Hyperscalers, host communities, and the deposit that changes the negotiation.
Two headlines are circulating right now, and both describe the same machines. The first comes from Sam Altman's September 2024 essay "The Intelligence Age," which declares that "a defining characteristic of the Intelligence Age will be massive prosperity." The second is Kyle Chayka's October 2025 New Yorker column asking, "Will A.I. Trap You in the 'Permanent Underclass'?" One promises abundance for everyone. One warns that abundance will be hoarded. Neither is fringe. One comes from inside the industry building the data centers; the other comes from the press covering it. Both are live bets on what the next decade does to the people who live near the buildout.
A school gym with folding chairs. The zoning hearing runs long. A resident waits her turn, steps to the microphone, and asks the planning board and the company's site team the same question: what's in it for us? She has heard the prosperity pitch. She has also read the underclass warning. Nobody at the table answers her, because nobody at the table is sure which headline she is living in. The question hangs and the meeting adjourns. The next hearing fills more chairs.
Policy analyst Michele Wucker has a name for a threat that is enormous and visible, charging straight at you while everyone looks away: a gray rhino, the coinage from her 2016 book of the same name. Anti-data-center sentiment is that rhino. Every operator can see it coming. Town by town, hearing by hearing, the resentment is gathering speed in plain view, and the prevailing response has been to watch it run.
So who actually decides which headline wins? The resident assumes the company does. The company assumes the regulators do. Meanwhile the moratorium vote happens in one building, the siting decision in another, the cash decision in a treasury department a thousand miles from the gym. Each party is answering the podium question without realizing the others are in the same conversation. Someone is deciding America's data-center decade. Nobody has named who.
Strip the rhetoric off a contested data-center siting and two honest inventories remain. The grievances of host communities hold up under scrutiny. So do the builders' reasons for building. Any strategy that treats either side as confused will fail in the room where it matters, the town hall, and so the working baseline has to be one both sides of the table would sign.
Community objections sort into two families, and the distinction matters because the two families demand different answers.
The first family is economic extraction. Construction payroll spikes, then drains away when the trades move to the next county. Tax abatements run for years before the school district sees its share. A running facility employs far fewer people than the construction phase suggested, so the permanent-jobs promise lands as anxiety instead of reassurance. And the company's cash, the largest pool of money the project ever generates, banks somewhere else entirely. Local lenders gain no new capacity to fund the businesses the boom was supposed to seed. Residents watch value get created on their land and deposited in someone else's.
The second family is physical and procedural. Water draw in places where wells already run short. Grid strain and interconnection queues that crowd household demand — and a fight over electricity rates that the evidence leaves genuinely contested, though towns argue it as settled. Noise, light, and diesel generator testing at all hours. Health fears voiced in plain language at podiums, including the word leukemia. Farmland converted and neighboring property values in doubt. Running through all of it is the procedural wound: projects assembled behind NDAs and shell entities, so the town learns the buyer's name after the land is optioned. Residents experience that secrecy as betrayal before a single foundation is poured.
The families need separating because they fail differently. Physical and procedural grievances can be addressed in permits and ordinances. Economic-extraction grievances concern where money flows, and no permit condition touches them.
The other side of the ledger is just as real. Compute demand is genuine and sustained, and the facilities will be built somewhere; the open variables are where and on what terms. Construction trades on these sites are well paid, with electricians earning meaningfully above comparable work. Substations and fiber outlast the project that paid for them; so do the road upgrades. A community that captures the build on good terms acquires infrastructure it could never have financed alone.
Sit through a hyperscaler community-relations briefing and a county public-comment session and one phrase surfaces in both rooms: what's in it for us? That shared question comes with a shared inventory. Community banks are the institutions residents already trust with their own money. Schools are the first place either side points when asked what prosperity should fund. Main Street small business is the test both sides apply to whether a boom was real. Both rooms distinguish money that circulates locally from money that leaks out, and both ask what remains in year ten, after the cranes are gone. Both rooms agree these things matter. The disagreement is over whether the project delivers them.
