The Facts Aren't as Settled as Alabama Is Being Told

Guest Opinion by Alicia Haggermaker

Dr. Bill Chitwood opened Monday's meeting the way a psychiatrist would: he told the room they are being psychologically manipulated every day, by every side, and that they shouldn't trust what they hear from him, from any speaker that night, or from anyone else — go verify it yourself. That was the standard the meeting set before a single claim was made.

I was one of the speakers who followed him. Regular readers here know where I stand and what I've spent years documenting, so I'll skip the introduction. The evidence isn't a matter of theory anymore. Declassified documents, official admissions, and public records now confirm claims that were dismissed as conspiracy for years — go read them yourself rather than take our word for it. Specific to what I raised that night: these hyperscale data centers are not necessary for what ordinary people actually need AI for. Mycelium — the community platform I built — is the proof of concept.

That's the context Perry Hooper Jr.'s recent piece is answering, whether it names us or not. Hooper's case is built on confidence: the facts have been studied, Alabama has reached its moment, and continued hesitation is now "its own form of risk." Senator Tuberville is right, Doug Jones's proposed moratorium is wrong, and the only real choice left is how fast Alabama moves forward.

That confidence deserves the same scrutiny Dr. Chitwood asked the room to apply to us. On the specific points Hooper leans on hardest — transparency and the underlying economics — the record doesn't support the certainty he's selling. And on what a real alternative to blind trust actually looks like, there's more built and running right now than a press release or an op-ed.

Transparency Has Already Been Tested Once

Hooper lists "insist upon transparency" among the protections Alabama can maintain while still competing for this investment, as though transparency and speed are naturally compatible. Huntsville's own history with this exact industry says otherwise.

In 2018, the City Council approved a development agreement and $6.6 million in incentives for a company known publicly only as "Starbelt" — its identity as Facebook wasn't revealed until weeks later. Mayor Tommy Battle thanked the council on the record for "being quiet about the agreement" while it moved through approval, and confirmed the city has used redacted development agreements before — citing the Toyota-Mazda deal as another example — and said it likely wouldn't be the last time.

So when Hooper writes that Alabama can insist on transparency without slowing down, the real precedent for a comparable deal in this state involved secrecy as a condition of speed, not a parallel track alongside it. If Alabama's leaders want to prove this time is different, the test is simple: will there be an NDA on Phase 2, yes or no? That's a question worth asking directly, on the record, before the next vote — not after.

No one can force a company or a city to choose transparency over an NDA, and that's not the argument here. It's a trade, and it should be treated as one: a project that runs its filings, comment periods, and commitments through an open, verifiable record builds something an NDA can't — a public trust record it can point to when someone asks it to prove good faith. A project that instead routes everything through closed channels doesn't get that. The choice stays theirs. So does what it says about them.

The Economics Aren't Settled Either

The facts underneath this exact industry are not settled at all. Right now, some of the most serious voices in American finance are locked in an open dispute over whether companies like Meta are even accounting honestly for the hardware these facilities are built around.

Investor Michael Burry — best remembered for correctly predicting the 2008 financial crisis — has publicly estimated that hyperscalers, Meta included, are understating depreciation on their AI chips by roughly $176 billion industry-wide between 2026 and 2028, by assuming a six-year useful life on hardware some analysts believe holds real economic value for closer to two or three years before it's outpaced by the next generation. Meta itself recently extended its own equipment's estimated useful life to 5.5 years, a change that lowered its reported depreciation expense by $2.3 billion. Whether that reflects sound accounting or a company leaning on generous assumptions to keep its earnings numbers looking healthy is precisely the kind of question serious financial analysts are still fighting over in real time.

If the people whose job is to get these numbers right can't agree on them, Alabama's leaders owe their constituents more than Mr. Hooper's confidence that the studying is already done.

The Concentration Wasn't Required — It Was Chosen

There's one more assumption in Hooper's piece worth testing directly: that facilities of this scale, in this one location, are simply what the technology requires. They aren't. Training the largest AI models today routinely happens across multiple, physically separate data center campuses — sometimes dozens of miles apart — linked by fiber to function as one combined system. NVIDIA's own networking technology, built specifically to link separate data centers into one functioning training system, is documented working at distances from roughly 500 meters up to hundreds of miles apart — tested at 10 kilometers with better performance than traditional networking, according to NVIDIA's own technical documentation — spreading a project across several smaller sites, each drawing less water and less power, is a proven, working alternative.

