Project Arrowhead: Inside Irwin County, Georgia's Data Center Fight

How a failed solar deal became the most controversial land battle in rural Georgia

I. The Room in Ocilla

On the night of February 26, 2026, the Ocilla-Irwin Planning Advisory Commission met in a municipal room that was not built for what arrived inside it. The agenda item read, in county shorthand, IC-SE-01-2026: a special-use request for a 546.35-acre “Irwin Forward Tech Park” on Ponderosa Drive, out near the Alapaha River, in the soft middle of south Georgia farm country.

Patrick Atwater was there. His 400 acres of hunting land sits next to the parcel. He said, on the record, that he had not been contacted by anyone about a data center arriving over the fence line. Marvin Peavy was there. His 1,500-acre family farm has worked that ground for generations; his concerns ran through water, through noise, through the shape of the place he calls home. The room filled with red shirts, a color the opposition had quietly agreed on, and people spoke for three hours about cooling water, about transformer hum, about what a 4.2-million-square-foot campus does to a cotton road at four in the morning.

The Planning Commission voted 4-0 to recommend approval.

Four days later the applicant’s counsel pulled the filing back, “without prejudice.” Six weeks later the same project reappeared, doubled in size, as a state-level Development of Regional Impact filing: 1,066 acres, 4.22 million square feet, 1.25 gigawatts at peak, $6 billion in capital expenditure, roughly 200 permanent jobs, Phase 1 complete 2030, full buildout 2034.

And across two public filings, one local hearing, one withdrawal letter, and one state portal record, no tenant name appears. Not one.

That absence is the story.

“I’ve lived here my whole life. Nobody called me. Nobody came by. And now the fence line is going to light up like a small city.”

— a sentiment expressed in many variations at the February 26 hearing, captured on the WWALS Watershed Coalition’s video archive, which has become the citizen record of this fight.

A brief note on that archive, because nearly every primary source in this article flows through it. WWALS Watershed Coalition is a 501(c)(3) nonprofit based in Hahira, Georgia, serving as the official Suwannee Riverkeeper under the national Waterkeeper Alliance.

Its formal mission is to protect the Withlacoochee, Willacoochee, Alapaha, Little, and Upper Suwannee river basins through water-quality monitoring, advocacy, citizen science, and public education. Its practical role, in fights like this one, has become much larger: WWALS shows up at planning commission and board of commissioner meetings across its basin counties with cameras, posts the videos publicly, writes up the proceedings with primary-source care, and maintains a searchable online archive at `wwals.net` that has effectively replaced the public reporter’s notebook for data-center-scale land-use decisions in this part of south Georgia. When a county does not livestream its hearings and the local newspaper does not send a reporter, WWALS is frequently the only public-record videographer in the room. Much of what a community one county over can learn about Project Arrowhead exists because John S. Quarterman and the WWALS team drove to Ocilla and pointed a camera at the dais. This piece relies on their archive throughout, and it is worth naming why that archive exists.

In the parallel county this piece will come back to, a woman named Jacqueline Lassetter in Coweta County, Georgia, lived through the same kind of room eighteen months earlier and watched her commissioners vote 3-2 for a project that turned out to be backed by Prologis, the largest industrial real-estate investment trust on earth. Lassetter’s name appears in the April 7, 2026 DeSmog investigation that finally pulled the curtain back on that deal. The same broker who appears in Coweta’s public record appears, physically, in the Ocilla room on February 26.

His name is Mike Lash. He is an Executive Vice President at CBRE Data Center Solutions in Atlanta. He was in the room. This article will come back to him.

But before it does, it needs to do something the people in that room deserve: present, cleanly and without false drama, what the public record actually shows about the fight they are in, what the public record shows about the fight next door, and how to tell the difference between the play they think they are facing and the play they are actually facing. Because those two things are not the same. And the gap between them is where communities in this position lose.

II. What the Public Record Shows

Eight days in February shaped the deal. Read them in order.

On February 16, 2026, WALB television reported that Irwin County was moving to dissolve its Industrial Development Authority. The county framed the action as “checks and balances” governance restructuring and specifically said it was separate from any data center proposal. A second rural south Georgia county had done the same thing weeks earlier. The dissolution vote and roll-call details are not yet in the retrievable public record.

On February 19, 2026, a bankruptcy court in the Southern District of Texas confirmed the Chapter 11 plan of Pine Gate Renewables, a multi-state solar developer. Embedded in that plan was the disposition of a Memorandum of Option that had been recorded against Sirrom Farms LLC’s parcels in Irwin County since February 2020. The option had originally run to Samsung C&T and then to Pine Gate for a project called Reedy Solar. The Chapter 11 plan abandoned it. For the first time in just under six years, the Sirrom parcels stood unencumbered by any development right.

On February 25, 2026, Premier Data Center Development, LLC was formed at the Georgia Secretary of State (Control Number 26043985). Organizer of record: Kelly O. Faber of Faber|Mabe, a real-estate law firm in Dacula, Georgia. Registered agent: CT Corporation System, in Lawrenceville. Mailing address: PO Box 212, Statham, Georgia 30666. No members are listed; Georgia does not require LLC members to be disclosed at formation, and this is the ordinary shape of a shell.

On February 26, 2026, the Planning Commission met and voted 4-0 to recommend approval of a 546.35-acre special-use request on the Fletcher Trust parcels alone. Present in the room, per the WWALS Watershed Coalition’s video archive: Mike Lash of CBRE Data Center Solutions; a civil engineer introduced only as “Joshua,” who described himself as having fifteen years of data center experience out of Alpharetta; and Gerald Pouncey, at that time recently transitioned from Chairman of Morris, Manning & Martin to Senior Counsel at Taft Stettinius & Hollister following the firms’ December 31, 2025 merger.

Named on the initial planning materials and in the Planning Commission packet were two additional firms whose presence tells its own story: Abernathy Development Company LLC of Lawrenceville, listed as the project’s Planning Consultant, and Kimley-Horn Associates of Mobile, Alabama, listed as the engineer of record. Abernathy is a forty-year Gwinnett-based retail and mixed-use developer with no documented prior data center project; Kimley-Horn is a national civil-engineering firm with a substantial data center practice. The mix of a local zoning-entitlement specialist paired with a national data center engineer is a signature of the Tier 4 archetype this piece will return to: a locally credible planning face in front, national-grade technical capacity behind.

On March 2, 2026, Pouncey and a Taft colleague named McCullers signed a withdrawal letter pulling the Planning Commission recommendation before it could reach the Board of Commissioners. The stated reason, reported via counsel email to the county: the filing “wasn’t ready” and “needed more specificity.” The withdrawal was “without prejudice,” meaning the filing could return.

It returned on April 10, 2026, not as a local rezoning but as DRI #4689 at the Georgia Department of Community Affairs: Project Arrowhead, 1,066 acres, an upward revision of roughly 520 acres, now straddling five parcels rather than the original footprint. The 520-acre expansion is not incidental. Between the withdrawal and the refile, Sirrom Farms was captured into the project. The original Planning Commission application had been drawn only around the Fletcher Trust tract. The refile added the adjoining Sirrom parcels, and with them two things the Fletcher-only footprint did not have: a critical-path connection to a main road, and the substation-ready ground that every data center of this scale has to sit near.

Both of those assets existed on Sirrom because the land had spent four years engineered for a utility-scale solar project that never got built. The solar work did not vanish when Pine Gate went bankrupt. It stayed as transmission proximity studies, easement work, and the substation footprint now drawn into the Arrowhead site plans. Whoever re-engineered the refile understood that the Fletcher tract alone could not carry the load. The refile reads as a project that discovered, mid-process, that it needed Sirrom.

