
- Nuclear experience
- Board governance, not operations or AP1000 delivery
Private diligence
Request investor access.
A private diligence path for qualified counterparties. Confidential material is shared under NDA, on request.
No transaction terms are published on this site.