That means the choice to concentrate this much draw at one Huntsville site isn't a law of physics. It's a business decision — and it's a decision whose benefits and costs don't land on the same people. The savings from building one large site instead of several smaller, distributed ones belong entirely to the company: less coordination overhead, lower construction cost, simpler operations. The consequences of that concentration — the water draw, the strain on the grid, the infrastructure built to serve one facility — land entirely on the community around it. That is not a neutral engineering tradeoff. It is a cost shifted off a corporate balance sheet and onto residents' water bills and electric rates, presented as a technical necessity when the technology itself proves it didn't have to be built this way.

What "Proof of Concept" Actually Means

Every institution you've dealt with — the utility, the PSC, zoning, the school board, the platform you post on, the app you check reviews on — was built to serve someone else's interests before yours. Not necessarily malicious. Just structural. The data doesn't talk to itself across agencies because no one who built those systems needed it to.

Mycelium is not a nicer app. It's an argument, built in software: that a community can own its own infrastructure for truth, coordination, and mutual accountability — and that infrastructure can be honest by design instead of by promise. It runs on hardware a community actually owns, not a facility drawing millions of gallons of water and hundreds of megawatts of power.

What it already does: a social feed for sharing local finds and updates, sorted by time instead of an engagement-chasing algorithm; scheduling and coordination tools built around a Maker's Guild of local tradespeople; a business review system that only counts confirmed, verified transactions, so a rating actually means someone did business with that person; a Lost & Found and OSINT case system with geographic clustering, surfacing patterns across missing-persons cases that a single overworked investigator might never spot manually; a whistleblower forum with tiered anonymity and encryption, so a report doesn't require sacrificing safety to be heard; environmental dashboards pulling live water, air, and energy data straight from EPA, USGS, and EIA public records; a local AI stack running entirely on community-owned hardware, nothing sent to a third-party company's servers; and Meshtastic mesh radio integration, so local coordination keeps working even when the internet and cell towers don't. And for those of us that remember the good ol’ days, it even has some of the customization capabilities that MySpace had and the chatrooms reminiscent of the AOL dial-up era. It could even have an open-source Napster for original content people want to share that isn’t locked behind corporate deals and copyright limitations. The possibilities are endless.

All of it running today on a single desktop computer. Not a data center. Not a subscription. A machine one person, with little formal coding background, built with AI as a development partner — because the tools to build real community infrastructure are now within reach of anyone willing to put in the work, not just a company with a nine-figure budget and a non-disclosure agreement. If you know anyone who can code, or wants to learn by contributing to something real, that is the single most valuable thing anyone could bring to this effort right now. Every feature below is designed and ready to build. What it needs is more hands than one.

The Standard This Platform Holds Everyone To — Including Itself

There's a specific failure every part of this platform is built to interrupt. The real danger isn't a bad human decision or a wrong AI answer on its own — those are catchable. It's the two combining. A person who defers to whatever a confident system tells them, paired with AI that's built to sound agreeable even when it shouldn't be — neither checks the other, and the errors compound instead of canceling out. That's the same mechanism behind institutions rubber-stamping each other's assumptions, agencies that don't cross-reference, nobody catching a contradiction because everybody assumed someone else already had. Every AI feature on this platform follows one non-negotiable rule: surface and suggest, never decide, never act. It flags a possible connection between cases. It flags a contradiction between two filings. It drafts a response for a human to review before it sends. It never closes a case, contacts anyone, or reaches a conclusion on its own.

That principle applies without exception — including to whoever runs the platform. A system that catches contradictions between agencies but treats its own admin's word as beyond question hasn't closed the blind spot, it's just moved it.

Which leads to the actual working standard: a case file, not a statement. A short rationale is easy to write and easy to hide behind — it captures a decision-maker's account of themselves, not the evidence the decision rests on. Anyone making a call that spends public money, uses community resources, or affects people's safety or standing should have to build an actual case file first — what evidence was reviewed, who was consulted and what they said, including anyone who disagreed and not just the people who agreed, what alternatives were considered and why they were rejected, and what the real cost and risk were, in specific numbers, not vague reassurance. "I talked to five people and this benefits everyone" is not a case. It's a statement dressed up as one. When someone's freedom is on the line in a courtroom, the state shouldn't get to skip presenting evidence because it's confident it's right, and a decision that spends millions of someone else's dollars, or reshapes a neighborhood's water and power, deserves at least that same discipline. Nobody gets a role or a track record that puts them outside the reach of a fair question — not a company, not an official, not the person who built this platform.