The DRI form names Fletcher Marcus D Trust (Angela F. Bryan, trustee, per deed records) and Sirrom Farms LLC (the Morris family; “Sirrom” is “Morris” reversed, per Georgia Secretary of State Control Number 12098617) as the property owners. The developer-of-record contact on the DRI is Russ Bryant, a landowner neighbor on the same road, whose personal contact information is now on a state filing as the formal point of contact for Premier Data Center Development, LLC. Whether Bryant holds equity in Premier DCD is not publicly disclosed.

The DRI lists no anchor tenant. The “Continuation of Previous DRI” field is marked No. On April 14, the Department of Community Affairs determined the filing warranted regional review by the South Georgia Regional Development Commission.

Six days cleared the land. One day formed the company. The next day a planning commission recommended the project. Four days later the applicant pulled it back. Six weeks later it returned, doubled, at the state level.

That is not how a project runs when it has an anchor tenant in hand. That is how a project runs when a developer is racing a political window and an entitlement clock.

---

III. The Man in the Room

Before naming the person, it helps to say plainly what CBRE is and what it is not, because the confusion about this is part of why data center fights like Arrowhead are hard for communities to read.

CBRE Group, Inc. is the world’s largest commercial real estate services and investment firm. It is a publicly traded company (NYSE: CBRE). It is not a data center developer. It does not own data centers. It does not build data centers. It does not operate data centers. What CBRE does is sit in the middle of commercial real estate transactions as a brokerage, advisory, and investment-management intermediary. It represents buyers and sellers of land and buildings. It advises corporate occupiers on site selection. It manages properties. It brokers leases. It handles capital-markets transactions for institutional real estate. CBRE’s Data Center Solutions practice, which is where Mike Lash works, is a specialized advisory and brokerage team that represents data center operators, hyperscalers, and institutional investors in the site-selection, acquisition, disposition, and sale-leaseback transactions that move data center real estate. CBRE also has an Agricultural Services practice, which brokers rural land. The two practices share a firm; internal handoffs between them, when a rural landowner is approached about a transaction that turns out to have a data center buyer pool attached, are the ordinary architecture of the business.

This matters because the way CBRE appears in a public record is as a deal intermediary, never as a principal. When a CBRE Data Center Solutions executive attends a rural county zoning hearing, they are there on behalf of a client, not as the party that will own or operate the facility. The client may be a hyperscaler. The client may be a wholesale colocation operator. The client may be a speculative developer with an institutional backer. CBRE does not disclose client identity absent authorization. A CBRE executive in a hearing room is thus simultaneously the most important tell a community has about who is serious enough to hire CBRE, and the least direct source of information about who the ultimate party is. That is the structural position Mike Lash occupies in this story.

Mike Lash is identifiable from his CBRE public profile and from his physical appearance in the WWALS video of the February 26 hearing. He is an Executive Vice President at CBRE Data Center Solutions in Atlanta, on Tim Huffman’s team. His public broker history, as CBRE itself advertises, spans QTS, Equinix, Compass, Digital Realty, Landmark Dividend, and Mortenson. In 2022 he brokered the QTS 615-acre mega-campus in Fayetteville, Georgia. In the DeSmog investigation of April 7, 2026, Lash is named by public records released to reporters as having met with Coweta County staff on February 11, 2025 alongside Prologis representatives and a local realtor named Leigh Ann Green. He is named again in those records as having returned to the Coweta County commission in mid-November 2025 to personally lobby for ordinance changes that raised the maximum building height from 60 feet to 70 feet and relaxed road-frontage requirements, both favorable to the Project Sail data center campus.

Project Sail, the Coweta project, is Atlas Development LLC on paper. In May 2025, after roughly nine months of shell-LLC operation, Prologis disclosed itself as the backer; a Prologis logo appeared on a site illustration; the general counsel of the county development authority, Sarah Jacobs, emailed Prologis vice-presidents thanking them for op-eds published under a Prologis name in the Newnan Times-Herald. A former Coweta County Development Authority president, Trae Westmoreland, walked directly from the authority into a newly created Prologis position titled Director of Data Center Policy. On April 10, 2026, the Coweta County Commission approved the Project Sail rezoning 3-2.

One broker. Two counties. Same year. Same playbook signature. The strongest analyst inference from the public record is that Lash is the connector. This is not a smoking gun on the Arrowhead backer; it is pattern.

Same broker, same playbook, two Georgia counties, same year. The broker is the pattern. The pattern does not tell you the tenant. It tells you the type of play.

Lash himself appears in no public record on Arrowhead other than as a participant in the February 26 hearing (visible in the WWALS video) and as an individual with a public role on the CBRE website. He has not been reached for comment for this piece; this piece makes no claim about his intent. The public record records his public roles. That is the line this article holds.

The broker is not the only tell. The lawyer is.

Gerald L. Pouncey Jr. signed the March 2, 2026 withdrawal letter alongside a Taft colleague named Stephen A. McCullers. Until December 31, 2025, Pouncey was Chairman of Morris, Manning & Martin, one of Atlanta’s most prominent corporate law firms. MMM’s reputation in the Southeast was built on technology, real estate, and infrastructure practice. Its long-standing data center client roster and its depth in REIT and infrastructure private-equity work made it one of the two or three go-to Southeastern firms for data center site entitlement. Pouncey himself has been recognized nineteen consecutive years by Best Lawyers in the environmental law category. He chairs the Council for Quality Growth and the Georgia Brownfield Association. His personal specialty is the fact pattern that arises when industrial redevelopment sites have environmental, permitting, wetlands, air, water, or electrical-infrastructure complexity. He is, in Southeastern industrial-real-estate practice, one of the people you hire when a site is complicated and the stakes are high.

On December 31, 2025, Morris, Manning & Martin was acquired by Taft Stettinius & Hollister, a nationally-scaled law firm headquartered in Cincinnati. Federal lobbying disclosures place Taft’s 2025 billings across thirty-eight clients at roughly $2.49 million. The merger brought MMM’s Southeast data center practice under the Taft brand, and Taft itself formalized its offering as a named “Data Centers, Digital Infrastructure & Powering AI” practice group. Pouncey transitioned from Chairman of MMM to Senior Counsel at Taft. Senior Counsel, at a firm of Taft’s size, is the post-chairmanship designation reserved for senior partners whose continuing value to the firm is client relationships and counsel rather than case volume.

The clients that Taft’s data center practice represents are hyperscalers, wholesale colocation operators, infrastructure REITs (the most common corporate structure for large colocation entities), and the private equity and pension-fund vehicles that back them. This is the client tier that pays Big Law senior-partner rates without blinking. A senior counsel of Pouncey’s profile at a post-merger firm of Taft’s scale commands hourly rates that are, in the Southeastern market as of 2026, in the range of roughly $1,000 to $1,500 per hour. An engagement that carries a project from initial local filing, through a withdrawn Planning Commission recommendation, through a state DRI refile with expanded scope, through regional review, and toward an eventual Board of Commissioners vote will run hundreds of hours across a small team. The total legal fee exposure on the February-through-April work already visible on Arrowhead is measurable in the low-to-mid six figures, and the downstream engagement is a multiple of that.

Someone is writing those checks. A Senior Counsel from a post-merger national firm with a named data center practice is not retained by a speculative rural landowner. He is retained by an institutional actor who is taking the project extremely seriously, who expects to defend the entitlement, and who is prepared to pay for the kind of legal work that shows up in public record only as the withdrawal letter and the refile.

Someone behind this is taking it very seriously. That is what the lawyer’s name tells you, even before you know who wrote the check.

