Israel Tightens Rules for Server Farm Construction Amid Energy Concerns

Israel's Electricity Authority is set to implement new regulations for server farm developers seeking grid connection permits, including substantial financial guarantees. These measures aim to manage the surge in demand for AI and cloud computing infrastructure, which has led to a massive number of applications for electricity connections.

The Electricity Authority had previously halted the issuance of new permits in July due to an overwhelming demand of 27 gigawatts, far exceeding the estimated need of 2 gigawatts for AI server farms, as defined by the Prime Minister's Office's new AI task force. This halt caused uncertainty among developers, who feared losing future contracts and marketing opportunities.

Under the proposed new rules, developers will be required to pay a guarantee of approximately NIS 2 million per megawatt. For a 100-megawatt project, this would amount to a NIS 200 million guarantee. For exceptionally large projects, such as a 1.1-gigawatt facility, the one-time guarantee could reach NIS 2.2 billion. These guarantees are intended to ensure developers are serious about their projects and possess sufficient financial backing, thereby deterring those without deep pockets.

In addition to the financial guarantees, the Prime Minister's Office's National AI Directorate will now play a crucial role in approving server farm projects. Led by former military intelligence official Erez Eshel, the directorate will assess developers' expertise, financial capabilities, and client connections. This marks a significant shift, granting the AI task force regulatory authority over these infrastructure projects.

The new criteria are expected to be finalized and published for public comment in November. The AI task force, comprising former senior officials from defense and finance ministries, will evaluate projects based on demonstrated knowledge, funding capacity, and existing client relationships, aiming to balance the need for advanced computing infrastructure with the management of limited energy resources.