The outcome record already runs the full spectrum. On one end sit negotiated agreements with enforceable terms: agreements that cap water use and tie noise limits to measured baselines rather than the operator's own estimates. On the other end sit dozens of moratorium and restriction actions by towns and counties across the country. What divides the two ends is preparation. Communities that arrived with terms drafted got terms. Communities that learned the buyer's name from a leaked filing reached for the only instrument left to them, the no.
That record settles the shape of the question. The contest at any given siting is over terms: which grievances get answered and how the commitments are enforced. Both inventories are legitimate, and the agreement that honors both is the one that survives the public meeting.
The Prosperity case is smaller than the pitch deck says and stronger than skeptics expect. The mechanism has a 17-year track record as StoneCastle reports it; the first-customer scenario is a model. Three kinds of figures run through the case: numbers verified against public sources, numbers the client supplied, and numbers produced by an illustrative model.
Forecast Hyperscaler data-center investment is forecast to approach $3 trillion globally over the next five years. That is projected spending; nobody has committed it.
Verified The committed layer is smaller and public: the four largest hyperscalers have guided roughly $600 to $725 billion in capital expenditure for 2026 alone. Both numbers are real. Only the second is money on the table.
Verified A typical operating data center employs roughly 50 to 200 permanent staff. Construction carries the headcount: 800 to 1,500 or more workers during the build phase, with data-center electricians earning about 30 percent above comparable trades. Fewer than 200 ongoing jobs is the honest figure, and any pitch that implies otherwise will be fact-checked at a podium.
Client-stated StoneCastle reports $233 billion in cumulative deposits placed since 2009, across a network of more than 1,000 community banks, 850-plus active today. Mechanics: large balances split into sub-$250,000 increments across the network, producing FDIC insurance up to $100 million per legal entity at the same yield, with daily liquidity. Network economics run on a spread of 10 to 50 basis points, per StoneCastle's materials. The sub-$250,000 redistribution mechanic itself is standard FDIC practice, used by reciprocal-deposit networks for over a decade.
Client-stated and illustrative Google operates an estimated 162 to 181 US legal entities. The bank count per county is a client working assumption, not a sourced figure: the call used roughly nine banks in a typical county footprint. At $250,000 of insured capacity per tax ID per bank, the model assumes 10 entities placing deposits across 12 banks in each of 50 data-center counties: roughly $1.5 billion in deposits. Move to 20 entities at the nine-bank working assumption and the figure is roughly $2.2 billion; hold the 12-bank assumption and it approaches $3 billion. Downstream, the same model projects on the order of 13,000 small-business loans and 40,000 to 45,000 jobs. Every one of those figures is a scenario output. None is a commitment or an observed result, and the local-GDP projections built on them inherit the multiplier correction below.
Client-stated Google spends about $27 billion a year on sales and marketing. That line sizes what Google already pays for visibility and reputation. The deposit program's cost at the cash-management level is zero: same yield, same liquidity, same insurance.
Two figures from the original deck do not survive sourcing.
The 2.4x local multiplier is unsupported as a precise number. AMIBA and Civic Economics retail studies put local recirculation in a band of 2x to 4x: money spent or deposited locally recirculates two to four times. Built on a band instead of a point, the $4.8 billion local-GDP projection becomes a range, and an illustrative one.
"Eighty-five percent of corporate cash sits in a few institutions" is also unsupported. AFP's corporate liquidity surveys put roughly 80 percent of short-term corporate cash in bank deposits, money-market funds, and Treasuries, and FDIC data put the four largest US banks at about 30 percent of US deposits. Concentration is real at those magnitudes; the original phrasing overstated it.
Local recirculation and cash concentration both hold at honest scale; the precise figures did not.