The engineer is the third tell.

Kimley-Horn Associates, the civil engineering firm listed as engineer of record on the Arrowhead planning materials, is a national multidisciplinary firm with a substantial data center practice. They are not obscure. They authored the Transportation Impact Study for QTS Data Centers’ 615-acre Fayetteville mega-campus (DRI #3813, “Project Excalibur”), the project that became Microsoft’s Fairwater AI superfactory. They have engineered at least six Georgia DRI filings for data center projects since 2022. And in a detail that matters more than anything else in this section: Kimley-Horn’s Mobile, Alabama office drew concept site plans dated February 1, 2025, for a ~546-acre data center on the same Irwin County parcels that became Project Arrowhead. Those plans were attached to the original February 26 filing that the Planning Commission recommended 4-0. The withdrawal on March 2 pulled those plans. The refile on April 10, doubled in size, landed on the same ground.

Kimley-Horn was drawing Irwin County data center plans a full year before Arrowhead went public. They were not hired in response to a February 2026 zoning filing. They were hired long before it. Someone commissioned a national data center engineering firm to draw a South Georgia campus in early 2025, when no public record of the project existed.

Three service providers are now visible on Arrowhead, and all three have documented ties to the same client: QTS Data Centers, a Blackstone subsidiary and one of the largest wholesale colocation operators in the country.

A caveat belongs here and it is an important one. None of these three firms works exclusively for QTS. Lash’s CBRE public profile lists QTS, Equinix, Compass, Digital Realty, Landmark Dividend, and Mortenson as clients. Kimley-Horn has engineered data center DRIs for at least five other developers in Georgia since 2022. MMM’s zoning practice represented dozens of large-scale Atlanta real estate clients. Any one of these names on Arrowhead would be unremarkable. All three, on the same project, with documented QTS relationships, is a pattern. This article treats it as a pattern, not as proof.

---

IV. The Coweta Playbook, Side by Side

Project Sail, 80 miles north in Coweta County, is now the most thoroughly documented data center land grab in Georgia’s public record, largely because of the April 7, 2026 DeSmog investigation by reporter Justin Nobel and colleagues. The contrast between what is documented there and what is documented at Project Arrowhead is not an academic exercise. It is the single most useful thing a community in Irwin County’s position can look at, because it shows what evidence has already surfaced on the older, riper fight, and what evidence has not yet surfaced on theirs.

The most parsimonious reading of that contrast is not that Irwin is a fundamentally different kind of project. It is that Irwin is a Tier 4 speculative-developer-led project whose operators appear to have watched what happened in Coweta and chosen to operate through different governance channels, with less email trail, through a dismantled (rather than a captured) local authority, using a developer-contact-of-record who is a local landowner rather than an institutional officer. That is an analyst’s inference, labeled as such.

There is a second, humbler reading worth holding alongside: Irwin may simply be earlier in the same arc, and the paper trail will accumulate on the same schedule Coweta’s did, nine to eighteen months from now. Both readings are live. The consequence for a community reading its own fight is the same either way: the Coweta artifacts are the rosetta stone. Whatever is documented there is worth searching for, carefully, in one’s own county.

V. The Fayetteville Template

The Coweta comparison shows Arrowhead’s political playbook. Fayetteville shows something harder to see and more important: the **development pattern**, from first land contact to tenant reveal, that the same service-provider team has already run to completion.

QTS Data Centers’ 615-acre Fayetteville campus, now home to Microsoft’s Fairwater AI superfactory, did not arrive as QTS. It arrived in three stages over five years, each behind a different name.

Stage 1 (2020): The stalking horse. DRI #3160 was filed by Oceanic Data Centers, a Jacksonville, Florida company run by President Joel Embry and Chairman Jason Cohen. Oceanic ran the first DRI, the first Planning and Zoning appearance, and the first site-plan approval, de-risking the property politically before QTS ever surfaced. A community resident attending that 2020 hearing would have seen a Florida-based company they had never heard of. They would not have seen QTS, Blackstone, or Microsoft.

Stage 2 (2022): The shell. DRI #3813 was filed under National Acquisition Co LLC, a Kansas-address shell entity. This was the filing Kimley-Horn engineered the Transportation Impact Study for. On June 30, 2022, Fayette County flipped the zoning from R-70 (two-acre residential) to BP Business Park in a unanimous special-called vote, three days after community opposition organized. The Fayette County Development Authority, not QTS, was the rezoning applicant, a classic Georgia incentive-enabling structure. CBRE’s Mike Lash and Tim Huffman then brokered the $153.8 million land sale to QTS. The Development Authority had acquired the land at roughly $12,559 per acre and flipped it same-day at approximately $1,479,235 per acre, a markup of roughly 118 times. The Authority netted approximately $75 million.

Stage 3 (2026): The named entity. Only in the current expansion filing, DRI #4603, does a QTS-branded entity, QTS Eastwood LLC, appear as the applicant. Six years after the first DRI.

The tenant was revealed even later. Microsoft was publicly identified as the anchor tenant in November 2025, when the first Fairwater building opened. QTS’s $4.6 billion green bond offering in April 2026 named Microsoft in the materials. Five years elapsed between the first DRI and the tenant reveal. Tenant concealment was not an accident. It was deliberate policy.

The Fayetteville pattern has a diagnostic shape: a locally-unfamiliar front entity files the first DRI and absorbs the political friction. A shell entity files the middle-stage DRI and captures the rezoning. The named operator surfaces only after entitlements are secured. The tenant surfaces only after construction is underway. At every stage, the community is making zoning decisions about an entity that is not the entity that will ultimately own and operate the facility.

The Arrowhead overlay is not conclusive, but it is specific. Premier Data Center Development, LLC, formed February 25, 2026, with a local landowner’s Yahoo email as the DRI contact, occupies the same structural position that Oceanic Data Centers occupied in Fayetteville in 2020: a locally-obscure front entity filing the initial entitlements. The same CBRE broker is visible. The same law firm lineage (MMM, now Taft) is retained. The same engineering firm (Kimley-Horn) drew the plans. The 3-stage DRI chain has not yet played out at Arrowhead, and it may never. But the service-provider team that ran the Fayetteville chain to a $4.6 billion bond offering is the same team that is standing in the room in Ocilla.

The Fayetteville template also provides the most important single number for the community to hold: five years. That is how long it took from the first DRI to the tenant reveal. A community being asked to approve zoning for Project Arrowhead in 2026 may not learn who the tenant is until 2031. The ask is not “approve a facility for a named company.” The ask is “approve a zoning change now, on the understanding that the company behind it may not be named for half a decade.”

Whether a community can live with that ask is a political question, not an intelligence question. But an intelligence question it is possible to answer is: what kind of actor makes that ask? The Fayetteville record answers it. The kind of actor that made that ask in Fayetteville was QTS, backed by Blackstone, building for Microsoft. The kind of actor making that ask in Ocilla is, on the current pattern of evidence, the same team.

---

VI. The Seven Types of Data Center Play

This is the section this article was written to deliver. The typology below is built from two years of Georgia public record, DeSmog’s reporting on Project Sail, the Georgia Department of Audits’ December 2025 economic-impact report and its January 2026 revision, and a broader scan of national data center development practice. It is meant to be legible to a community member who has never looked at a 10-Q. One named exemplar each. Three diagnostic tells each. A best-case and a worst-case framing. Where Project Arrowhead fits. Where to look.

Tier 1 — Hyperscaler-Direct. Meta, Google, Microsoft, Amazon Web Services, Apple.

