Lazy Digest #10: He Sold to Apple for $20 Million. This Week He Sold to OpenAI for 15 Times That.

Tue, 22 Sep 2026 09:00:00 +0100

Lazy Digest #10: He Sold to Apple for $20 Million. This Week He Sold to OpenAI for 15 Times That.

In 2015, Apple bought a small Israeli camera startup called LinX Imaging for about $20 million. Its founder went on to lead the team that built the iPhone’s Portrait Mode, then left to start a second company built on the same conviction: smartphone cameras had hit a wall that better lenses couldn’t fix, only better AI could. This week, OpenAI bought that second company for more than $300 million. Same founder, same core idea, fifteen years and roughly fifteen times the price apart.

A few hours later, on the other side of the ecosystem, a Nasdaq-listed American defense company paid up to $390 million for an Israeli firm that makes a part most people will never think about: the electronic switch that decides whether a missile is allowed to detonate. Nobody wrote a headline calling it romantic. It is, by some distance, the more important story of the two — and also the one that got a tenth of the coverage.

Squeezed in around both, before Rosh Hashanah’s tail cleared and Yom Kippur’s eve arrived, six more Israeli-founded companies raised a combined $297 million, and a two-month-old cliffhanger about a drone competition finally got a number attached to it.

Welcome to Lazy Digest #10 — Israeli tech, one week, researched with AI and filtered by a human who reads the primary sources. Estimates are labeled as estimates. Let’s go.

Fifteen Times, Eleven Years Apart

Ziv Attar grew up in Afula, studied optical engineering at the Technion, and worked as an optics specialist at Rafael before he ever started a company. His first one, LinX Imaging, fused a monochrome sensor with a color one to fake the depth and low-light performance normally reserved for cameras too big to fit in a phone. Apple bought it in 2015 for a reported ~$20 million and put Attar to work on exactly that problem: he co-led the Apple team that shipped Portrait Mode on the iPhone 7 Plus, alongside a colleague named Tom Bishop.

The two left Apple in 2019 with the same diagnosis, one generation later: smartphone cameras were hitting a “glass ceiling,” and the industry’s answer — bigger sensors, more lenses — was running out of room inside a phone’s thickness. Their new company, Glass Imaging, bet the fix wasn’t hardware but a neural network trained on one exact camera-and-lens combination’s quirks, correcting the image as it’s captured rather than edited after. They called it GlassAI. It shipped before most people had heard of the company: Chinese phone maker Honor uses it for zoom on its 600 series, and Motorola has tested it.

The Wall Street Journal reported on September 14 that OpenAI has acquired Glass Imaging for more than $300 million — roughly triple the ~$100 million valuation it carried out of a $20 million Series A in May 2025, and about 10x the ~$30 million it had raised in total across a 2021 seed (LDV Capital), a 2024 extension (GV) and that Series A (Insight Partners, GV, Future Ventures, Abstract Ventures). Neither company had confirmed the deal on the record as of this writing, which is now a familiar OpenAI pattern — the WSJ story broke it, and everyone else, including CTech, worked from there. Israel’s Foreign Ministry didn’t wait for confirmation before publicly celebrating it.

What OpenAI wants Glass Imaging for is genuinely unclear, and that’s the more interesting part. The company is building consumer hardware with Jony Ive’s old team following the $6.5 billion io Products acquisition, and a camera system that produces clean, artifact-free imagery straight out of capture is a plausible input layer for any device meant to see and understand a room without the user typing a description of it. That’s a hypothesis, not a confirmed roadmap. What’s confirmed is simpler and, in its way, more interesting: an engineer who spent a decade solving the same problem for two different buyers just closed a second exit worth fifteen times his first, and the buyer this time isn’t a phone company at all.

One clean footnote for continuity: issue #9 covered Anthropic walking away from Decart over a deal reportedly worth $6–7 billion, partly — per CTech’s sourcing — on discomfort with an Israeli company’s cap table and geopolitical optics. This week’s frontier-lab acquisition of an Israeli-founded company carried none of that visible friction. The difference may just be scale: a $300 million tuck-in draws far less scrutiny than a $6 billion strategic bet timed against an IPO. Worth watching whether that holds as a pattern or was simply this week’s exception.

