Trump officials announce data center and gas plants for former Ohio uranium site

A 10-gigawatt data center and up to 10 GW of new power generation (including 9.2 GW of natural gas) were announced for the Portsmouth 'PORTS Technology Campus,' with DOE saying construction will begin this year. SoftBank (via SB Energy) and AEP Ohio will build power and transmission infrastructure, including $4.2B in grid upgrades and $33.3B in Japanese funding tied to the gas component, and officials project thousands of jobs and excess capacity fed back to the grid to lower regional costs. The initiative is part of the U.S.-Japan Strategic Trade and Investment Agreement to expand AI infrastructure, though local opposition and a petition to ban mega data centers were highlighted.

Analysis

This project functionally creates a playbook for private capital to internalize the biggest cost and permitting friction of hyperscale AI sites — power and transmission — which compresses a key barrier to new large-scale data center supply. That lowers marginal hosting prices regionally over the medium term (12–36 months) as developers can offer deeper long-term PPA/infra bundles, pressuring incumbent cloud margins for commodity AI workloads while increasing demand for high‑value software and services layered on that infrastructure.

Oracle and its infrastructure partners have optionality beyond raw compute supply: they can capture long-term software licensing, managed services, and migration fees as new campuses sign multi-year deals. Hyperscalers that are not part of these strategic deployments face two second-order threats — accelerated competition for committed capacity and a need to underwrite more localized generation or accept higher grid exposure — which will show up in slower margin expansion for cloud IaaS hosting rather than immediate revenue declines.

Main policy and execution risks are political pushback (local ballot measures), state/regulatory cost recovery disputes, and equipment/permitting lead times; any of these can slip projects into 18–36 month timelines or force capital structure changes. Watch near-term catalysts: financing closings, PUCO/FERC filings, and partner MOUs — each is a discrete event that can re-rate risk premia and reveal whether claimed “no rate impact” transmission funding is real or shifted to ratebases.

Key Decisions for Investors

• Buy ORCL 12–24 month LEAPS (call or call-spread) to capture software/attach upside from infrastructure deals — target +30% move over 12–24 months for ~3:1 upside/downside (stop -25%). Rationale: Oracle is best positioned to monetize long-term licensing and managed services as campuses lock-in customers.
• Pair trade: Long ORCL / Short GOOGL (or GOOG) 6–18 months, size ~1:0.6. Trade objective: capture relative re-rating as infrastructure-linked software captures a larger share of AI hosting economics; take profits on 20–30% pair divergence, stop if GOOG rallies >15% on clear exclusivity/partnership announcements.
• Hedge political/regulatory tail: Buy 6–12 month put protection on META (put or put-spread) to guard against a near-term wave of local moratoria or higher regional power costs that could pressure incremental margins on AI/hosting and capex outlook. Risk/reward: limited premium outlay for protection against a 15–25% swing in sentiment-driven downside.