Exemplar: Microsoft Azure East US 3, Fulton and Douglas counties, announced late 2025, roughly 320 acres acquired in early 2026 for approximately $115 million.

Tells: ownership direct in the hyperscaler’s name or a named captive subsidiary; site control driven by a corporate real estate team, not a commercial broker; public announcement before or concurrent with development start.

Best case for community: LEED-Gold-grade sustainability commitments, direct employer-of-record accountability, fiber and power upgrades that serve the region.

Worst case: stranded capacity if the hyperscaler pulls out after zoning; closed-campus isolation; water depletion at full utilization.

Tier 2 — Hyperscaler-via-Developer. Microsoft through Stack, Google through Tract, AWS through captive partners; site selectors like McCallum Sweeney, Austin Consulting, Biggins Lacy Shapiro; general contractors Mortenson, DPR, Holder, Turner.

Exemplar: earlier-generation Google campuses built by Mortenson.

Tells: a site-selection consultant’s fingerprints in the pre-zoning correspondence, a named Tier-1 general contractor mobilizing on LinkedIn, a corporate real-estate-side attorney rather than a local real-estate attorney.

Best case: efficient land assembly with hyperscaler capital discipline.

Worst case: shell collapse, speculative build-to-suit, stranded zoning if the handoff breaks.

Tier 3 — Wholesale Colocation. QTS, Equinix, Digital Realty, Compass, Iron Mountain, CyrusOne, EdgeConneX.

Exemplar: Stack Infrastructure’s Lithia Springs campuses in the Atlanta market, IPI Partners-backed.

Tells: named corporate entity on signage and in pleadings from the outset, not a shell; dense metro-adjacent locations near fiber hubs; pre-leasing announcements; operational executives publicly visible.

Best case: competitive local capacity, diversified tenant base, predictable property tax revenue.

Worst case: capacity glut if the market saturates, fiber monopoly risk, tenant churn.

Tier 4 — Speculative Developer-Led with Hidden Institutional Backer. Prologis, sometimes Tract, occasionally Novva.

Exemplar: Project Sail / Atlas Development LLC with Prologis as the disclosed backer.

Tells: shell LLC formed weeks to months before zoning action; attorney-fronted organization (local real estate counsel rather than institutional officers); broker-driven public narrative that describes the campus as a “tech park,” “industrial flex,” or “logistics park” until after zoning; regulatory hearings staffed by brokers and counsel, not backer representatives; rapid disclosure within 60-180 days of approval, or quiet sale of the entitled site to a wholesale colo operator.

Best case: efficient land assembly, patient institutional capital, successful tenant capture.

Worst case: backer abandonment post-approval, shell-LLC liability vacuum, community distrust, stranded zoning.

Tier 5 — Rural Land Aggregation. CBRE Agricultural Services and landowner-side operators, stitching parcels for eventual resale to a Tier-1 through Tier-4 buyer.

Exemplar: the land-assembly layer that precedes many of the above tiers.

Tells: large acreage assembled quietly from multiple family owners; agricultural-zone conversion complexity; planning advisory (rather than commissioner) hearings used as early “test” reads; landowner-side counsel only, no institutional counsel. *Best case:* community-sensitive local farmer partnership with long-tail rural infrastructure benefits. *Worst case:* option expiration, speculative land bank, abandonment.

Tier 6 — AI-Compute Upstart. xAI, CoreWeave, Crusoe, Lambda, Together.

Exemplar: Crusoe Energy’s 2025-2026 GPU rollout.

Tells: venture or private equity capital in public press releases; GPU-density over network capacity; decentralized site selection driven by stranded energy; workload-specific (AI training); founder or CEO visibility in community engagement.

Best case: high-value compute jobs, advanced cooling innovation.

Worst case: stranded GPU capacity on model obsolescence, rapid operational failure.

Tier 7 — Bitcoin or Hybrid Mining-to-AI. Stronghold, TeraWulf, Marathon, Riot.

Exemplar: the Stronghold-Bitfarms consolidation of March 2025.

Tells: crypto-native operator history; stranded or curtailable power utilization; modular rapid-deployment facility design; community engagement often weak or reactive.

Best case: grid flexibility services, stranded-renewable utilization.

Worst case: crypto-price-collapse-triggered abandonment, noise and water complaints, potential state legislative ban.

Where Arrowhead fits: Tier 4, with moderate-to-strong confidence. The tells are almost all present. Shell LLC formed the day before the planning commission hearing. Local real-estate-attorney organizer. CBRE Data Center Solutions broker visible in person. Senior local counsel (Pouncey/Taft) leading the regulatory engagement. No anchor tenant disclosed. No general contractor mobilized. No hyperscaler-direct fingerprints (no Mortenson LinkedIn job posts in the area, no corporate real-estate officer at the hearing, no Azure or Google press release). And a state-level refile that doubled the footprint after the local withdrawal. Ongoing discussions are going on with someone in the background, informing scale of the development, but the plans remain fluid. There are no signs of the typical reference designs (pre-baked data center designs) that hyperscalers and tier one players bring to the table and drop into filings. This is a design on the fly, and a deal that has the haulmarks of being put together, rather than landed.

What this typology does not tell you is who the backer is. Capital availability in the data center market is currently universal: Prologis, Compass (KKR added $2 billion in March 2026 alongside existing Brookfield and Ontario Teachers stakes), QTS (Blackstone plus $8 billion portfolio securitization in flight), Tract, Aligned, Stack, Rowan, Microsoft, Google, Meta, AWS, xAI, all spending aggressively. One cannot reason from “X has a war chest” to “X is the tenant.”

The leading working theory for Project Arrowhead, from the public record, is QTS Data Centers (a Blackstone subsidiary), on the basis of the service-provider nexus documented in Section III and the Fayetteville development-pattern template documented in Section V. The three service providers visible on Arrowhead, Mike Lash (CBRE), Gerald Pouncey (MMM/Taft), and Kimley-Horn, all have documented engagements with QTS on Georgia projects. QTS is simultaneously running a 12-million-square-foot campus in Blakely (Early County, DRI filed March 2026, construction start July 2026) and expanding its Fayetteville campus (DRI #4603, April 2026), confirming that QTS has both the capital and the appetite for parallel South Georgia plays. Blackstone’s capital stack, including a $4.6 billion green bond sale in April 2026, provides the deployment capacity. This is a pattern-of-evidence working theory, not a confirmed identification. No QTS filing, no Blackstone filing, and no QTS press release names Arrowhead. No QTS-branded entity appears in any Irwin County public record. The place to look for them is structured debt raises from Blackstone that will be required to fund this site, if it gets to that stage. These can be found on the DealLogics Platform for those interested.

Prologis remains a secondary hypothesis on the basis of the Lash-Atlas-Prologis pattern at Project Sail, Prologis’s Q4 2025 earnings disclosure that 40% of its $4-5 billion development pipeline is directed at data centers, and Prologis’s existing footprint in adjacent Georgia counties. The Prologis hypothesis is weaker than the QTS hypothesis because it rests on a single shared broker (Lash), whereas the QTS hypothesis rests on three shared service providers plus a matching development pattern.

The hyperscaler-direct alternative (Microsoft, Google, Meta, AWS) is typologically off-pattern and is not advanced here as a serious hypothesis: hyperscalers buy, own, and develop in their own corporate names or through known captive vehicles, not through shell LLCs organized by a local real-estate attorney the week of a hearing. The Tier 4 shape of Arrowhead is evidence, not noise.

