Brig. Gen. (res.) Erez Askal, head of the National AI Staff in the Prime Minister's Office, stood up at AI Week 2026 at Tel Aviv University on Monday and said the quiet part out loud. "We are at the threshold of a different sovereign reality," he told the room. "Within a few years, national infrastructures will operate based on artificial intelligence; whoever controls it will control the switch of states."
Nobody in the hall was asking who controls the switch for the people on the other end of the wire. That's not the question the apparatus is built to ask.
The State's Monopoly, Now With GPUs
Askal used the opening lecture to launch the National Institute for Impact Oriented AI (NIIA), a body inside the Prime Minister's Office. Its shopping list: an energy infrastructure of 2 gigawatts, 100,000 GPU processors, a sovereign foundation model and an Israeli quantum computer. The plan is to take on ten challenges a year for five years, with a goal of getting five of the first ten to the implementation stage within a year. Each challenge gets a diagnosed barrier: regulation, data, technology or process.
The examples are mundane on purpose. Fourth-grade math. Traffic on Highway 6. The first meeting between a doctor and a patient. Nine national challenges in all, including health, aging, education, national resilience, equality and economic growth. The pitch is that AI will improve citizens' daily lives, and that solutions will be handed to bodies that can run them independently and for the long haul.
The man who runs this operation is a former chief of Unit 9900, Israeli military intelligence. Netanyahu appointed him on October 12, 2025, as the country's first AI chief, about one year ago and now in his second year in the post. He works alongside Israel National Cyber Directorate chief Yossi Karadi. The Jerusalem Post reported that the two have known each other for years from Israeli intelligence. A tight circle, drawn at the top.
Sovereignty Is Only Worth the Switch Itself
Askal's warning is worth reading closely. Israel and other countries, he said, will lose large aspects of their sovereignty to foreign companies within five to seven years if they don't control how AI is integrated into their infrastructure. Electricity, water, healthcare: all run by AI, and outsourcing it to multinationals based in the United States, China and elsewhere could cost Israel control of basic services.
Read that again. The complaint isn't that water and electricity and healthcare might be run by opaque systems nobody voted on. The complaint is that the wrong landlord might hold the keys. His answer is a domestic landlord. He also said Israel should partner with the dominant overseas AI producers while preserving control of AI in critical infrastructure. Dependence managed, not dissolved.
The June government decision gave Askal a budget of NIS 120 million, mostly for 2026, and authorization to fill 20 full-time positions. From 2027, NIS 13 million a year was due for salaries and operations, though Askal may spend it differently based on need. Netanyahu said the purpose is to make Israel "a world leader in the area of AI," adding that it will affect "the economy, security, science, industry, health, education – and on the State of Israel's international status." Askal said the decision "ensures the power and achievements of Israel for the upcoming decades."
Power, status, achievements. Every word aimed at the state's standing among states.
Who Gets Connected, Who Gets Counted
The institute is meant to link government ministries, industry, academia, civil society and philanthropy. That's the familiar architecture: civil society and philanthropy invited in as junior partners to an agenda written in the Prime Minister's Office. The AI bureau, the Post reported, will act as coordinator for the private sector toward clearer national goals. It will also pursue regulation to create order while avoiding overregulation that stifles innovation. Who gets to define "overregulation" isn't specified.
Askal's stated strengths are cybersecurity combined with AI and edge solutions, among them dealing with drones, emergency room surgery and the classroom. At a June 2 conference, about five months ago, he put it this way: "The ability to inject AI into the physical world and onto [solving] life's problems, to take an aircraft or drip irrigation and to make them 'smarter' is a critical growth engine for us." An aircraft and drip irrigation, in one breath.
He also pointed to AI security, using AI for cyber defense and defending AI systems. The base material offers no word on who audits any of it, who can refuse it, or what recourse a person has when a model sits between them and a clinic, a classroom or a tap. The institute promises "fair AI that can be trusted, with clear responsibility." Responsibility to whom isn't stated.
The Post noted Israel is considered dominant in cybersecurity but behind in broader AI, held back by limited space for data centers, thin natural resources for energy needs, and a cybersecurity success driven by the private sector rather than heavy government investment. So the state is now moving in to claim a field the private sector built, and calling it sovereignty.
The great powers, Askal said in June, are locked in a "crazy-paced arms race" over research, energy and modeling. His answer is to enter the same race at smaller scale, with the same logic: whoever holds the switch, holds the people downstream of it.
- israel
- artificial intelligence
- erez askal
- national ai staff
- technological sovereignty
- prime minister's office
- national infrastructure
- ai policy