Program status, per the 2026-06-09 call Three plain facts about the program itself:
| Item | Status |
|---|---|
| Named corporate adopters to date | Zero |
| Impact-measurement layer | Not yet built |
| Executive sponsorship | Not yet secured |
Zero adopters, no measurement layer, and no executive sponsor weigh as heavily in the arithmetic as the deposit math. A case built on a verified $233 billion mechanism, a verified $600-billion-plus capex year, a conceded jobs gap, and a modeled first-customer scenario in the low billions still clears headline scale. It clears it with those liabilities on the books, and only the honest accounting survives a town hall or a treasurer's diligence.
No company has run a larger live test of the standard corporate playbook against community resistance than Amazon with HQ2. The withdrawal from Long Island City is the cleanest public record of how each instrument performs under pressure, and every data-center county commission now in session is replaying the same script with smaller numbers.
By the public record of the announcement and its collapse: Amazon announced HQ2 in Long Island City in November 2018 with roughly $3 billion in city and state incentives attached and a promise of 25,000 jobs. Most of that $3 billion was forgone future tax revenue, money that would exist only if Amazon came. Amazon never said so in language a borough could repeat. Representative Alexandria Ocasio-Cortez asked why the $3 billion couldn't fund subways and schools instead, a question that was arithmetically confused and politically devastating, and opponents filled Amazon's silence with the simpler story: a cash giveaway to a trillion-dollar company. The simpler story won. Amazon withdrew in February 2019, losing an arithmetic argument it never made. The lesson generalizes: incentive math the company declines to translate becomes the weapon against it. People don't get the math, and a company that leaves the math unexplained has chosen the giveaway narrative on the community's behalf.
Hyperscalers fund some of the largest sales and marketing operations in corporate America. Against data-center negativity, that spend currently buys nothing. No campaign or prepared asset exists for the moment a resident stands up and asks what the town gets. The podium question is specific, and an ad answers a different question. The distance between the scale of the spend and the silence at the podium measures the failure: the machinery for telling stories at scale is already built and funded, and the one story it cannot tell is the one a county commission wants to hear, because that story requires money inside it.
Philanthropy fails structurally. A community grant is revocable and discretionary, and it arrives after the siting decision is already made — through the same opaque channels that produced the secrecy grievance in the first place. A town that learned its new neighbor's name only after the land closed is then asked to trust a gift from the same negotiating table. Read from the podium, the grant is reputation laundering: proof the company knows it owes something, paired with proof it intends to decide unilaterally how much and on what timetable.
The force that creates demand for a prosperity story is the same force that punishes a fake one. The resident at the podium experiences the company physically — water drawn from the same aquifer, land rezoned at the end of the street — so a response made only of claims gets tested against lived evidence, and lived evidence wins. A prosperity commitment that ships as communications lands as bread for the crowds, and the audience already knows the genre. The trap closes from both sides. Saying nothing leaves the giveaway narrative standing; saying something without money behind it confirms the narrative.
The town-hall refrain persists for a structural reason: no instrument any of these companies currently operates routes capital to the person asking the question. PR routes attention. Philanthropy routes discretion. Incentives route forgone taxes through municipal ledgers no resident reads. Until something routes money to where the grievance lives, "what's in it for us?" stays open, and every answer made of signaling sharpens it.
A hyperscaler's operating cash will sleep somewhere tonight. It sleeps every night, in the billions, and right now it sleeps in a handful of money-center institutions with no connection to the counties where the company pours concrete. Routing that same cash into the community banks of its host counties converts an unavoidable treasury decision into the negotiation instrument every prior playbook reached for and never had. Same cash, better outcome for the CFO. Your own team named the stakes on the June call: which headline wins is a capital allocation decision.
Three facts make the case. First, the cash already exists and must be parked; nothing here proposes new spending. Second, distributing it in sub-$250K increments across FDIC-insured community banks lets it sleep in the host county at comparable yield and same-day liquidity, with full FDIC coverage on every dollar. Third, the move costs the depositor almost nothing at the cash-management level; the treasurer's checklist comes back clean. What changes is the address.