Google is local rumor. The Change.org petition that circulated in Irwin County in early February asserted Google as the tenant with no primary source. The WWALS community archive itself hedges with the phrase “I’m told.” No filing, no logo, no land disclosure, no Google-adjacent general contractor mobilization has surfaced. Google may ultimately be the tenant; Google may never be the tenant. The public record does not support naming Google today. This article does not.

---

VII. The Two Community Camps, and What Each Is Missing

In every data center fight that has reached a public hearing in Georgia in the last three years, two community camps emerge. Arrowhead is not an exception.

The tax-counters are reading from a script the developer will soon be handing out. The script will contain a construction-job number in the thousands, an operational-job number in the hundreds, a property-tax revenue projection in the tens of millions annually, and a multiplier effect on local retail, services, and housing. Every line on that script is subject to independent audit. Almost all of it has been audited.

In December 2025 the Georgia Department of Audits and Accounts published a statewide economic-impact analysis for data center development. In January 2026 the same office published a revision. The revision cut the headline construction-jobs number from 28,350 to 8,505. It cut operational jobs from 5,471 to 1,641. It cut value-added by roughly 70 percent. Georgia’s data center sales and use tax exemption, the policy mechanism that makes almost all of this development economically viable at the facility level, cost the state $296 million in FY25 and is projected to cost $327 million in FY26. A prior Vesper: Public Intelligence piece, The Digital Land Grab: Georgia’s Data Center Wars, cited the same Georgia Department of Audits finding that roughly 90 percent of Georgia’s existing data centers would not have been built without the exemption, meaning the state is foregoing a third of a billion dollars a year to subsidize facilities that would otherwise have located somewhere else.

National benchmarks sharpen the picture. The average ratio of construction jobs to permanent jobs at a data center is roughly 10.7 to 1, meaning the thousand construction workers a community is told to expect translate into somewhere under a hundred permanent positions. The average cost per permanent job across data center incentive packages studied by Good Jobs First runs $1.95 to $2 million; in the worst example on record, Foxconn’s Genesee County Michigan proposal, the effective cost per permanent job reached $11.7 million. Closer to home, the QTS Fulton County example documented in the Atlanta Journal-Constitution traded a $45 million abatement for 15 to 20 permanent jobs, a per-job cost of $2.25 to $3 million. Timothy Bartik of the Upjohn Institute, one of the most-cited economists in the incentives literature, estimates that between 2 and 25 percent of firms change location decisions because of incentives, meaning 75 to 98 percent of incentivized projects would have located in the same region anyway.

The tax-counters are not wrong that Arrowhead would deliver some tax revenue. They are wrong that the math they are about to be handed bears any resemblance to the math that actually shows up in the second and third year of operation. The tax revenue and employment impact of a Tier 4 speculative-developer project depends entirely on whether a tenant is ever found and who the tenant turns out to be. The community is being asked to approve a zoning change now against a revenue model that does not yet have a counterparty.

The Google-believers are inhabiting a different story, and that story is what powers the tax-counters’ script. The two camps look like opposites, but they are really one closed loop: the Google rumor tells the tax-counters that the revenue is real, and the expected revenue tells them the Google rumor must be right, because that kind of money would only come from that kind of tenant. The loop is self-reinforcing and it produces the “counting the tax money” phenomenon in community Facebook threads and living-room conversations.

Where did the Google rumor come from? From nobody with a primary source. The February 5, 2026 Change.org petition circulated in Irwin County asserted Google as the tenant with no filing, no press release, no named source. WWALS Watershed Coalition, the single most rigorous citizen archive of this fight, uses the hedge “I’m told” whenever Google is mentioned, which is exactly what you write when you know you are repeating a claim for which there is no backbone. Community chatter then carried the rumor forward as if it had been confirmed. It has not been confirmed. No Google filing, no Google logo on a planning illustration, no Google-adjacent general contractor (DPR, Holder, Turner, Mortenson) has posted a superintendent role in Irwin or its adjacent counties, no Azure-or-Google real-estate officer has appeared at any hearing, and the project attributes in DRI #4689 look nothing like a Google-direct play.

If Arrowhead were a Tier 1 hyperscaler-direct project, much of what the Google-believers expect would be true. Tier 1 projects leave documentary fingerprints. None of those fingerprints exist on Arrowhead. The documented service-provider nexus points not to a hyperscaler but to a wholesale colocation operator, most plausibly QTS, as the institutional actor behind the shell. A hyperscaler may ultimately become the tenant of whatever facility QTS or another operator builds, exactly as Microsoft became the anchor tenant at QTS’s Fayetteville campus. But a hyperscaler is an end-user, not a developer, and no hyperscaler-direct fingerprint (no corporate real-estate officer, no named general contractor, no press release, no Azure or Google job posting in the area) exists on Arrowhead. The community that is told “Google is coming” is being asked to accept a fact that is not in evidence. What the evidence supports is that the same team that built Microsoft’s home in Fayetteville is assembling entitlements in Ocilla.

And there is a third fact that matters for both camps: a meaningful share of announced data centers never get built. The industry’s announce-to-commission attrition rate is, by any reasonable estimate, substantial. Projects stall in the interconnection queue, lose their tenant mid-cycle, fail to secure capital, get caught in moratoria, run into environmental review delays, or simply sit on entitled land for years waiting for a buyer that never arrives. Industry tracking by Data Center Dynamics, Data Center Frontier, and DCCKnowledge routinely flags projects that were announced with substantial fanfare in one year and are still “under development” or quietly shelved three to five years later. In Georgia specifically, Southern Company’s Q3 2025 disclosure of 12 gigawatts in “advanced negotiations” is a large number precisely because the utility knows that only a fraction of advanced negotiations become signed large-load contracts, and only a fraction of signed contracts become operating data centers on schedule. The gap between announcement and operation is filled with stalled projects, delayed projects, and quietly dead projects. A community that is being asked to approve a zoning change today against a project that may not operate until 2030, if ever, is being asked to accept a transaction risk that even the developer’s own pro-forma does not try to quantify in public.

That is the third thing both camps are missing. The tax-counters are counting money from a facility that may never exist. The Google-believers are expecting a tenant who has not, in any public record, been identified. Both are standing on ground that the developer’s own filing quietly concedes is provisional.

What both camps share is a frame that does not match the play. The frame is: “a data center is coming, and the question is how good or bad it is.” The play is: “a developer is racing an entitlement clock, and the question of who the operator is will be answered, if it is answered, sometime later.” Those are not the same question. The first is about a finished deal. The second is about a deal that is not yet finished, and whose final shape will be decided by whoever writes the check after the zoning is approved.

---

VIII. The Institutional Backdrop

The state policy context that surrounds Project Arrowhead is as important as the project itself. Georgia approved its 2025 Integrated Resource Plan through the Public Service Commission on December 19, 2025, authorizing 9,885 megawatts of new-build generation and introducing what is known as the “100 MW rule,” a regulatory mechanism meant to manage the large-load contract cluster forming as data center demand accelerates. Southern Company, Georgia Power’s parent, disclosed in its Q3 2025 earnings that it had roughly 12 gigawatts in advanced negotiations for large-load customers, with a committed build-out of 10 gigawatts of new capacity by 2030. The Georgia PSC docket (44280, with significant filings on February 27 and March 20, 2026) catalogs the queue of contracts that will determine which of those negotiations materialize. Arrowhead’s position in that queue, if it is there at all, is not yet disclosed.