What Actual Transparency Would Look Like

Right now, nobody outside a handful of officials can actually verify any of this in real time. Not the water figures, not the NDA status, not whether a utility's public claim matches what it filed privately. The institutions that hold each piece — city zoning, the Public Service Commission, the utility itself, environmental permitting, economic development agreements — don't coordinate with each other, let alone with the public. That gap is not hypothetical. It's the exact complaint raised by a candidate for the Public Service Commission at Monday's meeting, describing a process where "one arm doesn't communicate with the other."

There's a concrete way to close that gap, and it already exists in early form: an institutional accountability dashboard that takes filings, permits, and public commitments from every agency touching a single project and lays them side by side in one shared, timestamped record, flagging where they contradict each other. Right now that means uploading the documents by hand, because most of these records simply aren't available anywhere that software can reach automatically — no public API, no standard format, no single portal. That gap is itself worth sitting with: the fact that a resident has to manually track down and upload a permit, a filing, and a utility statement just to compare them side by side is a direct measure of how little of this process was ever built for public visibility in the first place.

The design goes a step further than tracking what's already public: a company can submit its own case file directly — its account of the good, the bad, and the ugly on a given project — but that submission isn't the last word. The public can add missing context, counter-documents, and direct questions to that same case file, so a company's self-reported account has to stand next to what the community actually knows, not replace it. A statement that goes unchallenged because no one had a place to challenge it isn't accountability. This is.

The same underlying model has a second, direct application: as an internal planning and coordination tool institutions themselves could adopt to fix that exact "arms don't communicate" problem, rather than one more system residents have to work around. Applied to companies and agencies rather than the platform's own admins, the trade works the same way: no one can force a project to run its filings through an open, verifiable record instead of an NDA — but the project that does builds a public trust record it can point to, and the project that doesn't, doesn't get one. The choice stays theirs. So does what it says about them.

A Verified Identity Doesn't Have to Stop at Reviews and Reports

A verified Mycelium account — one confirmed identity, community-vetted — could serve as the credential for civic processes beyond posting and reviewing, including transparent, verifiable voting: a cryptographically signed vote, tied to one account, aggregating to results a resident could check themselves rather than trust from a distance, potentially down to scanning a QR code at a polling place and pulling up live, verifiable precinct results directly.

The same verified-account system could also keep a voter roll honest automatically, with the right data-sharing partnerships — vital records offices for death certificates, DMV for change-of-address — cross-checking itself against real records instead of relying on manual purges that happen too slowly or not at all. A deceased account gets flagged and closed. A moved account doesn't just get deactivated — it transfers, so someone who relocates carries their verified standing to their new area instead of being disenfranchised or double-counted. That piece can't be built by software alone; it depends on institutions agreeing to share the data. But it's exactly the kind of partnership this platform is built to make a credible, documented ask for. Imagine not having to watch and wait for the votes to roll in during election time because you can just follow the progress in real time by scanning a QR code. Each vote added to the tally as it is made for each node, all updated live to the aggregate regions totals, with no sudden jump in tallies that doesn’t flag to the source of the anomaly.

This isn't hypothetical enough to dismiss as speculation, either — a version of this argument has already been made directly to Huntsville's leadership, and already been tested. In 2025, I presented a 70-plus page Wise City proposal to the full Huntsville City Council, including a community-backed stablecoin concept — backed by locally produced goods rather than federal debt, distributable through existing credit union infrastructure. Mycelium wasn’t built yet, but it could serve as the transparent tracking layer showing exactly where value moved and to whom. It was conceived as a resilience plan: a sovereign financial alternative a community could stand on if larger institutions failed it. I sent it directly to council member Michelle Watkins on September 16, 2025, at her request for a closer review. She never followed up. That’s why I built all of this out in the open and documented it on Facebook and UnPrecedentedTimes.org as it progressed.

