Lazy Digest #9: Anthropic Walked Away From $6 Billion. Nir Zuk Got $245 Million Anyway.

Tue, 15 Sep 2026 13:10:00 +0100

Lazy Digest #9: Anthropic Walked Away From $6 Billion. Nir Zuk Got $245 Million Anyway.

Three digests ago, Nvidia offered more money for Decart and lost anyway, because the founders wanted Anthropic instead. This week we found out why that preference was risky: Anthropic finished its due diligence, then said no.

On the same Tuesday, a company with no shipping product, no customers you can call and no disclosed revenue raised $245 million on the strength of one man’s name. The market’s mood this week, in one sentence: extremely picky about billion-dollar acquisitions, completely relaxed about writing nine figures to a founder who’s already built a great cybersecurity company once.

Welcome to Lazy Digest #9 — 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.

The Ninetieth Minute

We’ve carried Decart’s acquisition talks since issue #6, where Nvidia’s higher, cash-richer bid lost to Anthropic’s lower, stock-based one — a preference from the founders and lead shareholder Sequoia that we flagged, at the time, as a bet on Anthropic’s own IPO. Issue #7 set a rule: Decart doesn’t get its own line again until there’s a signed agreement.

There wasn’t one. On September 8, Bloomberg reported that Anthropic had finished due diligence on Decart and walked away. No signature — a collapse, which is the other kind of ending, and arguably more informative than a close. So it gets the line the rule was written for.

The number itself has never settled. Bloomberg and most outlets say roughly $6 billion. CTech, in its own reporting on the collapse, says "$6 billion to $7 billion," which happens to match the ~$7 billion figure this digest cited from CTech’s contemporaneous reporting back in issue #6. Two months, same negotiation, and the press still hasn’t agreed within a billion dollars — read that as a sign that nobody outside the room actually knew the number, including, quite possibly, some people inside it.

More interesting than the price is why it died. CTech offers two explanations beyond the obvious “IPO timing got complicated”: that Anthropic, having spent weeks inside Decart’s chip-efficiency stack during diligence, may have concluded it could build the equivalent in-house — diligence as an expensive technical education — and that “geopolitical considerations played a role, including questions surrounding Qatar’s significant investment in Anthropic and the possibility of a Qatari connection to the acquisition of an Israeli company.” Haaretz adds a third reason, sourced to people close to Decart: the deal broke over Anthropic’s plan to move most of the company’s operations and staff abroad.

Founded in September 2023 by brothers Dean and Orian Leitersdorf with Moshe Shalev, Decart last priced at nearly $4 billion in a $300 million round led by Radical Ventures in May, up from $3.1 billion the August before. Anthropic’s stock-based offer implied roughly a 50% premium on that mark. Nvidia — still an investor in Decart, and the party that lost this bidding war back in issue #6 — is the obvious next call: the only outside party left standing with both the capital and the motive to try again. Neither company has said anything about that possibility yet.

For any Israeli AI-infrastructure founder treating a frontier lab as the exit: the lesson isn’t only that a stock offer from a company that isn’t public yet can evaporate after full diligence. It’s that geopolitical optics around the buyer’s own cap table, and where they want your team to physically sit once you’re theirs, are now live deal-breakers — categories of risk nobody was pricing into a term sheet in 2023.

Two Hundred Ninety Million Dollars, No Product Yet

Palo Alto Networks founder Nir Zuk raised $245 million for his new company, Cylake, on September 8 — through a convertible note, not a priced round, which matters: Cylake hasn’t set, or exposed to scrutiny, an actual valuation. Zuk co-founded the company with Wilson Xu and SentinelOne co-founder Ehud Shamir, though the round was priced, and covered, entirely on Zuk’s name. Total raised is now $290 million, six months after a $45 million seed from Greylock, the same fund that backed Palo Alto Networks at the start. Lightspeed, Picture Capital and Redpoint put in the new money; Lightspeed co-founder Ravi Mhatre joins the board.

