The Dunkelman Problem
Marc Dunkelman's central argument in Why Nothing Works is not that American government is corrupt or captured or ideologically broken. It's that it has been systematically reorganized, over fifty years of individually defensible reforms, into an institution that is structurally incapable of executing at scale. The interstate highway system could not be built today. Not because we lack the engineering. Because we lack the institutional architecture to make a decision and follow through on it.
The graphite supply chain problem is a case study in exactly this failure mode. The diagnosis was correct. The strategic vulnerability is real. The technology to address it exists, is cost-competitive, and was developed in American national laboratories with American federal funding. The timeline to first commercial production — for a technology already at TRL 5–6 — is two to four years with committed capital and a government offtake agreement. The tariff revenue being collected is more than sufficient to fund the transition.
None of it happened. Not because the money wasn't there. Not because the technology wasn't ready. Because the institutional will to sign the paperwork wasn't there. Because somewhere between the diagnosis and the solution, the American government ran into its own reorganized architecture and stopped.
This piece is about what that architecture needs to look like — not in the abstract, but in the specific. One proposal, for one supply chain, using existing legal mechanisms. Then every serious objection to it, answered. Then the objections that survive, stated honestly.
The Hamiltonian Premise
Alexander Hamilton's Report on Manufactures (1791) made the case that the federal government had not just the authority but the obligation to develop American industrial capacity — to identify strategic industries, protect them while they scaled, and invest in the infrastructure that made them viable. Not socialism. The recognition that certain coordination problems are too large and too long-horizon for private capital alone to solve. That logic underwrote the transcontinental railroad, rural electrification, the interstate highway system, and the semiconductor industry in the 1980s. None of those were purely private achievements. All of them required a government willing to make a bet, hold it, and follow through.
Both parties have forgotten this. The Democratic Party, captured by a coalition that reflexively distrusts industrial development, has ceded the Hamiltonian vocabulary to its opponents while stripping the investment mechanisms that would make it real. The Republican Party has adopted the vocabulary — "strategic industries," "domestic production," "supply chain resilience" — while delivering only the punitive half of the policy, the tariff without the investment. The result is a government that can identify the problem and impose a cost for it, but cannot build the solution.
The graphite supply chain doesn't need a new ideology. It needs someone to sign three documents: an offtake agreement, a loan guarantee, and a bond authorization. What follows is a specific proposal for what those documents should say.
The Proposal: American Materials Bonds
The federal government should authorize a new class of retirement savings instrument — American Materials Bonds — federally guaranteed, fixed-term bonds available through 401k and IRA accounts, with proceeds directed exclusively to domestic critical material production facilities operating under DOD offtake agreements. The mechanism combines three existing instruments: Treasury-backed savings bonds, the DOE Loan Programs Office guarantee structure, and DOD Defense Production Act offtake authority. No new law is required. What is required is a Treasury rulemaking, a DOE loan guarantee authorization, and a DOD procurement commitment — bureaucratic actions, not legislative ones.
Instrument: Federally guaranteed fixed-term bonds, available through qualified retirement accounts (401k, IRA, 403b). Not tradeable on secondary markets. Hold-to-maturity only — early withdrawal forfeits the federal interest guarantee and reverts to principal only.
Proceeds: Directed exclusively to domestic critical material production facilities (graphite, cathode active materials, separator manufacturing) operating under signed DOD or DOE offtake agreements.
Guarantee structure: The federal government guarantees return of principal plus stated interest at maturity. The guarantee is contingent on hold-to-maturity. No early redemption window. The government's contingent liability is predictable, schedulable, and never subject to bank-run dynamics.
Offtake prerequisite: No bond issuance authorized for a facility without a signed government offtake agreement. The facility must have a buyer before it has a builder. The bond finances construction of a facility whose revenue stream is already contractually committed.
Two Options. One of Them Gets Done.