The Switch That Arms a Missile

Ondas Inc. (Nasdaq: ONDS) announced on September 14 that it is acquiring GATE Technologies, a Shoham-based company founded in 2006 that makes Electronic Safe & Arm Devices — the components that determine when a missile, rocket, loitering munition or UAV payload is allowed to arm and detonate — along with its Poland-based manufacturing affiliate, Bron Technologies. Per the Jerusalem Post, GATE’s fuzing technology is integrated into dozens of weapon systems worldwide and has been used in the Iron Dome program.

The structure is worth reading closely, because the headline number undersells the actual bet. Ondas is paying $205 million up front — $105 million cash, $100 million in its own stock, with about $22.5 million of that stock deferred up to nine months — plus up to $185 million more in earn-outs tied to financial targets running through 2028. That puts the maximum consideration at $390 million. Ondas expects GATE to generate $65 million in revenue this year, rising to $180 million by 2028, and to contribute more than $130 million of aggregate EBITDA over that period; roughly 80% of GATE’s revenue currently comes from outside the Middle East. All of that is Ondas’ own forward-looking projection, not an audited figure, but the shape of the deal — cash now, stock now, and the largest single piece contingent on three years of delivery — tells you the buyer is underwriting qualified production capacity and an existing customer base, not betting on a patent portfolio.

This isn’t Ondas’ first stop in Israel. In August, the company appointed former Mossad director David Barnea as global president and chairman of Ondas Defense — one of several former intelligence chiefs, alongside Tamir Pardo and Yossi Cohen, to move into the private sector recently. GATE is now the latest in a string of Israeli defense-technology acquisitions Ondas has made this year, alongside earlier deals for DZYNE and Cyberhawk, which together consumed roughly $325 million of Ondas’ cash in the third quarter alone.

Set this next to issue #8’s coverage of Anduril recruiting former IAF commander Amikam Norkin to build its Israel operations, and a shape starts to emerge: American defense-tech companies aren’t just buying Israeli IP anymore, they’re buying Israeli production capacity and hiring Israeli security leadership to run it. That’s a materially different relationship than a straightforward acquisition, and it’s one Israeli defense-tech founders should be pricing into how they think about their own eventual buyers.

The Blank XTEND Left Us

Two issues ago we flagged that XTEND had rung the NYSE opening bell on a valuation resting partly on an unpublished leaderboard. Issue #9 left the number blank. This week it got filled in.

The Pentagon’s Drone Dominance Program posted the Gauntlet II results on Tuesday, September 15. In the Close Quarters Battle segment: Neros first with 88.1 points, ORQA US second with 82.8, and XTEND Reality (NYSE: XTND) third with 80.2, ahead of Vector (78.2) and ModalAI (74.3). On the separate Deep Strike board — where XTEND didn’t compete — Perennial Autonomy topped Hyperscale, Neros and Skycutter, and neither Ondas nor Red Cat made either leaderboard at all, a detail Benzinga flagged under the headline “Red Cat, Ondas Miss the Cut.”

XTEND’s placement followed delivery and demonstration of 120 STRIKER systems, running its proprietary XOS operating system, in operationally realistic scenarios at Fort Carson in August. CEO Aviv Shapira called it “a major win… We delivered, we competed, and we proved what STRIKER is capable of,” and the company says it now has more than 12,500 systems deployed globally.

Third place is genuinely good — XTEND is the only publicly traded company on either leaderboard, so it’s the only one whose Gauntlet performance shows up in a stock price rather than just a press release. Two caveats belong alongside the celebration, though. The Department of War plans to buy roughly 60,000 drones across the ten finalists — cited elsewhere at around $300 million — from a program described as an “iterative $1 billion” initiative that fixes quantities at contract award, not strictly by rank. And the program’s own status reporting has noted at least two Gauntlet I winners weren’t yet shipping their earlier awarded orders as of July — placing well and delivering at scale are different achievements.

Six Rounds, $297 Million, Not Much in Common

Everything else that priced this week landed Monday through Thursday, before the holiday calendar shut the news cycle down again.

