This corpus (AskAmerica/askengine) has no pharmaceutical-manufacturing-facility table — search_catalog confirms "pharmaceutical" matches nothing in any schema (economic tables here go only to NAICS-sector/industry aggregates, e.g. econ.state_industry, econ.industry_gdp, which can't identify individual facilities or construction status). The answer below is built entirely from company disclosures, local/state government announcements, trade press (FiercePharma, Contract Pharma, PharmExec, Construction Dive), and a UNC Kenan-Flagler tracker built specifically to separate "broken ground" from "announced only" for tariff-exemption purposes. That tracker's table was the single most useful primary source and is cited directly below.
Real, shovels-in-the-ground activity is concentrated in three clusters, plus a scatter of single-site expansions — and it is running well ahead of what most headlines about "reshoring" imply, but a meaningful share of the biggest-dollar announcements (Lilly's Alabama and Virginia API plants, several Novartis/Roche/Sanofi/UCB sites) are still pre-construction as of this writing.
1. North Carolina's Research Triangle / "Pharma Corridor" (Holly Springs–Clayton–Sanford–Wilson–Durham) — the most active zone in the country. At least eight distinct facilities have actually broken ground or are under active construction here, several already partially operational: - Fujifilm Biotechnologies, Holly Springs — $3.2B biologics site; Phase 1 opened Sept 2025, drug-product line coming online in 2026, Phase 2 (doubling capacity) under construction for 2028. - Novo Nordisk, Clayton — $4.1B second fill-finish plant, under construction, targeting 2027–2029. - Amgen, Holly Springs — $1.5B drug-substance plant, under construction. - Genentech/Roche, Holly Springs — $700M metabolic-medicine plant, under construction, 2029 target. - Kyowa Kirin, Sanford — $530M biologics plant, under construction, 2027 target. - Johnson & Johnson, Wilson — $2B plant, under construction. - Merck, Durham — $1B vaccine facility, under construction (2025 target). - Biogen, NC — $2B antisense-oligonucleotide plant, under construction.
2. Indiana's Lilly footprint (Indianapolis + the new "LEAP" district in Lebanon). Both are actively under construction, not just announced — $4.5B in Indianapolis (targeting late 2027) and $4.5B in Lebanon (targeting 2027). This is the largest concentration of confirmed Lilly build-out outside its newer 2025-26 API announcements.
3. Chicago-area API cluster (AbbVie, North Chicago). AbbVie actually broke ground on a $195M API plant in fall 2025 (target: operational 2027), and followed with a second, larger $380M two-plant API expansion on the same campus with construction starting spring 2026 (target 2029) — one of the only confirmed under-construction API-specific (as opposed to biologics/fill-finish) sites outside Texas/NC.
4. Scattered single-site builds actually underway: Merck biologics/ADC plant, Wilmington DE ($1B); GSK vaccines plant, Marietta PA ($800M, targeting 2028); Roche glucose-medicine plant, Indianapolis ($550M); Vetter fill-finish, Des Plaines IL ($285M, targeting 2029); Hikma, Columbus OH; Gilead biologics, Foster City CA (targeting 2026).
5. Eli Lilly's Houston, TX API/GLP-1 mega-site ($6.5B) has moved from announcement to real construction — site permits were filed and construction activity was confirmed starting mid-2026, though the company itself still frames the full buildout as running to 2030.
Both clusters had existing pharma-manufacturing infrastructure, state/local incentive packages already in place, and multiple companies co-locating (shared labor pool, supply chains, utilities) — which is why announcement-to-groundbreaking conversion has been fastest there. By contrast, greenfield API announcements in Alabama and Virginia (states without an existing large-scale pharma manufacturing base) are taking longer to convert from announcement to visible construction.
(Note: search_catalog in the AskAmerica connector was queried and returned no pharmaceutical-manufacturing table; this finding — a genuine corpus gap — is reported rather than papered over with an ill-fitting proxy.)