Before explaining the bond structure, it is worth acknowledging that there is a simpler path — and being honest about why it probably won't be taken.
The graphite tariff collects an estimated $1.3–$1.9 billion in annual revenue. A first commercial-scale FJH or biographite facility requires $200–400 million in capital. One year of tariff collection funds the entire first facility, with money left over for a second.
The mechanism is a direct appropriation: Congress designates a portion of graphite tariff revenue to the DOE Loan Programs Office for critical mineral facility guarantees, combined with ARPA-E scale-up grants for TRL 5–6 technologies already in the pipeline. The administration announces a domestic graphite capacity initiative. Shovels go in the ground within eighteen months.
Why it probably doesn't happen: It requires Congress to appropriate designated revenue rather than letting it flow to deficit reduction. That means a budget fight. A whip count. A coalition that holds across two appropriations cycles. An administration willing to spend political capital directing tariff revenue away from the general fund. Every one of those steps is a place where the institutional capacity Dunkelman describes as missing gets tested. Option 1 is the better policy. It is also the one that requires the government to work.
The tariff revenue stays exactly where it is. Not a dollar is redirected. Not a budget line is touched. The general fund is unaffected. The deficit hawks have nothing to object to.
Three executive actions — a Treasury rulemaking, a DOE loan guarantee authorization, a DOD procurement contract — mobilize private retirement capital to build the same facility that Option 1 would have funded with public money. The government's only financial exposure is a contingent guarantee on bonds that, as argued below, is in all likelihood never triggered because the private market fills the facility's capacity before the government's offtake volume is ever called.
Why this is the realistic path: It costs the administration nothing visible. No new spending. No new taxes. No appropriation fight. No redirected revenue. The political exposure is a guarantee that probably isn't called, on a facility that produces a product American manufacturers are already desperate to buy. This works in any political environment — including one where the party in power has made fiscal restraint a centerpiece of its identity.
Option 1 is the better policy. It is faster, cleaner, and doesn't ask retirement savers to bear any duration risk. If the administration wanted to spend the tariff money on the problem the tariff was supposed to solve, this series would be shorter and the argument simpler.
Option 2 is the proposal that actually gets done. And in a government with the institutional capacity problems Dunkelman describes, the proposal that gets done is the better proposal. A perfect policy that requires Congress to work is worth less than a good policy that requires only three signatures. What follows is the three-signature version — with one more regulatory action that makes it urgent to start now rather than later.
The Demographic Window Is Closing
The capital pool available for long-duration industrial bonds is a function of the age distribution of current retirement savers — and that distribution is shifting fast in one direction. The baby boomer cohort is the largest concentration of retirement wealth in American history. The oldest boomers are 79 this year. The youngest are 61. The 5-year bond window for a 60-year-old closes in 2031. The 10-year window for a 55-year-old closes in 2036. After that, the money isn't sitting in retirement accounts anymore. It's in distributions, estates, and inheritance transfers that are far harder to mobilize for long-duration industrial investment.
Under the SECURE Act, inherited IRAs require full distribution within 10 years for most non-spouse beneficiaries. An heir who inherits a 401k today has a mandatory 10-year drawdown window — the money moves into consumption or conventional liquid investment almost immediately. It does not sit in long-duration industrial bonds. The mobilization window for this specific pool of capital is not open indefinitely. It is open now, and it is narrowing by year.
The fix is a fourth regulatory action: an IRS designation of a new qualified account category — American Materials Funds — as a sub-account within existing 401k and IRA structures, holding only federally guaranteed industrial bonds, and exempt from the 10-year mandatory distribution rule for inherited accounts.
The logic for the exemption holds on its own terms. The mandatory distribution rule exists to prevent retirement accounts from becoming perpetual tax shelters. An American Materials Fund is not a tax shelter. It is a designated industrial investment with a federal guarantee, a defined maturity, and a known payout schedule. The government already knows exactly when it pays out and exactly what it owes. There is no valuation ambiguity, no tax avoidance mechanism. The exemption isn't a loophole. It is recognition that a federally guaranteed bond with a fixed maturity is categorically different from an equity portfolio that could compound indefinitely.