Buildots — $130M, led by billionaire Eyal Ofer’s O.G. Venture Partners with Lightspeed, Intel Capital, Mohari Ventures (also a backer of SpaceX, Neuralink and Anduril), Qumra, Viola Growth and Avigdor Willenz. Total raised: $297 million since 2018. CEO Roy Danon told Bloomberg the round puts the company “within striking distance of” a $1 billion valuation — not a confirmed one. Buildots’ helmet-camera computer vision compares what’s actually built on a site against the schedule, and its ~400 employees now serve 100-plus large builders including Intel and Digital Realty, increasingly on AI data-center projects. Danon and co-founders Yakir Sudry and Aviv Leibovici are all Talpiot graduates — the same pedigree line that keeps showing up across this ecosystem regardless of sector.

MIND — $72M Series B, led by Crosspoint Capital Partners with existing backers YL Ventures and Paladin Capital Group, bringing total funding to $112 million barely a year after its $30 million Series A. Market estimates put the new valuation near $300 million — roughly a tripling. Seattle-headquartered, Israel-developed, founded 2023 by CEO Eran Barak (who sold his first company, Hexadite, to Microsoft for ~$100 million in 2017) with CTO Itai Schwartz and VP R&D Hod Bin Noon, a Unit 8200 veteran. MIND rebuilds data-loss prevention for a world where AI agents move data faster than static rules can track it, and claims 17x revenue growth over the past year — company-reported, unaudited, but consistent with how fast this specific sub-category has been raising all year.

Noetive — $41M seed, out of stealth after five months, led by Eclipse with Craft Ventures, The Westly Group, Swish Ventures and others. CEO Amir Frenkel did his undergraduate studies at the Technion before a decade at Meta, most recently as VP of Generative AI, and joined Eclipse in April as its first Chief AI Officer specifically to build this company — making Eclipse less an investor here than a co-founder with a fund attached. Noetive pairs a self-improving AI model with its own sensing hardware to build an “intelligence of record” for factories, logistics and energy operators — physical-world companies this AI wave has mostly skipped.

Loora — $22M Series B, led by Union Tech Ventures, taking total funding to $43.25 million. Founders Roy Mor and Yonatan Levin met in Sayeret Matkal, the IDF’s most elite special-operations unit, before building an AI English tutor now claiming 15 million users and ARR “in the tens of millions” — self-reported, unaudited, but backed by a stated plan to double headcount from 30 to 60 within six months.

FIZE Medical — $20M Series B, first close, co-led by Japan’s Asahi Kasei Medical and Rapha Capital Management. Dual-based in Modi’in, Israel and Jericho, New York, founded by Noam Levine and now run by CEO Dror Zerem, previously CEO of VisionMap (sold to Rafael). FIZE’s real-time urine-output monitor for ICU fluid management is already deployed in Japan, the US and Europe; the round deepens Asahi Kasei’s role from distributor to investor and adds European distribution on top of its exclusive Japan deal.

Keewano — $12M seed, out of stealth, led by Hetz Ventures with a16z speedrun, Remagine Ventures and DIG Ventures. Founded 2024 by Mark Kardashov with three co-founders who share two prior exits (TestProject to Tricentis in 2019, Devalore to Abra in 2022) and gaming-analytics scale from Plarium. KeewanoDB is a database built for AI agents doing open-ended reasoning rather than humans running predefined dashboard queries, and Hetz’s Judah Taub told Dealroom the round “ranks in the 95th percentile among all-time seed deals for AI startups in Israel.”

Add it up and you get $297 million across six rounds — modest next to issue #8’s $1.4 billion week, roughly in line with issue #9’s $334 million. The more persistent signal isn’t the total, it’s the geography: Glass Imaging (Los Altos), MIND (Seattle) and Noetive (San Francisco) were three of this week’s four largest funding-and-exit stories, and none of the three is an Israeli-incorporated company. Issue #9 flagged this as “two of four.” This week it’s three. That’s not enough data points to call a trend line, but it is enough to say the pattern hasn’t reversed.

Everything Else Worth Your Time

No Israeli tech layoff, shutdown or restructuring was announced this week. Given the holiday calendar compressed the news cycle to four working days, treat this as an artifact of the calendar rather than a signal that the broader 2026 layoff wave has ended.

Dvash Group and Hebrew University launched a $50 million-target deep-tech fund on September 16, run independently by Yoram Dvash with the university and its tech-transfer arm Yissum, aimed at quantum, dual-use, defense and cybersecurity spinouts. The $50 million is a target, not capital already closed.

