01Two engines, and which one actually pays
The project has two revenue engines and they behave differently. The instinct is to treat the CO₂ side as a safe floor with product revenue as upside on top. The comparables do not support that reading.
The CO₂ engine does not carry the project. Every tonne captured earns §45Q, at the utilization rate where carbon goes into product and the sequestration rate where the balance goes into the well. On 850,000 t/yr that is roughly $66M a year for twelve years. Against capture blocks that cost about $1.0B all in, it services roughly a third of the capital. Run the CO₂ side alone and it never pays back on §45Q, and it still does not clear with a certified LCFS pathway and a carbon removal attribute stacked on top. Breakeven on the CO₂ blocks alone is about $300 per tonne of effective credit value, in a range of $254 to $353 across the capital comparables.
So the product engine is the condition, not the upside. The commercial strategy page documents the industry record, which is that producers with more than 100 t/yr of installed capacity book under $1M of annual revenue. This model therefore drives tonnage from a contracted ramp and sizes conversion capacity to follow it, so capital is never committed ahead of a signed contract. What it does not do is treat that engine as optional. Without a book of business the capture project does not finance.
02Mass balance
The carbon splits two ways. What is sold becomes graphene; the balance of the stream is injected. Both halves earn a credit.
| Line | Value | Basis |
|---|---|---|
| CO₂ captured | 850,000 t/yr | Biomass flue gas at Mesquite Lake |
| Conversion efficiency | 75% | Sizing model default; the range is 40 to 95 percent |
| Carbon yield | 0.2727 t C / t CO₂ | Stoichiometry, 12/44 |
| Physical yield cap | 173,864 t/yr | The most graphene the stream can produce at 75 percent conversion |
| CO₂ consumed as feedstock | 3.67 t per t sold | Drives the §45Q utilization tier |
| CO₂ injected | balance of the stream | Drives the §45Q sequestration tier |
| O₂ coproduct | 2.67 t per t C | Sold at the merchant rate on the share the market absorbs |
| Electrical load | 6.0 MWh/t + 1.0 MWh/t CO₂ | Electrolysis plus capture parasitic, behind the meter |
Conversion consumes a minority of the stream in every case. At the phase 1 book the plant converts about 183,000 t of CO₂ into product and injects about 667,000 t, so the credit stack stays dominated by the sequestration tier and shifts toward utilization only as tonnage grows. The injected balance is what earns the sequestration tier and the removal attribute, which is why the sequestration floor is a design constraint rather than a leftover.
03Three cases
The constraint that decides the shape of this business is that world graphene demand is about 3,800 tonnes a year and this plant's carbon yield is forty six times that. No case sells one product into one market, so each case is a book of business across ten segments rather than a tonnage forecast.
| Case | What it assumes | Full book | Blended price | Multiple of world demand |
|---|---|---|---|---|
| A · Anchor book | A short list of anchor contracts, weighted to the segments that qualify fastest. Enough to prove the process and lock the §45Q utilization tier. | 25,000 t/yr | $38,619/t | 6.6x |
| B · Phase 1 as scoped | Five to eight anchor customers across every tier. Battery-grade graphite and concrete admixture carry the volume, specialty carries the margin. | 50,000 t/yr | $31,033/t | 13.2x |
| C · Full conversion | Every segment at or near the volume the market map says it can absorb. This is what full conversion of the available carbon requires, not a demand forecast. | 150,000 t/yr | $14,127/t | 39.5x |
Price is an output here, not an input. Each segment carries a price today and a price at full-conversion volume, and the model interpolates between them in proportion to how much of that segment the project is filling. Case B lands at $39,300 per tonne at its published mix, which is the same blended figure the sizing model reaches independently. That agreement is a check on the mix, not a coincidence.
The book of business is editable. The live product mix model lets you reshape it segment by segment and watch the capital, the credits and the returns move with it.
04Results
All three cases, 25 years, all equity, computed in the workbook and reproduced line for line by the live model page.
| Metric | A · Anchor book | B · Phase 1 | C · Full conversion |
|---|---|---|---|
| Full book | 25,000 t/yr | 50,000 t/yr | 150,000 t/yr |
| Peak capital deployed | $1,270M | $1,436M | $2,096M |
| Peak cash shortfall | $982M | $918M | $724M |
| Revenue, year 10 | $1,167M | $1,752M | $2,314M |
| EBITDA, year 25 | $742M | $1,179M | $1,531M |
| Project IRR, unlevered | 38.5% | 51.5% | 70.0% |
| NPV at 15 percent | $1,833M | $3,196M | $4,467M |
| Payback | Year 5 | Year 4 | Year 3 |
| Credit stack over the credit period | $778M | $758M | $679M |
Underwrite case A. The returns rise with volume because the fixed CO₂ blocks are already paid for in every case, so incremental tonnage carries only its conversion capital. That is a real effect, and it is also why the higher cases should not be read as the plan: case C requires every one of ten segments to reach the top of its published absorbable range, and each of those segments carries its own qualification cycle measured in years. Case A clears a fifteen percent hurdle on a book small enough to name the customers.