That address is the point. A hyperscaler already treats land, power, fiber, and cooling as infrastructure choices, negotiated parcel by parcel and county by county, with teams of specialists assigned to each. Cash placement belongs in the same category and currently gets none of the same attention. Where your cash sleeps is an infrastructure choice: a zoning-grade decision, made today by default in a treasury back office, available to be made by design at the negotiating table. A company that arrives in a county with its deposits already routed locally walks into the town hall holding an answer instead of an apology.
The instrument is built to be held from both sides, and the second side is part of the design rather than a leak to be managed. In a CFO's hands it is a treasury action with a community story attached. In a CMO's hands it is the one response to data-center backlash that involves money actually moving rather than messaging about money. In a town's hands the same sheet becomes a demand: if Meta is coming, why aren't they depositing in our community bank? We want that question asked at podiums. A demand the company can meet without disturbing yield or liquidity is the cheapest concession in the entire siting negotiation, and the company that volunteers it before being asked owns the headline instead of chasing it.
The deposit answers one community concern, the charge that the money leaves town, and only that one. A company that routes its cash locally still has to win every other argument on that argument's own terms.
The mechanism underneath is proven: reciprocal-deposit networks have moved cash this way at hundreds-of-billions-of-dollars scale for more than a decade. The credibility infrastructure around it is still being built — measurement and attribution are commitments today, not yet running systems. We would rather you hear that from us now than from a skeptical reporter later. The instrument works the day a treasurer signs; the proof that it worked, the kind a CMO can put on a stage, takes that build-out. The gap dictates the order of operations: sign the treasurer first, build the proof second.
The creative work already exists. Four artifacts came out of the June 9 working session and the draft copy behind it: a pitch addressed to the AI companies as a class, a pitch addressed to the data-center builders, a news-segment script set in Jackson County, West Virginia, and a future-state press release written in Google's name. Each one carries something the other three cannot. The umbrella pitch carries the thesis; the builders pitch, the mechanism. The script is where the emotion lives. Proof, such as the program has, comes from the release. Each artifact needs one rewrite before its hardest audience, which in one case is Fiserv legal and in another is a microphone at a town hall.
Track A speaks to Google, Meta, Anthropic, and Amazon at the highest altitude any of the four artifacts reaches. Its opening device, two mocked-up headlines promising extreme prosperity and a permanent underclass, sets the size of the decision before any product appears. The builders pitch starts at the county line; track A starts at the species.
That altitude is also its soft spot. Arun's caution from the call stands: Google is not yet hated the way data centers are. The two-headline premise describes where AI-company sentiment is heading, while track B's premise is already standing at microphones. A pitch that opens on crisis the audience does not yet feel reads as melodrama, and a CMO who feels no fire will not buy an extinguisher.
So track A should open conversations and never close them. It earns the meeting; track B earns the deposit.
On the page, that means track A runs as the door-opener spread and hands off to the builders pitch within two pages. Any line treating AI-company resentment as a present condition rather than a direction of travel comes out.
Track B is the only artifact written for both readers at once. The CFO gets the cash-management case in four lines: insured, competitive rate, liquid, no new risk. The CMO gets the placement logic: the cash moves to banks near the data centers, where the company's reputation problem actually lives. That pairing exists nowhere else in the set.
Its weakness shows up in a vocabulary count. Every noun in the current draft belongs to the treasurer: deposits, liquidity, rate, insurance, multiplier. The words spoken at podiums, water, trucks, the grid, the school, appear nowhere. The CFO pitch and the podium never meet on the page, which means the CMO, the reader who must answer the podium, has nothing here to say to a town. The dollar mechanics and the lived grievance have to share sentences, since they already share a geography.
The "How it works" block carries the rewrite: every mechanism clause should end on something a resident can see, the deposit funds the loan, the loan reopens the storefront, the lending capacity makes the school bond bankable. Every sentence carries both halves.
The script is the only artifact that makes anyone feel the flip. Its structure is a tale of two towns compressed into one: the same county on two timelines, hostile meeting in the cold open, body shop and elementary school by act two. That changing-timelines turn is built for clipping, and the Robert Reich doodle register the team named as the tone reference fits it. Nothing else on the table produces a shareable forty seconds.
