When Israel's cabinet approved a comprehensive National AI Program on June 16, 2026, the headline numbers were striking: 100,000 sovereign processing units, a national quantum computer built on Israeli-developed technology, a new National AI Institute connecting government and industry, and a declared goal of establishing Israel as "a global AI superpower." Prime Minister Netanyahu described AI as "not just another technology — it is a revolution" that would shape the country's economy, security, science, and international standing. Brigadier General (ret.) Erez Askal — appointed head of the National AI Directorate within the Prime Minister's Office last October — declared the resolution "ensures Israel's power and prosperity in the coming decades."
This is a serious program grounded in serious prior analysis. Whether it delivers will depend on governance choices that received far less attention in the cabinet announcement.
From Blunt Warning to Ambitious Program
The June initiative did not emerge in a vacuum. The Nagel Committee — led by former National Security Council head Yaakov Nagel — issued a stark assessment that Israel was "not in the appropriate position to accelerate in the field of AI," citing inadequate computing infrastructure, a severe shortage of skilled researchers, and the absence of a governing strategy. The committee's 2025 report used the word "severe" eight times. Its recommendations included a NIS 25 billion ($6.6 billion) five-year investment, the purchase of at least 60,000 GPUs, and the creation of a dedicated AI headquarters inside the PM's office.
The June 2026 program exceeds those compute targets, aiming for 100,000 processing units — well above the committee's ask. That ambition is calibrated against a real global gap: China's 2024 quantum technology investment alone exceeded $15 billion, and Gulf states like Saudi Arabia and the UAE are deploying $40 billion or more in AI infrastructure. Against that scale, Israel's program is substantial for a country of its size, but modest relative to the powers it aspires to match.
Building on Private-Sector Momentum
The program's architects can point to a formidable private foundation. Israeli AI startup funding is projected to reach $7.9 billion in 2025, up from $4.9 billion in 2024. In the first half of 2025, 64% of Israeli companies raising rounds of $50 million or more were AI companies. The intersection of AI and cybersecurity alone is set to attract $2.5 billion — nearly double the prior year. Israel is not building an AI sector from scratch; it is trying to ensure its strong private base is matched by sovereign infrastructure that cannot be switched off by foreign suppliers or geopolitical disruption.
That rationale is coherent, especially for a country operating in a high-threat environment where compute capacity is a genuine national security asset. Israel also became the first country to sign a joint AI declaration with the United States under the Pax Silica initiative, with Askal and US Under Secretary of State for Economic Affairs Jacob Helberg signing in Jerusalem on January 16, 2026. The declaration covers joint R&D, training programs, and semiconductor cooperation — extending Israel's access to the American technology ecosystem at a strategically important moment.
The program also includes domestic workforce measures: AI degree programs launching at Israeli universities in October 2026, national retraining mechanisms for displaced workers, and acceleration hubs designed to convert national challenges into applied AI solutions.
The Governance Tension
Critics of the program's structure raise concerns that deserve a fair hearing. Placing the AI Directorate inside the Prime Minister's Office — rather than the Ministry of Innovation, Science and Technology — concentrates decision-making at the apex of government. Askal was appointed without a competitive tender; the Civil Service Commission justified the exemption on urgency grounds. The directorate received NIS 120 million in initial budget, with just NIS 13 million annually from 2027 for operations — a figure that looks structurally underweight for the stated ambitions.
There is also genuine jurisdictional complexity. The Israel National Cyber Directorate (INCD) has operated as the country's lead civilian cyber body since 2014 and still lacks permanent legislation after a decade. As AI capabilities become central to both offensive and defensive cyber operations, the boundaries between the AI Directorate and the INCD will require careful management. Former cyber officials are divided: some see natural complementarity, others anticipate a repeat of the Shin Bet-INCD turf disputes that took years to resolve.
"The AI directorate's broad mandate mirrors the INCD's original 2014 charter in scope — and the INCD still doesn't have permanent legislation enshrining its authorities." — Jerusalem Post analysis, 2026
Regulation: The Right Instinct, an Incomplete Architecture
On the regulatory side, Israel is pursuing a risk-based sectoral model rather than a sweeping EU AI Act-style framework — the correct instinct for a small, innovation-dependent economy. The Privacy Protection Authority published draft AI guidelines in February 2025 applying existing privacy law to algorithmic systems, and Amendment 13 to the Privacy Protection Law took effect in August 2025, strengthening enforcement powers considerably. A potential "Framework Law" for AI — covering algorithmic discrimination, liability for autonomous systems, and AI-generated content — remains under discussion but has not advanced to legislation.
The absence of a comprehensive AI statute is not inherently a problem; it allows regulation to develop iteratively as risks materialize rather than locking in rules calibrated to 2025's AI landscape. But PPA draft guidelines are not a substitute for clear statutory rules on high-stakes applications — employment screening, public benefits administration, biometric surveillance — where the cost of getting it wrong falls on individuals with limited recourse. The PPA indicated it will step up active enforcement of its expanded powers in 2026; that is a meaningful signal, but guidance enforcement is not the same as legislation.
What Success Requires
An AI program's worth is tested in implementation. Israel has a credible compute target, an internationally anchored US partnership, a booming AI private sector, and a directorate with direct access to the prime minister. The risks are the mirror image of those assets: centralization without codified accountability, aspirational GPU numbers that may strain energy and procurement timelines, and a governance vacuum between the AI Directorate and existing cyber institutions.
The June 16 program is a necessary step for a country that cannot afford to fall behind in the infrastructure layer of the next technological era. Calling it sufficient would be premature.