05Capital
The CO₂ blocks are built once for the whole stream. Conversion capacity is added only as tonnage is contracted.
| Block | Cost | Timing rule |
|---|---|---|
| Capture island and injection wells | $800M | Built once, 60 percent in year one and 40 percent in year two. Sized for the full 850,000 t/yr stream, so it does not scale with graphene demand. |
| O₂ cryogenic recovery | $100M | |
| 220 kV substation | $80M | |
| Oxygen recovery | $100M | |
| Admin, control, safety | $25M | |
| Kiln potlines | $5,222 per annual tonne | Added only when capacity is contracted. Derived from the sizing model's $940M, $110M and $30M for a 180 kt/yr build. |
| Purification and spec | $611 per annual tonne | |
| Rail spur and loadout | $167 per annual tonne | |
| Contingency | 10 percent | On everything above, per the sizing model's own stack |
This is the capital discipline the strategy page asks for, expressed as a formula. Capacity built never falls and never runs ahead of the tonnage under contract, so the capital that produced the industry's defining failure, kilns installed against demand that never arrived, cannot be spent by this model.
06What has to be true
Each of these is a condition the model depends on, with the consequence if it fails.
- §45Q survives review at both tiers. The utilization tier requires an audited end use and a lifecycle analysis; the sequestration tier requires the injection permit and an MRV plan. Remove §45Q entirely and the base case loses roughly $66M a year for twelve years. A tax opinion comes before any of this is relied on.
- The injection path is permitted. The sequestration tier assumes the balance of the stream is injected on site. If the permit path fails, that CO₂ has to be vented, the sequestration credit disappears, and the project becomes a graphene business with a capture island attached.
- Power stays behind the meter. The model runs power at $45/MWh. At $90/MWh the base case loses about $41M a year at year-25 volumes, and considerably more in the capacity cases where electrolysis load dominates.
- Offtake is signed before capacity is built. Capacity in this model follows contracts by construction. If capital is committed ahead of offtake, the model no longer describes the project being built.
- Realized price holds on the declining curve. Price is the dominant driver once volume is meaningful. A 25 percent shortfall against the curve takes roughly $585M off year-25 EBITDA in the anchor case.
- The plant basis is settled. This model uses 850,000 t/yr of captured CO₂, which is the instruction. The sizing model page runs 840,000 t/yr, and the energy-island page describes a different biomass configuration. One project basis needs to be canonical before the model is taken to a counterparty.
07The workbook
The model is a live-formula workbook, not a picture of one. One input tab drives everything.
| Tab | What it holds |
|---|---|
| Assumptions | Every input, each with its source. The case selector is one cell. Change an input and the whole workbook reflows. |
| Revenue Ramp | The three tonnage cases, the declining price curve, and the yield-cap check. |
| Mass Balance | Carbon, CO₂ utilized and injected, O₂ coproduct, power and continuous load. |
| Capacity & Capex | Capacity ratchet, capital spend by block, cumulative capital. |
| Credits | §45Q at both tiers over the twelve-year period, transfer realization, O₂ sales. |
| Opex | Power, conversion, injection, consumables, labor, royalty, formulation, selling and administrative. |
| P&L · Cash Flow · Returns | 25 years, all equity, with IRR, NPV, payback and peak capital. |
| Sensitivity | Price, §45Q denial, power cost, and the case where graphene never sells at all. |
| Sources | Every input traced to a published page, or marked as introduced by the model. |
Download the workbook: SAVRN_Imperial_Valley_CO2_to_Graphene_Financial_Model.xlsx. Live formulas throughout; the case selector is Assumptions!B4.
Built from the sizing model and the commercial strategy. Inputs marked "model" in the Sources tab are introduced here and appear on neither page: technology royalty, selling and administrative cost, formulation and logistics, the merchant O₂ share, injection cost, and the discount rate.
Basis and disclosures
- Purpose
- This page is prepared by SAVRN for evaluation by the recipient. It is not an offer to sell or a solicitation of an offer to buy any security or interest, and it is not investment, legal, accounting, or tax advice. Any offering would be made only through definitive documents.
- Forward-looking content
- Scenarios, model outputs, ramp schedules, and pricing bands on this page are estimates built on the assumptions and sources shown. Actual results will differ. SAVRN undertakes no obligation to update this page as conditions change.
- Tax credits
- Amounts shown for §45Y, §45Q, §48, and §48E are estimates of eligibility under current statute and guidance. Each depends on facts not yet fixed, including prevailing wage and apprenticeship compliance, begin-construction date, domestic content, energy-community designation, prohibited-foreign-entity sourcing, and placed-in-service timing. Transfer pricing under §6418 is market-dependent. A tax opinion is required before any amount on this page is relied on.
- Model outputs
- Outputs move with the inputs on this page. EBITDA is stated before depreciation, amortization, interest, tax, technology license and royalty payments, selling and administrative cost, outbound logistics, and product formulation cost. Volumes assume offtake at the stated prices; the industry record on that point is set out in the commercial strategy page.
- Estimate class
- Capital and operating figures are AACE Class 5, order-of-magnitude (−50% / +100%), based on parametric analogies rather than engineered quantities. They are not a FEED or FEL-1 estimate.
- Third parties
- Companies, products, contracts, and prices named on this page are drawn from the public sources listed and are used for market reference. No affiliation, endorsement, or supply relationship is implied except where a signed agreement is stated.
- Data as of
- August 2026, from the sources listed on this page.
- Distribution
- This page is served without search indexing and is shared by link. Treat it as SAVRN business information and share it only with parties who need it.