Two problems live here, and they are the engagement's heaviest. First, the format. A fabricated news segment, with invented quotes attributed to a named real company, is already the sticking point with Fiserv legal, and the current draft hangs its entire defense on one endnote that no clip will carry. The format's greatest strength and its legal exposure are the same property. Second, the resolution. The cold open's anger is about the water draw, the truck routes, the strain on the substation; by the closing aerial, a deposit has answered all of it. A financial instrument resolves the money grievance and touches none of the physical ones, and the script's overclaim is precisely the line opponents will screenshot.
The fiction has to declare itself inside the frame: a persistent on-screen scenario marker and a dated "from 2031" device, never a single endnote. The correspondent gets one line conceding what the deposit does not fix. And the new timeline should land in concrete detail a camera can hold, a science room, say, or the county road finally repaved, with Diana's credit-union warmth in the banker's mouth instead of sentiment in the narrator's.
The release does the one thing the program has lacked: it puts a name, a place, a bank count, and a dated announcement on what has been a concept. As the named-prospect artifact aimed at Google, where a real conversation already exists, it is the closest thing to evidence the pipeline can produce before an actual signing. Its projections are scenario math, labeled as such, and must stay labeled in anything that ships.
Two of its most natural lines cut both ways. "Costs you nothing" is the whole point to a CFO and an admission to a town: read aloud at a hearing, it proves the company conceded nothing. "Daily liquidity" reassures the treasurer that the cash is never trapped, and tells the town the money can leave overnight, that the commitment is revocable at will. Both lines are accurate. Both will be quoted by the wrong reader.
A unilateral corporate announcement invites the question of what the town agreed to. A release written as Google signing the financial chapter of a community agreement, with the county and the banks named as parties in the lede, answers it before it is asked. "Costs you nothing" moves into CFO-facing materials only, and town-facing copy pairs every liquidity mention with a stated commitment: a deposit term with a renewal date, backed by a published local-lending report.
None of the four changes touches strategy; each is a day of rewriting. The script's format question is the only one with a clock on it, because it is already sitting open on a desk at Fiserv legal.
Three parties surround every data-center vote, and none of them sits with the town.
The incumbent deposit network has the rails. IntraFi runs a placement network of more than 3,000 banks, by its own published count, and can already move corporate cash into community banks at scale, yet it has no community-facing layer, no artifact a county commissioner has ever held, and no reason to build one, because its customer is the depositor. Hyperscalers have community-affairs teams and grant programs, with no financial instrument behind either. Towns have land-use lawyers and Community Benefits Agreement precedent that already reaches data centers directly — Lancaster, Pennsylvania's agreement binds data-center owners to roughly $20 million in community contributions alongside water restrictions and noise limits — and yet when negotiations reach the financial pages they have nothing to ask for beyond taxes and fees, because no one has told them a capital demand is theirs to make. The town side of the table is empty. The one-sheet that fills it is the product's distribution channel and its credibility proof. For the town, it is the route to a siting it can call a win.
A town, or a coalition of towns negotiating as one, puts six terms on the table before the vote.
The keystone is a non-substitution clause. The deposit covenant sits alongside water-use caps, noise and light limits, ratepayer protection, and clean-energy commitments, and it buys none of them off. The deposit answers the extraction grievance alone; a town that accepts a larger balance in exchange for relaxed water terms has been bought, and the sheet says so on its face. That clause protects the negotiating town, and it protects the depositing company too, since a deposit that visibly purchased silence would curdle into the corporate-signaling backlash already waiting for it.
The same pre-vote window covers where and how the facility gets built. Distance from homes and wells. Campus configuration and acreage. The point of grid interconnect and who pays for it. The water source, including whether cooling runs on municipal supply or closed-loop systems. Each is negotiable for a coalition that arrives organized with terms in hand, and fixed for a town that learns the details after the rezoning notice.