ScaleUp House announced its first cohort — roughly 20 companies selected from about 30 that pitched, including Guidde, Brandlight, BloomX and Tevel Aerobotics Technologies. The program, run by Hilla Ovil-Brenner with Calcalist, LeumiTech, Google for Startups and Sela, targets a gap this digest keeps circling back to: real product, real traction, still missing the sales-and-brand “muscle” to go global.

OpenAI added Israel to ChatGPT’s self-service Ads Manager on September 16, alongside the UAE, Saudi Arabia and Turkey — advertiser account access only. OpenAI hasn’t said when, or whether, ads will actually start appearing to Israeli users.

On the continuity ledger: Anduril’s appointment of Amikam Norkin to lead its Israel operations remains unofficial and unchanged since issue #8. IAI’s IPO remains “closer than ever,” per chairman Boaz Levy, with no new movement and the October 27 election still the practical gating factor. UVision’s Nasdaq listing had no new milestone this week, following last month’s CFO transition (Eyal Rubin replacing Ronny Barak) amid continued pressure on its target valuation. Decart gets no line this week, per the standing house rule — nothing new and concrete happened.

What I’d Do With This Week

If you’re building anything with a genuine, narrow technical edge: Attar’s second exit is the argument for depth over breadth. Glass Imaging won by solving one specific optical problem better than anyone else, twice, for two different buyers eleven years apart. If your pitch needs three paragraphs to explain what makes you different, sit with that.

If you sell physical components into defense supply chains: study the Ondas/GATE structure, not the funding number. Buyers are underwriting production capacity and delivery record with earn-outs stretching three years out, not paying up front for a patent. Proving you can ship at scale is now worth more than IP alone.

If you’re competing in a Pentagon program: XTEND’s third place is real, but Gauntlet I winners still not shipping as of July is the more important data point. Placing on a leaderboard and fulfilling an order are two separate businesses — model your revenue on the second one.

If you’re raising a seed in agent infrastructure: Keewano’s round and MIND’s third milestone in two years both show the category still commands a premium — but both wins came from teams with prior exits already behind them. Budget time to build that résumé, not just the pitch.

If you’re tracking where Israeli-founder value actually lands: three of this week’s four biggest stories were companies incorporated outside Israel. Not new at this point in the year, but it hasn’t reversed either.

See you next week. Same lazy hands, same working head.

With 💜, Alex Sherman

Research assisted by AI, every claim read and filtered by a human. Primary sources: CTech / Calcalist (English and Hebrew), Globes, Jerusalem Post, Times of Israel, Ynetnews, Bloomberg, WSJ (via TechCrunch/SiliconANGLE secondary reporting), SecurityWeek, DefenseScoop, Benzinga, PR Newswire, GlobeNewswire, Ondas’ SEC 8-K, Dealroom, VentureBurn and VC Cafe’s Firgun newsletter (Sept 18 edition). Estimates, not disclosed figures: Glass Imaging’s >$300M price (WSJ, unconfirmed by either company); Ondas’ up-to-$390M GATE consideration ($205M base is SEC-disclosed, the $185M earn-out is contingent on unmet 2028 targets); Buildots’ “within striking distance of $1B” (CEO commentary, not a disclosed valuation); MIND’s ~$300M valuation (market-source estimate). Company-provided and unaudited: MIND’s 17x revenue growth, Loora’s user/ARR figures, Keewano’s token-savings claims, FIZE Medical’s patient-monitoring volumes. FIZE’s founding year is reported inconsistently (2011 vs. 2020) and flagged rather than silently resolved. XTEND’s Gauntlet II placement is a ranking, not a confirmed production order. Per the policy set in issue #7, Decart returns only on a new, definitive development; none occurred this week. Rosh Hashanah’s tail and the approach of Yom Kippur (sundown Sept 20) compressed reporting into Monday–Thursday. Previous issues: #1, #2, #3, #4, #5, #6, #7, #8, #9.

Dvash Group on Sherman - Israeli Startups

Lazy Digest #10: He Sold to Apple for $20 Million. This Week He Sold to OpenAI for 15 Times That.

Lazy Digest #10: He Sold to Apple for $20 Million. This Week He Sold to OpenAI for 15 Times That.

Fifteen Times, Eleven Years Apart

The Switch That Arms a Missile

The Blank XTEND Left Us

Six Rounds, $297 Million, Not Much in Common

Everything Else Worth Your Time

What I’d Do With This Week