Israel's Military Reservist Techies Can't Catch A Break

Reserve duty rates that worked for brief conflicts become devastating during extended mobilization.

Welcome to Israel Tech Insider. I’m Amir Mizroch, a communications advisor and former EMEA Tech Editor at The Wall Street Journal. This newsletter is about connecting the dots to give you a sharp, insider’s take on what’s really going on in Israel’s tech industry.

As always, feedback to amir@israeltechinsider.com

---

And so began another round of reserve duty.
Suddenly, I’m no longer a computer guy, analytics, development, AI.
The one next to me is no longer a senior executive at a robotics startup. The one next to him is no longer a pediatrician at Maccabi (health organization). The one further along is no longer a freshly discharged soldier studying for the psychometric exam and wanting to be a mechanical engineer.
Now they are a group of people busy protecting the homeland.
In reality, completely voluntarily.
What a crazy place we live in.

THAT WAS a tweet from Tsoof Bar Or, a data science student who has been called up, again, for reserve military duty. This time in Lebanon. Why do I start the newsletter with Tsoof’s tweet? What does this have to do with Israeli tech?

Israeli law mandates military service for all Jewish men at 18 (Israeli Arabs are exempt, as are women with religious exemptions). Under a 1950’s agreement between Haredi parties and the state’s founder David Ben Gurion—who tried to reconstitute their Torah communities from the ashes of the Holocaust—ultra-Orthodox 18-year olds were exempt from military service in lieu of seminary study. And while in the early days of the state that deferment amounted to only several hundred haredim a year, that has now ballooned to tens of thousands of military-aged men getting a pass each year. Even in normal times the unequal burden of military service was becoming unsustainable. 21 months into a brutal seven-front war with a depleted and bone-weary army, it has become outrageous.

Many of the reservists now in their 3rd and even 4th tours of duty come from the tech industry. Tsoof is one of them. The casual despair in his tweet captures the brutal arithmetic eating away at Israeli tech: tens of thousands of professionals pulled from tech roles, mid-pitch deck creation, mid fundraising, mid product development, mid everything, again and again. Some are sent to Gaza. Some to Lebanon. Some to Syria. Some to the West Bank. Some to Iran and Yemen. Some to a combination of all these places in our forever wars.

According to the Israel Innovation Authority's November 2024 survey, 30% of tech companies identified prolonged military reserve duty affecting employees or their spouses as the primary employment-related challenge since October 7, 2023, with operational disruptions escalating proportionally to the percentage of workforce called up for duty. Project delays are becoming endemic. Companies fail to reach development targets. Some reservists return to find their positions eliminated. Shockingly, some are fired or quit after military service.

All this while 80,000 military-age ultra-Orthodox men collect government stipends without serving. There’s a double whammy here: not only are these haredi men not taking the place of worn-out reservists, they’re also highly, highly unlikely to join the tech workforce anytime either. That’s because most of them don’t believe in stem cells, let alone study STEM.

This matters. Israeli tech contributes over 20% of GDP and 54% of exports, making workforce disruptions a national economic security issue. Current trends show a narrowing base carrying an expanding burden while a growing exempt population extracts resources.

Haredim argue that Torah study provides spiritual defense. Perhaps. It’s hard to tell. In any case it rings hollow to combat engineers working on pitch decks between firefights while their wives hold down the home-fort.

This is why what happened last week is so dangerous. Last week, instead of taking the opportunity to contribute to the state in its darkest hour, both Haredi political parties quit the government rather than accept even modest sanctions on draft-dodgers, leaving Netanyahu with a minority government. His response? Fire Defense Committee chair Yuli Edelstein—the lone voice demanding equality of burden—and install a compliant replacement who'll codify permanent exemptions for an entire demographic. This effectively guarantees the exemption system will remain untouched through at least 2026, when coalition agreements expire. Let’s hope the war is over by then.

Member of Knesset Yuli Edelstein.

Either way, the military service arithmetic is unsustainable. Tech workers serve 10-20 times more reserve duty than peacetime norms—some combat veterans logging over 450 days of military service since October 2023. Meanwhile, 8,300 tech employees have emigrated since the war began, stripping institutional knowledge at rates that exceed replacement hiring. It's a brain drain concentrated among the exact demographic carrying the military and hi-tech forward. While exact numbers remain elusive—anecdotal evidence suggests the exodus may be slowing—the strain on reservist techies is unmistakable.

As an aside: the situation amongst non-techie reservists is much, much worse. For employees in the services and manufacturing sectors salaries aren’t nearly as high as in hi-tech. And consider the freelancers and small business owners: 33,000 businesses closed in the first half of 2025, compared to only about 19,000 that opened, according to a report published by the Self-Employed Forum of the Histadrut (Union of National Workers).

Critics might argue that Israel's tech sector has weathered previous conflicts. But three factors make this crisis different: the duration (already exceeding the 1973 Yom Kippur War), the demographic trajectory (Haredi population doubling every 16 years), and the global competition for tech talent and investment that didn't exist in previous conflicts.

Reserve duty rates that worked for brief conflicts become devastating during extended mobilization. Sustained military commitments without workforce diversification threaten long-term competitiveness. The sector's resilience has limits, and those limits are becoming visible in emigration statistics and operational stress surveys.

