SoftBank to invest $3 billion in Ohio factory for OpenAI data center, The Information reports

SoftBank will invest up to $3 billion to convert the former Lordstown, Ohio EV plant into a facility to build modular data-center equipment for OpenAI, with manufacturing due to begin early next year and a small working data-center on site as a demonstration; the units are intended to serve OpenAI sites including Milam County, Texas. The move builds on SoftBank’s $375 million purchase of the site in August and follows the sale of a $5.8 billion Nvidia stake, and is tied to the Stargate plan (five U.S. AI data centers, ~$500 billion) and a broader White House–announced joint venture in which SoftBank pledged $18 billion. The deal materially advances industrial-scale deployment of AI infrastructure at a time when OpenAI is targeting roughly 30 GW of capacity (estimated cost ~$1.4 trillion) but faces very high per‑GW costs and no advertising or cloud business to underwrite the buildout, so SoftBank’s commitment could hasten capacity rollout while concentrating execution and financing risk.

Analysis

SoftBank plans to invest up to $3 billion to convert the former Lordstown, Ohio EV plant—which it bought in August for $375 million—into a factory to build modular data-center units for OpenAI, with manufacturing expected to begin early next year and a small working data center on site as a demonstration. The move is explicitly tied to the Stargate initiative alongside OpenAI and Oracle to build five U.S. AI data centers (the report cites a ~$500 billion plan) and sits within a broader White House‑announced joint venture in which SoftBank pledged $18 billion.

This follows SoftBank’s sale of a $5.8 billion Nvidia stake to fund CEO Masayoshi Son’s AI push; OpenAI’s public targets cited by the article include building roughly 30 gigawatts of capacity at an estimated $1.4 trillion cost and ultimately adding about 1 gigawatt per week, while current costs run above $40 billion per gigawatt. The Lordstown conversion accelerates vertical integration and deployment speed via modular, portable units that can be scaled to Milam County, Texas and other sites.

Key risks are concentrated execution and financing exposure: SoftBank and OpenAI are shouldering large capex without OpenAI having advertising or cloud revenues to underwrite the rollout, which increases the likelihood of further capital raises, partnership shifts, or execution delays. Market signals in the accompanying data label sentiment as mildly positive but speculative, so near‑term valuation action may depend on delivery milestones and cost reductions.

Key Decisions for Investors

• Consider selective, sized exposure to SoftBank‑linked securities to capture upside from an accelerated AI infrastructure play, but limit position size given concentrated capex and execution risk
• Monitor near‑term execution milestones—start of manufacturing early next year, first deliveries to Milam County and demonstration center commissioning—as binary catalysts for sentiment and valuation repricing
• Avoid assuming OpenAI will self‑fund large buildout costs; prepare for potential dilution or additional asset sales and prefer liquid hedges or options if holding long positions tied to this rollout
• Track related suppliers and Oracle partnership updates for early revenue signals from modular unit production, and reallocate quickly if cost‑per‑GW or timeline deterioration is reported