The Refusal Is a Data Point Too

Several of the capabilities described here — cross-agency data center accountability, an honest voter roll, real institutional partnership on trafficking or public safety work — can't be built by software alone. They need an institution to say yes: share the data, accept the referral, integrate the record. That's not a weakness of the design. It's a deliberate feature of it.

Because the request itself becomes evidence. If a community offers a utility, an agency, or an office a transparent, verifiable system — one built specifically to make good-faith cooperation visible and easy — and that offer goes unanswered or gets refused, that refusal is worth documenting with the same rigor as anything else on this platform. This platform, this piece, and the questions in it aren't going away after this vote. Whatever Huntsville's leadership does next with the questions raised here — about Phase 2's water draw, about whether an NDA gets signed again, about whether any of this gets answered on the record before the next vote, won’t either. Don't assume bad faith. Don't assume good faith either. Build the system that makes the honest answer checkable, and let the response speak for itself.

That's a standard Alabama's leaders — on every side of this debate — should be willing to meet, if the confidence in this deal is as well-founded as Mr. Hooper says it is.

Alicia Haggermaker is an investigative journalist, systems analyst, and community organizer based in Huntsville, Alabama. She publishes at Unprecedented Times (unprecedentedtimes.org) and is the founder of the Mycelium Network — a sovereign community coordination platform built as an alternative to the infrastructure this series documents.

unprecedentedtimes.org | mycelium.unprecedentedtimes.org

Opinions do not reflect the views and opinions of ALPolitics.com. ALPolitics.com makes no claims nor assumes any responsibility for the information and opinions expressed above.

SOURCES;

The 2018 Starbelt/Facebook NDA and Mayor Battle's quote

Why did Huntsville City Council keep Facebook deal secret? — WHNT.com https://whnt.com/news/huntsville/why-did-huntsville-city-council-keep-facebook-deal-secret/ (Contains the direct quote: "To my colleagues on the city council, thank you for being team players... and also being quiet about the agreement as we move through it.")

$6.6 million in incentives, May 24 council approval, Toyota-Mazda precedent context

Facebook to Invest $750 million in New Data Center — City of Huntsville (official) https://www.huntsvilleal.gov/facebook-to-invest-750-million-in-new-data-center/

Alabama Jobs Act investment credit ($86.25M) and 30-year revenue projection ($297.3M)

Governor Ivey Announces Facebook to Invest $750 million in New Data Center in Huntsville — Office of the Governor of Alabamahttps://governor.alabama.gov/newsroom/2018/06/governor-ivey-announces-facebook-to-invest-750-million-in-new-data-center-in-huntsville/

Michael Burry's $176 billion depreciation claim (primary reporting)

'Big Short' investor Michael Burry accuses AI hyperscalers of artificially boosting earnings — CNBChttps://www.cnbc.com/2025/11/11/big-short-investor-michael-burry-accuses-ai-hyperscalers-of-artificially-boosting-earnings.html(Contains the direct quote: "Understating depreciation by extending useful life of assets artificially boosts earnings - one of the more common frauds of the modern era.")

Meta's specific 5.5-year useful life extension and earnings impact

Michael Burry revives bear case for AI chips amid GPU depreciation concerns — CryptoBriefing https://cryptobriefing.com/michael-burry-bear-case-ai-chips-gpu/ (Confirms: "Meta has adopted a 5.5-year schedule for its GPU assets.")

The counter-argument / complicating view on chip useful life (for balance)

Why Michael Burry Is Wrong About AI Depreciation (But Still Might Be Right) https://interestingengineering.substack.com/p/why-michael-burry-is-wrong-about (Useful if challenged on the depreciation point — presents the hyperscaler-side argument that older chips still generate revenue via inference even after they age out of frontier training use.)

Multi-campus/distributed AI training being standard, deployed practice

How to Connect Distributed Data Centers Into Large AI Factories with Scale-Across Networking — NVIDIA Technical Blog (primary source, from the chipmaker itself) https://developer.nvidia.com/blog/how-to-connect-distributed-data-centers-into-large-ai-factories-with-scale-across-networking/ (Confirms NVIDIA's Spectrum-XGS Ethernet links separate data centers — from roughly 500 meters apart up to "tens or hundreds of miles, across cities or even states and countries" — into one functioning AI training system. Tested at 10 km between sites with up to 1.9x better performance than standard networking.)