What is disclosed, and it is a meaningful tell, is that Georgia Power is actively conducting an engineering study on the 1.25 gigawatt load that Project Arrowhead has requested. A Georgia Power engineering study at that scale is not a casual data request. It is a paid, structured, multi-month utility workstream with a low-six-figure minimum price tag, performed by the utility only after a customer has made a formal request and committed to the deposit. Someone, on the developer side of Arrowhead, has already written the check for that study. The identity of the payer is not in the public record. But the existence of the study means one thing unambiguously: the project is far enough along that a utility-grade engineering workstream has already been commissioned and is running. That does not happen on a speculative pitch with no operator behind it. It happens when a serious institutional actor, behind the shell, has decided the project is real enough to spend real money on.

The timing tells its own story. The Phase 1 completion date on the DRI is 2030, and the full buildout 2034. That 2030 number is not incidental. The restrictive use covenants on the Sirrom parcels, documented in the deed records, are structured to expire in 2029. Phase 1 complete 2030, covenants off the land 2029: the data center’s first operational year lines up exactly with the first year the land is legally unencumbered to host it. A utility engineering study running now, a 2029 covenant expiry, and a 2030 Phase 1 completion do not read as three independent schedule facts. They read as a single coordinated timeline. Either the developer team planned the buildout to slip cleanly in behind the covenant release, or the covenant release date has been the real timeline lever all along. The public record does not resolve intent, but the alignment is too clean to ignore.

Meanwhile, at the federal level, the buildout economics of rural data center deals depend on transmission interconnection timelines that have stretched from roughly 18 months a decade ago to 4 to 7 years in PJM, MISO, and SERC queues today. A site with an abandoned solar project already studied for transmission proximity arrives with a head start that greenfield sites cannot match. That head start is part of why Pine Gate Renewables’ bankruptcy matters far beyond Pine Gate.

---

IX. The Failed Solar Deal, and Why It Is a Warning Sign

This is the tradecraft lesson that distinguishes an Arrowhead-aware community from a community that will be surprised by the next one.

The Sirrom Farms parcels carried a recorded Memorandum of Option in favor of Reedy Solar beginning in February 2020. The option passed through Samsung C&T and eventually into the hands of Pine Gate Renewables, one of the largest utility-scale solar developers in the Southeast. For more than four years, the land was nominally “slated for development” as a solar project. The option work built a scaffolding that nobody outside of those two families would have seen: survey lines, easement negotiations, environmental baselines, transmission proximity studies, and, most importantly, relationships. An agricultural broker, most plausibly from the CBRE Agricultural Services line of business, was likely inside that scaffolding from an early stage: interfacing with the Morrises, coordinating with Pine Gate, handling the landowner side of the option.

On November 6, 2025, Pine Gate Renewables filed Chapter 11 bankruptcy in the Southern District of Texas (Case 4:25-bk-90669). On February 19, 2026, the bankruptcy court confirmed Pine Gate’s plan. The Sirrom option was abandoned or rejected, like many other solar options in Pine Gate’s portfolio across multiple states. Six days later, Premier Data Center Development, LLC was formed. The day after that, the Planning Commission recommended the project 4-0.

Read the sequence as a single move. A broker with four years of relationship capital, parcel intelligence, transmission analysis, and family trust on a rural piece of ground is, the day that solar encumbrance dies, in possession of the single most valuable thing in the current data center market: an entitlement-ready site with warm-body landowner relationships and a pre-built infrastructure story. The CBRE Agricultural Services hand can pass that package, internally, to the CBRE Data Center Solutions practice where Mike Lash sits. That pass is not speculative; it is the ordinary architecture of a commercial real-estate firm that happens to have both practices under one roof. Whether it happened here, in the exact form described, is an analyst inference the public record supports as plausible but does not confirm.

There is one more detail that makes the Sirrom role on Arrowhead specifically load-bearing, and it is the one most likely to be missed. The original Planning Commission filing covered only the Fletcher Trust parcels. The state-level refile doubled the footprint precisely by pulling Sirrom in. Sirrom is where the critical-path road access to the campus runs. Sirrom is where the site plans place the substation ground that every hyperscale data center requires. And Sirrom is the land that spent four years being pre-engineered for utility-scale solar, which means its transmission proximity and substation footprint had already been studied, mapped, and walked. The developer team did not just add acreage in the refile. They added the precise piece of land that the failed solar project had already engineered for power delivery. An entirely greenfield substation and road design would have taken a year of civil engineering. Sirrom’s solar scaffolding gave them most of it for free. The failed solar project was not merely a relationship into Sirrom. It was, in the civil-engineering sense, half the work already done.

Pine Gate Renewables’ multi-state bankruptcy is not an Irwin County footnote. It is a southeast-wide leading indicator.

The lesson that generalizes is this. If a community sees a failed or stalled utility-scale power project on rural Georgia agricultural land, whether solar, wind, gas peaker, or biomass, in the five years prior to a data center proposal, that same land is a candidate for a data center follow-up play. The relationships are there. The parcel intelligence is there. The broker network is there. The transmission studies are there. The option work is already paid for. All that is needed is a tenant pool and a shell.

Pine Gate filed Chapter 11 with assets across the Carolinas, Georgia, Tennessee, and beyond. Every one of those bankrupt options is a candidate signal for the next Project Arrowhead. Local governments watching for data center encroachment should be reading their bankruptcy court filings as seriously as they read their rezoning agendas, because the entitlement clock for the next fight started the day the solar developer filed its petition.

---

X. What We Cannot Say

Intellectual honesty is the coin of this kind of work. Here is what the public record does not yet show and what this article does not claim.

We cannot name the backer. QTS Data Centers (Blackstone) is the leading working theory on the basis of the three-provider service nexus (Lash, Pouncey/Taft, Kimley-Horn) and the Fayetteville development-pattern match. Prologis remains a secondary hypothesis on the Coweta pattern. Neither QTS nor Prologis appears in any Arrowhead public filing. The absence of confirmation is itself consistent with a Tier 4 play in which the backer has not yet been disclosed by design, and with QTS’s documented practice of operating through unnamed front entities for three to five years before surfacing.

We cannot name the anchor tenant. No tenant is listed on the DRI. No tenant name is in the WWALS archive. No CBRE deal sheet referencing Arrowhead has surfaced. No corresponding Azure, Google, AWS, or Meta press release has been identified. “Google” is local rumor, and this article records it as such.

We cannot confirm whether Russ Bryant has equity in Premier DCD, only that his personal phone and email are the DRI point of contact for the developer.

We cannot read the motive for the Irwin County Industrial Development Authority’s dissolution. The county said it was about checks and balances and specifically denied a connection to the data center application. The timing is suggestive; the paper trail is not yet retrievable.

We cannot confirm whether the CBRE Agricultural Services hand who likely brokered the original Sirrom solar option is the same individual or on the same team as the CBRE Data Center Solutions hand who is now visible at hearings. The internal architecture of CBRE’s practices makes the pass plausible. The pass itself is not documented.

All of these are live open questions. Some of them will resolve in the next 60 to 180 days as the DRI regional review proceeds, as a BOC vote eventually happens (if Arrowhead returns in local-zoning form), and as the inevitable post-approval disclosures in the trade press land. Some of them may never resolve cleanly. A community making a decision about its own land cannot wait for every open question to close. It can, however, decide on the basis of what type of play it is facing, which is knowable today, rather than on the basis of who the tenant turns out to be, which may not be knowable for another year.

---

XI. An OSINT Toolkit for Communities

These are the moves any community can make, without a budget, to read its own data center fight. They are ordered by ease.

1. Georgia Secretary of State LLC formation lookup. Search `corporations.sos.ga.gov` for the applicant entity. Note formation date, organizer, registered agent, registered agent city, mailing address. A formation date within 60 days of a zoning action is a Tier 4 tell. An organizer who is a local real-estate attorney (not an in-house hyperscaler counsel) is a Tier 4 tell. A registered agent of record with CT Corporation System or Corporation Service Company is a default shell tell, neither incriminating nor exonerating. A Statham or Dacula mailing address for a project in Irwin County is a Tier 4 tell.

