$3 billion AI data center coming to the Upstate

A $3 billion AI data center is planned for the Upstate, according to WYFF Greenville, representing a sizeable capital investment into regional technology infrastructure. While details on the developer, timeline and incentives were not provided, the project could meaningfully boost demand for construction, power and local suppliers and create opportunities for utilities, real-estate and infrastructure vendors in the region.

Analysis

Market structure: A $3B AI data‑center build in Greenville primarily benefits AI GPU suppliers (NVDA, AMD), large-scale data‑center REITs/colocation operators (DLR, EQIX, QTS) and engineering/construction contractors (J, FLR). Locally it boosts demand for power/transformer capacity (NEE, ABB) but increases regional supply — pressure on smaller regional colos and spot wholesale pricing in the Southeast could rise within 6–18 months.

Risk assessment: Tail risks include energy grid constraints or rate shocks that raise opex by >20%, GPU supply shortages delaying monetization, or local permitting pushback causing >12‑18 month delays. Immediate (days) impact is sentiment; short term (weeks–months) is supply‑chain and contractor revenue; long term (quarters–years) is utilization rates and pricing power for colocators.

Trade implications: Favor long positions in NVDA (6–12 months) and selective data‑center REITs (DLR, EQIX) while hedging energy/capex risk via utilities (NEE) or power equipment names (ABB). Consider relative‑value: long large diversified REITs vs short smaller/regionally exposed peers (CONE/QTS) for 3–9 months; use options (3–6 month call spreads or put spreads) to control capital and gamma exposure.

Contrarian angles: The market underestimates grid/inflation drag and overestimates local pricing power — several $3B builds nationally could produce temporary overcapacity and compress FFO by 5–15% in weak demand cycles. Historical parallels (2017–2019 colo expansions) show utilization lags; hedge positions and wait for utilization data (first 12 months) before levering up.

Key Decisions for Investors

• Establish a 2–3% long position in NVIDIA (NVDA) over the next 4–8 weeks for a 6–12 month time horizon; set a stop at 15% below entry and a profit target of ~40% within 12 months, size options exposure with a 0.5–1% notional 3‑month 20% OTM call spread to cap downside.
• Allocate 2% to Digital Realty (DLR) and 1% to Equinix (EQIX), dollar‑cost over 4 weeks; take profits if combined share price gains >25% or if REIT guidance on FFO falls >5%, implement 12% stop‑loss on each position.
• Implement a pair trade: long DLR 1.5% vs short CyrusOne (CONE) 1.5% for 3–9 months to express scale/contract diversification over regional exposure; unwind if relative performance gap exceeds 15% or if 10‑yr Treasury yield falls >50 bps.
• Buy downside protection: purchase a 6‑month put spread on a data‑center REIT ETF or DLR sized to 0.5–1% notional (e.g., 5–10% OTM put spread) and add 1% long NextEra Energy (NEE) to hedge potential grid/energy risk and rising opex; exit hedges if implied vol falls >20% or utilization data in 12 months confirms demand.