One test decides whether a siting cleared the bar, and the dashboard from the demand list already runs it: audited outcomes in that county, after the fact, measured against the town's own terms. Projections are excluded entirely; a forecast counts for nothing. A facility that passes gives the company a proof no advertisement can match. A facility that fails hands the town its enforcement case.
We expect the sheet to do its own marketing. A resident holds it up at a town hall and asks why the company building next door keeps no balance at the bank on Main Street. Neighboring towns in the same utility corridor negotiate the same terms together, so the company cannot shop for the unprepared town one county over. AI assistants, seeded with the demand list, surface the deposit term whenever anyone asks what a data center owes its neighbors. And the trades channel pulls the company toward local payrolls on its own: water cooling and closed-loop retrofits are specialist work, and the contractors who do it are local. Every venue where the sheet appears is a sales conversation nobody at Fiserv had to schedule. The network incumbent could match the rails tomorrow. It cannot pick up the town side of the table without becoming a different company, and that vacancy is the position on offer.
The four ways this play dies are already visible. None requires bad luck. Each is a specific behavior with a specific defense, and in every case the defense costs less than the failure it prevents.
The deposit answers the money grievances: the dollars that leave town while benefits never arrive, the community bank that watched the construction boom from the sidelines. It has nothing to say to the second family, the physical and procedural grievances inventoried earlier. The fastest credibility kill available is an artifact that lets the deposit claim standing on that second ground. A resident asking about her children's sleep or her well water is asking a question money does not answer, and the moment a one-sheet implies otherwise, the deposit stops reading as a remedy and starts reading as a payoff.
A deposit that arrives after the siting vote, revocable at the depositor's discretion and announced through the same opaque channels that produced the NDA grievance, is bread for the crowds — the greenwashing trap closing right on schedule. It inherits every liability of the playbooks it was built to replace, because it places the money on the company's timeline rather than where the grievance lives.
The future-state press release and the news-segment video script are the strongest creative assets in the package and the largest current legal exposure. The fake-article device is already the biggest sticking point with Fiserv legal and copy review. That is true today, before any external reader has seen a draft. A speculative artifact that a reader could mistake for reporting hands opponents a fabrication story and puts the Fiserv relationship at risk before the first dollar moves.
The program status table above stands unchanged: no adopters, no measurement layer, no sponsor. A town playbook that promises an audited dashboard which does not exist converts the credibility asset into a liability the first time a town clerk asks to see the numbers. The deposit's whole advantage over the incentive playbook is that it is verifiable. Shipping the promise before the verification exists surrenders that advantage in public, to the exact audience the play was built to win.
Together the four disciplines are cheap. A clause that rides along with every artifact. A sequencing rule on when the money lands. A label on a creative device. A gate on the release calendar. The four failures all destroy the same thing: the credibility that separates a covenanted deposit from a press release. The asset survives on discipline that costs sentences and sequencing. It dies on shortcuts that cost the play.
The next 30 days are a sequencing problem, and the first move is already known. Google goes first. Josh Siegel's relationship with Neil Wilcox is live, and an active discussion involving Google and a Native American bank means the program has, for the first time, a prospect with a name attached. Every other move on the calendar either feeds that approach or protects it.
Owner: Josh Siegel, with Shur supplying the artifact. The future-state press release, repaired to specification, is the document that goes in. The Wilcox relationship sets the meeting; the Google and Native American bank discussion gives it a concrete frame to land in. Josh takes one conversation with a real counterparty and a finished artifact in hand; no dates or commitments are promised on Google's behalf.
Owner: Josh, inside Fiserv. Josh secures executive sponsorship first, then commissions the measurement specification. Town-facing material ships only after both. The sponsorship-before-publicity discipline becomes a dated commitment: both actions land on the calendar inside the 30 days, with the gate written down so nothing town-facing slips past it.
Owner: Shur, with Fiserv legal. Shur executes the four named repairs now and decides the format question on the press release up front, with Fiserv legal in the room: label it plainly as a scenario, or cut it. A decision either way beats a draft that sits in legal purgatory while the Google window is open. No artifact reaches Google or a town in its current state.
