Israel Plans Its Own AI Model to Reduce Foreign Dependence, Protect National Infrastructure

National strategy calls for 100,000 GPUs, an Israeli quantum computer and AI-driven solutions to health care, education and transportation challenges

By Shula Rosen

Israel is pursuing technological independence in artificial intelligence through plans for a domestically controlled AI model, expanded computing capacity and a national quantum computer, as officials seek to ensure the country retains authority over systems increasingly essential to its infrastructure.

Brig. Gen. (res.) Erez Askal, who heads the National AI Directorate in the Prime Minister’s Office, presented the strategy Monday during AI Week at Tel Aviv University.

His proposals include developing 2 gigawatts of energy infrastructure and acquiring 100,000 graphics processing units (GPUs), alongside investments in specialized personnel and technological capabilities.

Askal described AI independence as a matter of national sovereignty, arguing that countries could become increasingly dependent on whoever controls the technology supporting their essential systems.

“We are on the threshold of a new reality of sovereignty. Within a few years, national infrastructure will operate on the basis of artificial intelligence. Whoever controls it will control the switch that powers nations.”

Separately, the government announced the establishment of the National Institute for Impact-Driven Artificial Intelligence, which will concentrate on translating technological advances into improvements in public services.

Operating within the National AI Directorate, the institute will bring together government agencies, academic institutions, private companies, nonprofit organizations and philanthropic groups.

Its mandate encompasses nine national challenges, including health care, education, population aging, economic growth, equality, environmental protection and climate issues, and national resilience.

Among the potential projects are improving mathematics instruction for fourth-grade students, addressing traffic congestion on Highway 6 and improving initial consultations between physicians and patients.

The institute expects to pursue 10 to 14 projects annually, with a goal of moving five of its first 10 initiatives into practical implementation within one year.

Askal also outlined a five-year program targeting 10 challenges annually, with each project examining obstacles involving regulation, data availability, technology or administrative procedures.

Successful initiatives are intended to move beyond experimental programs into sustained operation by organizations capable of managing them independently.

The strategy also calls for adaptable regulations, stronger cybersecurity safeguards and accountability mechanisms governing AI development and deployment.