2. DRI portal. `apps.dca.ga.gov/DRI/AppSummary.aspx` lists every Development of Regional Impact filing in the state, with initial and additional forms attached. Read the “Continuation of Previous DRI” field. Read the developer contact. Read whether an anchor tenant is named. A “No” on continuation paired with a recent local-zoning withdrawal is a soft tell that the project has moved out of local control and into state review, which is usually favorable to the developer.

3. Georgia Department of Audits incentive reports. `audits.ga.gov` publishes the statewide economic-impact reviews and their revisions. The January 2026 revision to the December 2025 data center report is the one that matters for current-year decisions. Read both. The delta between them tells a community what the realistic job and revenue numbers look like after peer review.

4. Good Jobs First Subsidy Tracker. `subsidytracker.goodjobsfirst.org` catalogs every major subsidy and abatement deal in the country. Search the county and the state for prior data center awards. The per-job costs are the sharpest rebuttal to an overoptimistic local-revenue pitch.

5. County planning commission video archives. In south Georgia, the WWALS Watershed Coalition (`wwals.net`) has become the single most useful citizen archive for data center-adjacent hearings. In north Georgia, county government YouTube channels and DeSmog’s investigative coverage play the same role. Watch the hearings. Identify every person who walked up to the podium. Cross-reference names against CBRE’s, JLL’s, and Cushman & Wakefield’s Data Center Solutions team pages.

6. Broker-of-record visibility. The public team pages at CBRE, JLL, and Cushman & Wakefield’s data center practices list the brokers working the Georgia market. If one of those names appears in a hearing video for your county, the call is almost always a commercial-capital-backed play, not a local landowner-initiated project. CBRE’s `cbre.com/people` pages are searchable.

7. General contractor mobilization on LinkedIn. Mortenson, DPR, Holder, and Turner are the four general contractors most frequently retained on hyperscaler-direct (Tier 1) and hyperscaler-via-developer (Tier 2) data center projects in the Southeast. Their LinkedIn job postings for project engineers, superintendents, and field supervisors in a specific county, dated three to twelve months before a rezoning hearing, are a near-certain tell that a hyperscaler-grade project is real. The absence of such postings for Arrowhead is one of the most important negative findings in this report.

8. USPTO trademark filings. `uspto.gov` is searchable for project names and campus brandings. “Project Arrowhead” as a trademark would be an escalation signal. It has not been filed.

9. Georgia PSC dockets. `psc.ga.gov` publishes large-load contract filings through Docket 44280. Large-load customers for 100+ MW deals file interconnection requests that eventually become public.

10. Service-provider nexus mapping. Identify the broker, the attorney, and the civil engineer on the project. Then search each one against known data center operator engagements in public record, DRI filings, press releases, and court dockets. If two or more of the three have documented relationships with the same operator, that operator is the working hypothesis. The Arrowhead example: all three visible service providers (CBRE’s Lash, Taft’s Pouncey, Kimley-Horn) have documented QTS ties. Any one alone would be unremarkable. Three together is a pattern. Map the nexus before you name the tenant.

11. The failed-power-project warning sign. Cross-reference the parcel in question against PACER bankruptcy filings, the Department of Energy’s LPO database, and state utility commission abandonment dockets. Any failed or abandoned utility-scale power project, solar, wind, gas, or biomass, on the same land or on contiguous land, in the prior five years, is the strongest single predictive tell we have identified for a data center follow-up play. Pine Gate Renewables’ multi-state portfolio of abandoned options is the current hot zone. So is the long tail of stranded solar interconnection applications at every Southeastern independent system operator.

None of these moves produces a smoking gun. Taken together, and read against the seven-tier typology, they give a community the language to describe the play it is actually facing, in time to make a decision with its eyes open.

---

XII. Back in the Room

On March 17, 2026, the Irwin County Board of Commissioners held a public hearing on a draft data center ordinance. A closed-loop cooling provision was on the agenda, one of the single most important protections a rural county can write into its ordinance before a project lands. The WWALS Watershed Coalition has the video. The ordinance has not been finalized. Whether it is finalized before or after DRI #4689 completes regional review will be one of the single most consequential decisions Irwin County makes this year.

Patrick Atwater is still on his 400 acres. Marvin Peavy is still on his 1,500. Russ Bryant is still a landowner neighbor on the same road, and his phone is still the number on the DRI. Angela Bryan is still the trustee of her father’s estate. The Morris family is still on Sirrom. The land has been “going somewhere” since February 2020: a solar project that was never built, a data center campus that may or may not be built, a set of restrictive covenants that expire in 2030. The pattern is not that the land was ever going to stay where it was. The pattern is that the relationships that were built the first time were always going to be available for the second.

Jacqueline Lassetter, 80 miles north, went through a version of this fight eighteen months earlier, lost the 3-2 vote, and is now part of the public record of Project Sail, quoted in the DeSmog investigation that finally unmasked Prologis. The documentary trail she and her neighbors built through their own public-records requests is why Irwin County has a playbook to read from at all. That is how this work compounds. One community’s FOIA becomes the next community’s primer.

The fight in Irwin County is not with a phantom Google. It is not with a tax-revenue mirage. It is with a Tier 4 speculative developer-led play, with every professional service provider on the project carrying documented ties to QTS Data Centers, a Blackstone subsidiary that has already run this exact development pattern to completion eighty miles north in Fayetteville. No committed anchor tenant has been named. No QTS entity appears in any filing. But the broker is the same. The law firm lineage is the same. The engineer is the same. And the playbook, a shell LLC filing initial entitlements while the real operator remains unnamed for years, is the same playbook that took five years to produce a Microsoft tenant reveal in Fayetteville. The community is being asked to approve a zoning change on that basis, on land whose development scaffolding was paid for by a solar project that went bankrupt in another state. It is a play that has a shape. That shape is knowable. And communities that know the shape can make decisions that communities looking for a phantom cannot.

“The fight is not with a phantom Google or a tax-revenue mirage. The fight is with a play that has been run before, eighty miles north, by the same team, to a $4.6 billion bond offering. The community that knows the Fayetteville template can read its own fight clearly. The community that does not will be told a story about Google until the real operator is ready to surface.”

The people who were in that room on February 26 were not confused about whether something was happening to their county. They were clear. They were organized. They wore red shirts and they spoke for three hours. What they were missing, through no fault of their own, was the language for the kind of play that was happening to them. If this article has one job, it is to supply that language. The seven tiers. The diagnostic tells. The Coweta contrast. The Georgia Department of Audits revision. The failed-power-project warning sign. The CBRE broker of record. The shell LLC formed the day before the hearing.

Four-o against, and then pulled back. The next hearing will happen. The next filing is already in state review. When the community returns to the podium, they will return with a different piece of paper than the one they had last time.

That is what Vesper: Public Intelligence is for.

---

Sources

Primary sources are cited in the order they appear in the text. All are publicly accessible as of the date of publication.