The practical effect is significant in both directions. A 58-year-old who inherits a parent's 401k today must distribute it within 10 years and pay ordinary income tax on every distribution. If a portion of that account is in American Materials Funds, that portion carries to the heir with the bond terms intact — they collect the guaranteed return at maturity, pay tax then, and the capital stays invested in the facility for the duration it was designed to serve. The demographic cliff gets pushed out by a generation.
It also makes the bonds more attractive to the original investor. A 62-year-old considering a 10-year bond today might hesitate — what happens to the position if they die in year six? Under the American Materials Fund exemption, the answer is: it passes intact, the heir collects the guaranteed return at maturity, and the facility gets the full duration it needs. The investment decision stops being complicated by estate planning anxiety.
One IRS rulemaking. The demographic urgency problem, the intergenerational transfer problem, and the long-duration investor hesitation problem — all addressed in a single regulatory action that costs nothing and requires no appropriation.
The Yield Curve Is the Policy
The bond program is structured as a deliberate yield curve — longer duration gets a higher guaranteed rate, shorter duration gets a lower rate but attracts the capital-heavy cohort who need liquidity sooner. This is not a subsidy. It is a maturity premium, the same logic that governs every bond market in existence. The federal guarantee is what makes it accessible to retirement savers who would otherwise require a credit rating and a prospectus to evaluate the underlying project risk.
| Bond Term | Guaranteed Rate | Target Demographic | Typical Principal Available | Capital Profile |
|---|---|---|---|---|
| 5-year | 3.5% | Ages 40–55 — peak savings, pre-retirement window | $50,000–$300,000+ | High principal, modest rate — bulk of capital mobilized here |
| 10-year | 4.5% | Ages 30–45 — mid-career, long runway | $15,000–$100,000 | Medium principal, competitive rate — growth-oriented savers |
| 20-year | 6.0% | Ages 22–35 — early career, low principal but high duration tolerance | $2,000–$25,000 | Low principal, attractive rate — small capital individually, large signal collectively |
| Program goal: $2–5B mobilized across all tranches for first graphite facility network | One year of graphite tariff revenue would fund this outright | |||
The 5-year bond is the workhorse. A 48-year-old with $180,000 in a 401k puts $40,000 in at 3.5% — a guaranteed $47,800 at maturity, better than a CD, out before retirement is a real question. The 20-year bond is the statement: a 27-year-old with $8,000 saved puts $3,000 in at 6% and gets $9,621 at maturity at zero credit risk. Individually a rounding error. Collectively, across millions of early-career workers, a demand signal that tells private capital the program has broad public support — which is its own form of de-risking.
The Timeline Works
When the first 5-year bonds mature, the facility is not a speculative project. It is a producing asset with a government customer, generating revenue against a signed offtake contract. The federal guarantee is not being called to bail out a failed experiment. It is honoring a bond that financed a facility that worked. That is categorically different from the risk profile the word "guarantee" implies to most readers.
Where You Build It
The standard siting objection — NIMBY opposition, environmental review timelines — doesn't apply here. The ideal sites for FJH and biographite facilities are places where no one else wants to build, because the feedstock is waste. The best co-location is adjacent to the waste stream being consumed:
Municipal solid waste processing facilities. Already zoned industrial. Already permitted for waste handling. The graphite facility adds a conversion step to material that is already there and already being managed at cost. The neighborhood association that would oppose a new factory has no standing to oppose an upgrade to an existing permitted facility.
Tire stockpile and recycling sites. The United States generates approximately 300 million scrap tires annually. Many are stockpiled illegally or managed at public expense. An FJH co-location converts a liability into a feedstock. The permitting argument is reversed: you are solving a problem, not creating one.