Owner: Shur drafts; StoneCastle selects the market. The one-sheet answers the town-hall demand list, item by item, in the language a resident would use at a podium. StoneCastle picks one real community-bank market near an announced site, and we test the one-sheet there once, taking honest notes on what landed, before any wider distribution.
Owner: jointly commissioned; we scope it. Is there a distance and a configuration at which a hyperscaler's arrival is net-positive for a town by default? Nothing in the record answers that today, and no one selling the deposit mechanic should pretend otherwise. The study converts the open question into a deliverable with a deadline. Until it reports, the question stands.
We ran every numerical claim from the 2026-06-09 call and the draft creative copy through a verification pass before drafting. Verdicts below; the corrected phrasings appear in Section 3.
| Claim as received | Verdict | Phrasing adopted in Section 3 |
|---|---|---|
| ~$3T hyperscaler investment in US data center buildout | Partially verified | Restated as a global, five-year forecast; the verified near-term figure is big-4 2026 capex guidance of roughly $600-725B |
| Operating data center employs ~50-200 permanent staff | Verified | Stated plainly, paired with the verified 800-1,500+ construction workforce |
| Data-center construction electricians earn ~30% above comparable trades | Verified | Stated plainly |
| StoneCastle: $233B cumulative since 2009, 1,000+ banks, up to $100M FDIC-insured per legal entity, same yield, daily liquidity | Client-stated | Attributed to StoneCastle's materials; figures carried as StoneCastle reports them |
| Google scenario: $1.5-2.2B deposits, ~13,000 loans, 40-45K jobs, ~$4.8B GDP | Illustrative | Framed as scenario modeling ("the model assumes"), with inputs attributed to the 2026-06-09 call |
| 2.4x local multiplier | Unsupported as a precise figure | Replaced with the 2-4x local-recirculation band the literature supports |
| 85% of corporate cash concentrated in a few institutions | Unsupported | Replaced with ~80% of short-term corporate cash held in deposits, money funds, and Treasuries; the top-4 banks hold roughly 30% of US deposits |
| ~9 banks per county average | Unsupported | Carried as an explicitly unverified client working assumption; the verifiable context is roughly 4,500 FDIC-insured institutions across ~3,100 counties, unevenly distributed |
| Google spends ~$27B/yr on sales and marketing | Client-stated | Consistent with Alphabet's reported sales and marketing expense; carried as client-stated |
| Amazon HQ2: ~$3B incentives, 25,000 jobs, Nov 2018 announcement, Feb 2019 withdrawal | Public record | Widely reported figures from the announcement and withdrawal coverage |
| IntraFi network of 3,000+ banks | Company-published | Carried with in-text attribution to IntraFi's own published count |
| Lancaster, PA community benefits agreement: ~$20M data-center contributions | Reported (2025 coverage) | Carried with in-text attribution; verify against the agreement text before any external use |
The structural findings in Sections 6 and 7 come from discourse network analysis. We built an InfraNodus graph from two corpora: the money-mechanics discourse (call notes, StoneCastle materials, creative copy) and the community-grievance corpus. The graph's cluster structure separates cleanly into a deposit-mechanics cluster and grievance clusters (water and rates, jobs, secrecy), with three structural gaps between the money mechanics and the lived grievances. The message repairs in Section 6 and the opening in Section 7 target those three gaps.
The conversation behind this report, mapped. Each dot is an idea from the June 9 call and the creative copy; a line means two ideas kept appearing in the same breath. Drag the dots. Three bridges are missing.
The financial mechanics — insured entities, corporate cash, risk profile — share no language with the town’s story of roads, schools, power, and meetings. The CFO pitch and the person at the podium never appear in the same sentence.
The “change the timeline” creative hook never touches the concerns it is meant to transform. Health fears sit isolated, connected to jokes and search beats rather than to any mitigation language.
The deposit story and the money-at-scale mechanics run as separate conversations, joined only by the words “community bank.” The town playbook in this report is the artifact built to join them.