Cylake is building what it calls “operational sovereignty” security — an AI-native platform running entirely on-premises or in a private cloud, for institutions that structurally cannot put their security stack on someone else’s servers. CMO René Bonvanie says the capital funds the hardware, storage and AI infrastructure needed to run the whole stack on a customer’s own premises — an unusually vertically integrated bet for a security company. Beta is promised by year-end, general availability in 2027. No design partners named. No revenue disclosed.

None of that is a criticism, exactly. It’s a description of what $290 million actually buys before a product exists: not a business yet, but a bet on Zuk personally, who spent two decades building and running Palo Alto Networks before stepping down as CTO last August. Cylake is HQ’d in Sunnyvale, not Tel Aviv — which puts it next to Covenant, below, on the growing list of the biggest Israeli-founder raises going to US-incorporated companies rather than Israeli ones. We flagged that pattern hard in issue #8; this week it’s two for two on the two largest stories.

A Missile Factory Opens in Grand Prairie, Texas

Covenant spent two years in total secrecy and came out of it on September 9 with a working product, a Texas factory, and $250 million-plus raised across three rounds nobody had heard about — Andreessen Horowitz, Founders Fund, Lux, 8VC, Aleph, Lightspeed and Altimeter, at a reported valuation north of $1 billion.

The product is Anthem: a ground-launched, long-range cruise missile carrying more than 440 pounds, aimed at a mid-six-figure unit cost. For comparison, the U.S. Navy lists the Tomahawk Block V at roughly $3.64 million. That gap isn’t a rounding difference — it’s the entire pitch. Covenant opened a 105,000-square-foot plant in Grand Prairie, outside Dallas, targeting 1,000 units in year one and 5,000 annually after that, plus a second line in Leipzig and a 32,000-square-foot R&D-and-manufacturing site in northern Israel. The company has booked roughly $150 million in orders already, plus a separate ~$70 million U.S. Navy contract to adapt Anthem into a naval variant.

Founded in 2024 by Israeli-American Michael Kaufman — ex-Thiel Capital, and, per the company’s own account, recently relocated to Israel himself to help build that side of the business — with CTO Ofir Dayan. Same shape as Cylake: US headquarters, US manufacturing, Israeli engineering. Two of the week’s biggest stories, both built by Israeli founders, neither one an Israeli company on paper.

The other defense story this week was XTEND, which rang the NYSE opening bell the same day Cylake announced — but the number that actually prices the stock, the Pentagon’s Gauntlet II drone-competition leaderboard, still hasn’t been published. Worth knowing before you buy the ticker: XTEND didn’t finish in the top 11 at the previous round, Gauntlet I. This one isn’t a formality.

Everything Else Worth Your Time

Five smaller funding rounds and two clean exits filled out the rest of the week, all landing before Thursday — Rosh Hashanah began at sundown on the 11th and the outlets went quiet through the weekend.

Wint raised a $36M Series D co-led by LIP Ventures and Inven Capital for its AI leak-detection platform: 30,000-plus systems sold, nearly 600 enterprise customers, and a December strategic investment from property group Grosvenor already on the books. Cymphony ($25M Series A, founder Shy Dekel) and Euno ($23M Series A, backed by the founders of Wiz, Cyera and Eon, and separately named to Gartner’s “Coolest Vendor Innovations in Data Management” list for 2026) are both, in their own words, securing the seam between human employees and AI agents — a seam so many companies are now funded to secure that the category looks less like a frontier and more like a subdivision. HydroSight ($5M seed) is the actual outlier: seabed intelligence from marine survey data, for ports and energy operators planning underwater infrastructure. Small, specific, nothing to do with agents.