Rust Belt brownfields. Former steel mills, chemical plants, and refineries — already zoned heavy industrial, already served by rail and power infrastructure, already on the EPA's remediation list in many cases. Brownfield redevelopment has federal fast-track permitting pathways specifically because redevelopment is preferable to continued vacancy. The congressional district that contains a former steel mill has been trying to attract exactly this kind of industrial investment for thirty years.
EPA Superfund sites with PFAS contamination. As documented in Part 3, Rice University's FJH process simultaneously destroys PFAS forever chemicals and converts the activated carbon into graphene. The EPA has a multi-billion dollar PFAS remediation obligation. A facility that closes a Superfund site while producing battery-grade graphite is not asking for environmental approval. It is providing one. The permitting conversation is entirely different when you show up offering to solve a problem the government has already agreed it needs to solve.
The hardest part of siting these facilities is deciding which existing problem to solve first.
First Attack: Why This Fails
That is the proposal. Now let's break it.
A federally guaranteed, fixed-rate bond with a defined maturity is, by definition, among the most prudent instruments available. The federal guarantee eliminates credit risk. The hold-to-maturity structure eliminates liquidity risk for the issuer and prices that risk explicitly for the participant. The rates proposed — 3.5% to 6% depending on duration — are competitive with prevailing Treasury yields plus a maturity premium. A plan fiduciary who excludes these bonds from the investment menu while offering equity funds with higher volatility and no federal guarantee has a more difficult fiduciary argument to make, not a simpler one.
But here's what no one in the policy debate has said out loud yet: the federal offtake agreement is probably never filled. Not because the program fails — because it succeeds too well for the government to be the interesting buyer.
The pricing structure is the mechanism. The government signs an offtake agreement at $4.50/kg — above the biographite minimum selling price of $3.30/kg, guaranteeing the facility's economics during construction and qualification. Private investors who want in early — bond subscribers, co-investors, OEM supply chain partners — get access at the Chinese pre-tariff market price of $4.20/kg. They get the cost parity price. Then the government's committed price sits at $4.50/kg, and anyone who shows up after commercial production begins pays at least $5.00/kg. First mover advantage is priced into the structure. The OEMs who want American graphite cheapest need to commit early.
Now consider what happens at year four when the first facility comes online. It is producing at roughly 3% of total North American battery-grade graphite demand — call it 50,000 tonnes against a market needing 1.5 million tonnes. The facility cannot fill every order in year one. It is physically impossible. So who gets first allocation? The early private capital partners and the OEMs who committed to offtake agreements at $4.20/kg. Ford, GM, Stellantis, Samsung SDI — any cell manufacturer with a North American facility has been watching a 160% tariff inflate their graphite costs for years. The chance to buy domestic material at pre-tariff Chinese parity, with supply chain security and no customs exposure, is not a hard sell. They will be in line before the facility ships its first tonne.
The government's $4.50/kg volume commitment sits at the back of that line by design. Private buyers get first dibs, because we actually want OEMs buying American parts — that is the stated goal of the entire policy apparatus that created the tariff in the first place. By the time the private demand is satisfied and the government's turn comes up, there is a reasonable chance the facility is already running at capacity against commercial orders. At which point the DOD procurement office can quietly note that its allocated volume has been deferred to the next production tranche, and no one is harmed. The guarantee was the insurance policy. The insurance was never claimed because the house didn't burn down.
The federal offtake agreement is not the exit strategy. It is the confidence signal that makes everyone else move first. That is precisely the kind of thing a government with a long time horizon and a credible balance sheet can do that private capital alone cannot — and it may cost nothing at all.
The mitigation is qualification staging. The offtake agreement should not activate full commercial volumes until the facility has passed an 18-month qualification period producing at 10% capacity against automotive specification. Bonds issued in year 1 finance construction and qualification. The federal guarantee backstops the bond regardless of qualification outcome — which is how the credit risk is properly allocated: the government bears the technology risk, not the retirement saver. This is the same structure used in DOE loan guarantees for first-of-kind energy projects. The participant gets their guaranteed return. The government absorbs the technical risk. That is a legitimate use of the federal balance sheet.
