The anchor-tenant paradox: what Microsoft's Licking County pause did to every pro forma around it

Everyone in this business is trained to fear the county. The angry hearing, the moratorium, the commissioner with a grudge. Licking County, Ohio spent 2024 and 2025 demonstrating that the county is sometimes the least of your problems, and the demonstration cost Microsoft's neighbors more than it cost Microsoft.

The facts run like a controlled experiment. In October 2024, Microsoft announced $1 billion across three central-Ohio data-center sites: $420 million earmarked for New Albany, where the city council approved a 15-year tax abatement, plus sites in Hebron and Heath. One month later, on November 20, 2024, Cologix announced it had acquired 154 acres in Johnstown, in the same county, from an affiliate of the New Albany Company, planning an eight-building, 800 MW campus of roughly 2.0 million square feet at over $7 billion. Two hyperscale bets, one county, thirty days apart.

By April 2025, Microsoft had paused all three of its sites. Cologix kept building.

If the entitlement was clean, what failed?

Microsoft's entitlement was genuinely clean. That is what makes the Licking County pause file uncomfortable reading. The abatement was approved. The zoning worked. What failed was the thesis underneath: co-locate compute next to Intel's announced $28 billion semiconductor megafab and ride the Silicon Heartland corridor. In early 2025, Intel publicly pushed the fab's commercial launch to at least 2030, and the anchor that justified central-Ohio timing stopped anchoring anything on this side of the decade. Stack tariffs on imported construction materials on top, plus a global Microsoft statement about "slowing or pausing some early-stage projects," and the $1 billion became a pause notice. Microsoft committed to honoring its existing road and utility agreements, said the Hebron and Heath land might return to farming for now, and separately noted it still expected to spend more than $80 billion globally on infrastructure in the same fiscal year.

Read that combination again. Not a retreat from the asset class. A repricing of one region's timing, driven by a different company's construction schedule.

Licking County Commissioner Tim Bubb called Microsoft's explanations "vague," and referenced the company reevaluating its need for capacity expansion. That is what it sounds like when a local government discovers it holds anchor-tenant risk it never underwrote. New Albany's 15-year abatement, for what it is worth, remains in place; the city's position is that it delivered on its commitments and the pause is a corporate decision, not a municipal failure. True, and cold comfort. An abatement on a paused pad produces neither jobs nor the political story the council approved it for.

Our pre-announcement read on the Licking County thesis was 42 out of 100, and the largest single deduction was the anchor-dependency line, because Intel's delivery record on the fab timeline was public before Microsoft committed a dollar.

Why did Cologix proceed while Microsoft paused?

Partly because the two companies were making different bets dressed in similar clothes. The Cologix Johnstown file shows an operator with four existing Columbus data centers totaling 500,000 square feet and 80 MW, buying from an aligned seller, inside a city whose mayor, Donald Barnard, publicly called the deal "confirmation that our updated planning policies are paying off." Johnstown had pre-built its zoning envelope for exactly this filing. First-phase construction was anticipated to begin in 2025, phased, matched to Cologix's own commercial rhythm rather than to Intel's.

The structural read: Microsoft's Licking thesis was a derivative of Intel. Cologix's Johnstown thesis was a continuation of Cologix. When the anchor slipped, the derivative repriced and the continuation did not.

Neither project escapes the state-level weather, to be fair. Policy Matters Ohio published a critical analysis of Ohio's 75% data-center sales-tax exemption in January 2025, under a title that called the tax breaks indefensible, and an organized statewide critic is a durable feature of any 15-to-20-year Ohio hold. That risk applies to Cologix and Microsoft alike. It just is not the risk that separated their 2025 outcomes.

And the anchor effect is real when it works, which is the paradox's other half. Next door, the New Albany corridor file documents what anchor gravity built: the city unanimously rezoned 1,689 acres into a purpose-built Technology Manufacturing District in 2022 after Intel's fab announcement, and the district has never denied a data-center application. Google went from $600 million in 2019 to $1.7 billion. Meta expanded to $1.5 billion and now has Project Prometheus, a 1-gigawatt cluster, under construction. AWS committed $3.5 billion and then bought 400 more acres for $116 million in 2025. Four hyperscalers and more than $15 billion, without a single denial.

Same anchor. It pulled $15 billion into one corridor and stranded $1 billion in the same county.

What did the pause do to everyone else's numbers?

This is the part the headlines skipped. Every land option, every speculative industrial pad, every gas-plant proposal, every township road budget within the Microsoft blast radius had quietly booked hyperscaler anchor demand as a given. Township hearings on data centers and the two proposed natural-gas plants in western Licking County were drawing standing-room-only crowds by late 2024, and Signal Ohio's December 19, 2025 feature framed the whole Ohio boom as running into political resistance, with transmission costs as a public grievance. The pause handed that opposition its best exhibit: even the demand is not certain. New Albany's council can fairly say it delivered everything it promised. The pro formas around the paused sites cannot say the same, because their load-bearing assumption lived in a Redmond capital-allocation meeting they had no visibility into.

Anchor adjacency is a demand assumption wearing an entitlement costume.

So underwrite it as demand. Pull the anchor out of your model and see what the site is worth without it; that number is your floor, and the gap between floor and headline is exactly the exposure you have to somebody else's earnings call. Ask when the anchor's own infrastructure actually delivers, because Intel's track record on the fab timeline was public information before October 2024, and a thesis predicated on a 2025-2026 commercial launch always had to carry that delivery risk. Any central-Ohio screening read after early 2025 should assume 2030, and price township-level politics one township at a time rather than treating Licking County as a single jurisdiction with a single mood.

My read is that the market is still mispricing anchor risk in the corridors where the anchor is a promise rather than an operating plant. I expect at least one more announced hyperscale project in the Columbus metro to pause or shrink before Intel's fab reaches commercial operation, and I expect it to happen on a site whose zoning was never in doubt. If everything announced in central Ohio through 2027 proceeds on schedule, I got this one wrong and the region got a miracle. Before your next anchor-adjacent deal closes, write down the anchor's delivery date and whose signature it depends on. If the signature is not yours, that line is risk, and it belongs in the model at a number bigger than zero.

This analysis is a source-cited research summary drawn from public records, not legal advice. It can contain errors and should be verified independently before any investment decision.