Netanyahu's coalition depends entirely on preserving a system where 14% of the population—growing rapidly through demographics—contributes nothing tangible to national defense while the economic engine funding Israel's existence slowly breaks down. A system where the productive subsidize the exempt while carrying disproportionate defense burdens cannot persist indefinitely. Netanyahu has chosen demographics over defense, ideology over innovation. The only question now is whether Israel will course-correct before its tech industry becomes another casualty of this war.

Go tell the Spartans our software update will be delayed.

---

Israel Has a National AI Program. On Paper.

Professor Itzik Ben Israel—a retired Air-Force major-general, former head of the Israel Space Agency and chairman of the National Council for R&D—told venture investor Guy Katsovich on the “Katsovich Podcast” that Israel’s long-heralded national artificial-intelligence plan has yet to move beyond the drawing board.

Commissioned by Benjamin Netanyahu in 2018, Ben Israel and Professor Eviatar Matania (the founding chief of Israel’s National Cyber Directorate, now at the Hebrew University) convened 300 specialists in 15 task forces to draft a strategy that would push the country into the world’s top five AI powers. The report, delivered in 2019, called for state spending of roughly NIS 2 billion ($540 m) a year for five years, later trimmed to NIS 1 billion during talks with the finance ministry. Three elections, a brief Bennett government and the current coalition’s focus on judicial reform and war funding meant the cabinet never voted on the proposal; no dedicated budget line exists.

The Israel Innovation Authority has redirected small sums from its regular grant pool, but big-ticket items—national computing infrastructure, health-data platforms, autonomous-transport regulation—remain unfunded.

The stakes are unusually high for a country of nine million that relies on technology for both prosperity and security. A decade ago Ben Israel’s cyber initiative knitted together defense know-how, academic research, and venture capital; Israel now captures roughly 40% of global private cyber-security investment and fields a quarter of the world’s unicorns in that niche.

Artificial intelligence promises similar leverage, yet demands heavier, more co-ordinated groundwork: sovereign GPU farms to curb dependence on foreign clouds, shared data trusts to train models in Hebrew and Arabic, and a regulatory lane that lets firms export defense-grade software without breaching export controls. Without such scaffolding, Israeli start-ups must rent scarce compute from abroad, researchers struggle to keep pace with American and Chinese labs, and the armed forces risk dependency on foreign infrastructure at the very moment AI is shaping the next phase of war.

Failing to invest in a national AI program is not simply a budgetary oversight for Israel—it is a risk to the pillars of its future as an agile, resilient state.

Prof. Itzik Ben Israel on Guy Katzovich’s Podcast.

---

Noteworthy

• Radiology artificial intelligence firm Aidoc raises $150M. The company, which has raised $370 million total since 2016, plans to leverage this investment alongside partnerships with Nvidia and AWS to expand its AI operating system aiOS and achieve coverage of 90% of clinically relevant diseases within three years. Aidoc holds over 20 FDA clearances spanning radiology, cardiology, and neurology for conditions ranging from aortic dissection to intracranial hemorrhage.

• Chip startup Hailo has unveiled the Hailo-10H, a palm-sized chip that lets laptops, cars, cameras and cash registers run ChatGPT-class language models and advanced vision algorithms locally, in real time, on just 2.5 watts. By moving heavy AI from distant cloud servers into the device itself, the chip slashes latency, subscription and bandwidth costs, locks sensitive data inside the hardware, and keeps critical functions alive even when networks fail. Built on a purpose-built “data-flow” architecture that wrings strong performance from miserly power budgets, the Hailo-10H gives smaller manufacturers the kind of AI punch once reserved for Big Tech’s server farms—threatening cloud revenue streams, accelerating in-car autonomy and physical-security analytics, and heralding an era where powerful generative AI is ubiquitous, private and pay-once.

• Cybersecurity startup Daylight emerged from stealth with $7 million in seed funding to automate the chronically broken managed detection and response (MDR) market—where organizations pay premium prices for mediocre, cookie-cutter cybersecurity services that still require extensive internal resources to actually resolve incidents. The startup deploys AI agents that can autonomously interview employees, integrate organizational knowledge, and close 50% more investigations than traditional MDR providers who merely offer recommendations, promising to finally deliver the 24/7 security operations that most companies need but cannot afford to build internally.

• Raja Giryes, a Professor of Electrical and Computer Engineering at Tel Aviv University, goes on Sabbatical at Apple AI Research.

• Investor Jack Levy is returning to climate technology by joining Uriel Klar as a partner at Epsilon, a venture studio focused on climate-tech. Epsilon are launching their fourth cohort, inviting 30 founders to participate in a four-month program starting in late September, aimed at helping them launch sustainable companies.

• Big language models usually think one word at a time, guzzling compute and cash, but an Intel-Weizmann team just found a way to make them sprint. Their study upgrades “speculative decoding”—the trick where a nimble helper model drafts several words and the heavyweight model simply rubber-stamps or deletes—so the two models no longer need to share the same private token “alphabet.” Three new algorithms let any small model pair with any large one, slashing generation time by up to 2.8× without hurting accuracy, and the code already ships in Hugging Face. Translation: faster chatbots, lower cloud bills, and fewer tech-vendor handcuffs, bringing real-time AI responses a step closer to your phone.