The broader industrial policy failure argument conflates cases where the government bet on the wrong technology with cases where it bet on the right technology and didn't follow through. Flash Joule Heating won a peer-reviewed publication in Nature and has been independently replicated. This is not a government picking a winner from a field of competitors. It is a government funding the commercialization of a technology that has already won.
Second Attack: What Survives
The defenses above are real. But intellectual honesty requires identifying what they don't fully answer.
Both surviving objections deserve the full answer, not a concession.
On the regressive design problem: the hold-to-maturity requirement that protects the federal guarantee does not have to be absolute. Allow early withdrawal with full return of principal and forfeiture of the federal interest guarantee only. A lower-income saver who hits a medical emergency or a job loss gets every dollar they put in, immediately — they simply don't receive the guaranteed return. They are not punished for needing liquidity. They are not rewarded for providing it. The federal guarantee structure is preserved because the government's contingent liability only triggers on bonds that were held to term, which are precisely the bonds that financed a facility that had time to be built and qualified. The regressive critique is not eliminated — the program still rewards those who can afford to wait — but it is substantially mitigated. Someone with $22,000 in a 401k can put in $3,000, know their principal is always accessible, and decide how much of their timeline to commit. That is not a trap. That is an option.
On the institutional capacity problem: the objection is correct in general and overstated in the specific case. Dunkelman's failure mode applies to projects that require sustained multi-agency coordination across decades, shifting political coalitions, and ongoing bureaucratic execution — the interstate highway system, the nuclear weapons program, the moon landing. This proposal requires three bounded executive actions: a Treasury rulemaking, a DOE loan guarantee authorization, and a DOD procurement contract under existing Defense Production Act Title III authority. After those three documents are signed, the government's active role is essentially finished. Capital mobilization runs through existing 401k infrastructure — the Treasury doesn't manage it, Fidelity and Vanguard do. Construction is done by private contractors on commercial timelines. Technology licensing is a commercial transaction between the developer and the facility operator. The government's ongoing obligation is to honor contracts it has already signed — which doesn't require institutional capacity, it requires not actively canceling them. That is a much lower bar than building the interstate highway system.
The Defense Production Act context matters here specifically. DPA Title III has a demonstrated execution track record — COVID vaccine manufacturing scale-up, domestic rare earth processing, semiconductor packaging capacity. The DOD procurement apparatus moves differently from civilian agency culture, by design and by necessity. The three-document structure of this proposal routes through the part of the American government that has retained the institutional capacity Dunkelman describes losing elsewhere. That is not an accident. It is a design choice in the proposal, and it is the right one.
Hamilton understood that the argument precedes the conditions. The Report on Manufactures was never fully implemented in his lifetime. The industrial policy it proposed took decades to become the reality it described. The argument was right before the political conditions existed to act on it. That is not a failure of the argument. It is the nature of ideas that are ahead of their moment.
The technology is there. The timeline is achievable. The financing structure is legal and workable. The sites are available and no one else wants them. Both surviving objections have answers that hold. The only thing missing is the will to sign three documents — and the political conditions that make signing them possible.
Industrial policy is not a speech about what you want to build. It is building it. When someone in American government is ready to build it, the blueprint is here.
← Part 3 — The Steelman and the IndictmentThe technology that could have closed the graphite gap was sitting in federally funded labs. This piece documents what was available and why it went unused. Continue Reading — Part 5 — What a $35,000 American EV Actually CostsAmerican Materials Bonds proposed. Now build the vehicle they'd fund. A ground-up BOM for a domestic EV — every component, every supplier, every cost. →
Source corrections, data updates, and substantive pushback welcome. If any cited source revises its conclusions, flag it here and the analysis will be updated.