1. WWALS Watershed Coalition video archive and hearing logs, `wwals.net`, specifically coverage of the February 26, 2026 Ocilla-Irwin Planning Advisory Commission hearing and the March 2, 2026 applicant withdrawal: https://wwals.net/2026/03/01/datacenter-recommended-approval-irwin-county-planning-commission-2026-02-26/ and https://wwals.net/2026/03/16/irwinville-data-center-withdrawn-by-applicant-irwin-county-commission-2026-03-02/

2. Justin Nobel et al., “How Data Center Developers Staked Their Claim in Rural Georgia,” *DeSmog*, April 7, 2026: https://www.desmog.com/2026/04/07/how-data-center-developers-staked-their-claim-in-rural-georgia/

3. Georgia Department of Community Affairs, DRI #4689 Project Arrowhead application summary and initial form: https://apps.dca.ga.gov/DRI/AppSummary.aspx?driid=4689

4. WALB, “2nd South Georgia County Moves to Dissolve Industrial Authority,” February 16, 2026: https://www.walb.com/2026/02/16/2nd-south-georgia-county-moves-dissolve-industrial-authority-what-it-could-mean-future-data-center-builds/

5. Georgia Secretary of State corporations database, Control Numbers 26043985 (Premier Data Center Development, LLC) and 12098617 (Sirrom Farms LLC): https://ecorp.sos.ga.gov/BusinessSearch

6. Pine Gate Renewables Chapter 11, Case 4:25-bk-90669, U.S. Bankruptcy Court for the Southern District of Texas, plan confirmed February 19, 2026: PACER.

7. Recorded Memorandum of Option, Reedy Solar on Sirrom Farms LLC parcels, Irwin County deed records, Deed Book 00410, page 00347, dated February 7, 2020.

8. Mike Lash public profile, CBRE Data Center Solutions: https://www.cbre.com/people/mike-lash

9. Kelly O. Faber professional profile: https://www.avvo.com/attorneys/30019-ga-kelly-faber-527625.html

10. Morris, Manning & Martin / Taft Stettinius & Hollister merger announcement, effective December 31, 2025; Gerald Pouncey transition to Senior Counsel.

11. Georgia Department of Audits and Accounts, Economic Impact of Data Center Development in Georgia, December 2025 report and January 2026 revision. Coverage: The Current GA, Capitol Beat News Service.

12. Georgia Public Service Commission, 2025 Integrated Resource Plan (approved December 19, 2025) and Docket 44280 (large-load contract cluster filings, including documents 225624 of February 27, 2026 and 225969 of March 20, 2026): https://psc.ga.gov

13. Southern Company Q3 2025 earnings call disclosure on 12 GW advanced negotiations and 10 GW 2030 build-out commitment.

14. Prologis Q4 2025 earnings call transcript, data center development pipeline disclosure: https://www.theglobeandmail.com/investing/markets/stocks/PLD-N/pressreleases/37156844/prologis-pld-q4-2025-earnings-call-transcript/

15. Propmodo, “Prologis Wins Approval for Massive Georgia Data Center Project”: https://propmodo.com/prologis-wins-approval-for-massive-georgia-data-center-project/

16. Sebastian Moss and Dan Swinhoe, *Data Center Dynamics* coverage of DRI #4689: https://www.datacenterdynamics.com

17. *Newnan Times-Herald* editorial pages, Prologis / Kent Mason op-ed of May 2025.

18. *The Citizen* and *Newnan Times-Herald* coverage of the Coweta County Commission 3-2 approval of Project Sail, April 10, 2026.

19. Microsoft Azure blog, East US 3 region announcement, Fulton and Douglas counties: https://azure.microsoft.com/en-us/blog/microsoft-will-establish-its-next-u-s-datacenter-region-in-georgia-s-fulton-and-douglas-counties/

20. *Atlanta Journal-Constitution* coverage of the Microsoft Fayetteville AI Superfactory and QTS Fulton County abatement: https://www.ajc.com/business/2025/11/microsofts-newest-ai-superfactory-opens-at-sprawling-fayetteville-campus/

21. Good Jobs First Subsidy Tracker, data center awards: https://subsidytracker.goodjobsfirst.org

22. Timothy J. Bartik, Upjohn Institute for Employment Research, “Who Benefits from State and Local Economic Development Policies?” and subsequent incentive-elasticity work.

23. Change.org petition, “Protect Irwin County from DATA CENTERS,” February 5, 2026.

24. Vesper: Public Intelligence, *The Digital Land Grab: Georgia’s Data Center Wars* (prior Substack edition), for the Georgia Department of Audits 90 percent finding.

25. KKR closes $2B investment in Compass Datacenters, March 16, 2026 (announced December 5, 2025); advised by Citi for KKR, Deutsche Bank for Compass.

26. *Atlanta Business Chronicle*, “Westside development could threaten BeltLine loop connection, officials say,” July 23, 2021 (naming Carl Westmoreland of Morris, Manning & Martin as QTS attorney on BeltLine easement): https://www.bizjournals.com/atlanta/news/2021/07/23/qts-realty-data-center-atlanta-beltline.html

27. CBRE press release and Metro Atlanta CEO, “CBRE Facilitates Sale of 615 Acres in Fayetteville, GA for Largest Multi-Tenant Data Center Campus in the World,” August 2022 (naming Mike Lash and Tim Huffman as CBRE brokers on QTS Fayetteville acquisition): http://metroatlantaceo.com/news/2022/08/cbre-facilitates-sale-615-acres-fayetteville-ga-largest-multi-tenant-data-center-campus-world/

28. Georgia DCA DRI #3160 (Oceanic Data Centers / Fayetteville, 2020), DRI #3813 (National Acquisition Co LLC / Fayetteville, 2022), DRI #4603 (QTS Eastwood LLC / Fayetteville expansion, 2026): https://apps.dca.ga.gov/DRI/Submissions.aspx

29. *The Citizen* (Fayetteville), coverage of QTS land sale ($153.8M), opposition, and rezoning: https://thecitizen.com/2022/08/04/fayetteville-data-center-site-sells-for-153-million-including-75-million-to-development-authority/

30. *DataCenterDynamics*, “QTS targets two data center campuses in Georgia,” March 13, 2026 (Blakely and Fayetteville expansion): https://www.datacenterdynamics.com/en/news/qts-targets-two-data-center-campuses-in-georgia/

31. *DataCenterDynamics*, “QTS files to expand Fayetteville data center campus in Georgia,” April 2026: https://www.datacenterdynamics.com/en/news/qts-files-to-expand-fayetteville-data-center-campus-in-georgia/

32. Bloomberg, “Blackstone-Backed QTS Kicks Off Bond Sale,” April 6, 2026 (QTS $4.6B green bond, Microsoft named in offering materials): https://www.bloomberg.com/news/articles/2026-04-06/blackstone-backed-data-center-firm-qts-kicks-off-bond-sale

33. *AJC*, “From peanuts to processors? South Georgia gets largest data center pitch yet,” March 2026 (QTS Blakely / Early County): https://www.ajc.com/business/2026/03/from-peanuts-to-processors-south-georgia-gets-largest-data-center-pitch-yet/

34. *AJC*, “Microsoft’s newest AI superfactory opens at sprawling Fayetteville campus,” November 2025 (Microsoft Fairwater tenant reveal): https://www.ajc.com/business/2025/11/microsofts-newest-ai-superfactory-opens-at-sprawling-fayetteville-campus/

35. Taft Stettinius & Hollister, “Taft Completes Merger with Morris, Manning & Martin,” effective December 31, 2025: https://www.taftlaw.com/news-events/news/taft-completes-merger-with-morris-manning-martin/

36. Kimley-Horn QTS Fayetteville Transportation Impact Study, referenced in DRI #3813 filing and WSJ/Yahoo coverage of Project Excalibur.

---

Vesper: Public Intelligence is an open-source intelligence publication for public distribution. This piece is based solely on publicly available sources. No private, commercial, or client work is represented in this report. Inferences are labeled where they depart from primary-source evidence. Corrections and additional public documents are welcomed at the publication’s public intake.