On the exit side: Prytek acquired Scanovate, a bootstrapped digital-identity company that had never taken outside money in eleven years and serves 250-plus customers including some 120 financial institutions, for ~$50 million in its first phase — a clean, unglamorous win for capital efficiency. And Kiteworks bought Bonfy.AI, an AI data-security startup founded by Gidi Cohen (who sold Skybox Security for $375 million in 2018) and Danny Kibel (former CEO of Idaptive, later CyberArk’s global R&D head after that acquisition) — terms undisclosed, CTech estimates tens of millions. Two founders with real prior exits, doing it again in eighteen months instead of a decade.

Add it up and the week’s disclosed venture total lands around $334 million across five rounds — a 76% drop from issue #8’s $1.4 billion. The honest read isn’t that the ecosystem cooled. It’s that #8’s number was never a trend — it was two days of backlog clearing at once, right after the school year started — and week two just looks like August again. Rosh Hashanah didn’t help.

What I’d Do With This Week

If you’re pitching a frontier lab as your exit: price in more than valuation risk. The Decart collapse suggests acquirers are now weighing where your team physically sits and who’s on their own cap table, not just what your technology does. Surface those questions early, not in week nine of diligence.

If you’re raising off a founder’s name rather than a shipping product: Cylake is the ceiling right now, not the floor. $290 million pre-revenue works if you’re Nir Zuk. It’s a much harder pitch for anyone whose last company didn’t define a category.

If you’re building anything with a physical, exportable product: Covenant’s unit-cost math — mid-six figures against a $3.64 million incumbent — is the actual argument, not the funding number. Investors are paying for the gap between those two prices, not for the missile itself.

If you’re job-hunting in Israeli tech right now: defense is the visible growth story. Covenant and XTEND are both hiring in Israel regardless of what the AI-funding cycle is doing elsewhere. It’s a narrower door than software used to be, but it’s open.

If you’re an ecosystem watcher: stop treating any single week’s total as signal. This digest alone has now shown $315M, a near-total lull, $1.4B and $334M inside two months. The pattern is spikes and troughs, not a trend line — the number worth tracking is the rolling monthly average, which nobody quotes because it’s less exciting than either headline.

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: Bloomberg, CTech / Calcalist, Globes, Haaretz, SecurityWeek, VC Cafe’s Firgun newsletter, and company/press releases including Cylake, Covenant and Kiteworks. Estimates, not disclosed figures: Decart’s deal value ("$6bn" per Bloomberg and most outlets, “$6bn–$7bn” per CTech — genuinely unresolved, not a typo), Covenant’s >$1B valuation (“reportedly”), Bonfy.AI’s price (CTech estimate, “tens of millions”), and Scanovate’s up-to-$100M shareholder-value figure (Prytek’s own projection). Company-provided and unaudited: Wint’s water-savings and damage-prevention figures, Cylake’s beta/GA timeline, and Covenant’s $150M order book and 200-plus test flights. Cylake’s $245M is structured as a convertible note; no post-money valuation has been disclosed. Anduril’s reported hire of Amikam (also rendered Amiram) Norkin to lead its Israel operations, IAI’s IPO/bond-confidentiality standoff, and UVision’s Nasdaq process are all still unresolved and unchanged since issue #8 — not repeated here, watched for next issue. Per the policy set in issue #7, Decart’s story was to return only on a definitive resolution; a walk-away counted as one. Rosh Hashanah began at sundown on September 11 (continuing through September 13, outside this window) and closed the Israeli news cycle for the back half of the 6–12 September window. Previous issues: #1, #2, #3, #4, #5, #6, #7, #8.

Scanovate on Sherman - Israeli Startups

Lazy Digest #9: Anthropic Walked Away From $6 Billion. Nir Zuk Got $245 Million Anyway.

Lazy Digest #9: Anthropic Walked Away From $6 Billion. Nir Zuk Got $245 Million Anyway.

The Ninetieth Minute

Two Hundred Ninety Million Dollars, No Product Yet

A Missile Factory Opens in Grand Prairie, Texas

Everything Else Worth Your Time

What I’d Do With This Week