Bloom Energy (BE) shares jumped 22% on April 14, 2026, after the company announced an expanded partnership with Oracle to supply up to 2.8 gigawatts of solid oxide fuel cell (SOFC) power for AI data centers. Of the total deal, 1.2 GW is contractually confirmed for 2026-2027 U.S. deployment, while the remaining 1.6 GW remains an option.
AI training clusters require 10-100x the electricity of conventional servers. Connecting new U.S. facilities to the power grid takes 3-7 years on average due to permitting constraints. Bloom Energy's SOFC can be deployed in approximately 55 days -- faster than gas turbines (90+ days) or solar-plus-battery (100+ days). This removes the grid bottleneck slowing hyperscaler expansion.
Jefferies analyst Dushyant Ailani doubled his target price from $97 to $187, upgrading from Underperform to Hold. The 1.2 GW contract could add ~$3.8 billion in revenue, raising 2026E consensus by 20% and 2027E by 51%. However, the target of $187 remains below the current stock price of $198.71 -- the stock already appears fully priced.
For Korean investors: Doosan Fuel Cell signed an AI data center power supply MOU in October 2025 and expects 1-2 U.S. project wins in Q1 2026. KB Securities rates it Buy (target KRW 50,000; current approx KRW 34,900), but 2026 operating loss is projected at KRW 24 billion -- a theme play until hard contracts materialize.
My read: The Bloom Energy-Oracle deal is the first major proof that AI data centers will bypass the traditional grid for on-site power generation. The structural tailwind is real. But the thesis breaks if Oracle cuts capex or competing technologies -- like small modular reactors -- achieve comparable deployment speed in 2-3 years.
For the full analysis in Korean, visit Snakestock.
