FROG · JFrog Ltd. · Stock research

Current Thesis

Prior coverage set the momentum-leg break at a weekly close below $82 and that condition was met twice — $80.16 on 2026-07-24 and $79.79 on 2026-07-31 — into a drawdown that bottomed at a $73.28 low close on 2026-07-29. The 2026-08-06 Q2 print reversed it. Revenue came in at $163.772M, +29% YoY, above the guided $154M–$156M range and above the $155.637M consensus; non-GAAP EPS $0.27 versus $0.24 expected; and the FY26 revenue range moved to $648M–$652M from $628M–$632M, with non-GAAP EPS to $0.96–$1.00 from $0.93–$0.97. The full-year raise was larger than the quarter's outperformance against its own guided range, which puts incremental dollars into the back half rather than just banking a beat.

The narrative leg on offer is the same one, now with a growth number that accelerated instead of decelerating: Artifactory as the provenance and governance layer for artifacts produced by AI coding agents, monetized through cloud migration and security attach inside an installed base. Cloud revenue was $87.5M, +53% YoY, and crossed to 53% of total from 45% a year earlier. RPO reached $659.0M, +38% YoY. Trailing four-quarter net dollar retention was 121%. Customers above $1M ARR numbered 97, +59% YoY.

Eleven firms raised targets on 2026-08-07 — UBS, TD Cowen and Benchmark to $120; Guggenheim, JPMorgan, Oppenheimer, Needham and BTIG to $115; KeyBanc $114; Truist $110; Piper Sandler to $90 while staying Neutral — and DA Davidson followed at $115 on 2026-08-08. Price closed $96.17 on 2026-08-14. The next dated event is swampUP, 2026-09-01 to 09-03 at The Glasshouse in New York, where the 2026-08-12 announcement says new product introductions land in the executive keynotes.

Bullish and bearish views on JFrog Ltd.

The model's bull view on JFrog Ltd. (FROG), in brief: Growth accelerated through the guided deceleration (2026-08-06): $163.772M, +29% YoY, against +26% in Q1 2026 and an FY26 guide that had implied roughly 18.5% at the prior midpoint. The bear view: The stock has not made a new high on strictly better information. Both cases follow in full.

Bull Case

• Growth accelerated through the guided deceleration (2026-08-06): $163.772M, +29% YoY, against +26% in Q1 2026 and an FY26 guide that had implied roughly 18.5% at the prior midpoint.
• Cloud crossed the majority line and is compounding faster than the whole (2026-08-06): $87.5M, +53% YoY, 53% of revenue versus 45% a year ago; management lifted the full-year cloud growth frame to 41%–43%.
• Contracted backlog outruns reported revenue (2026-08-06): RPO $659.0M, +38% YoY, against 29% revenue growth — the forward book grew faster than the print.
• Land-and-expand is measured, not asserted (2026-08-06): NDR 121% on a trailing four-quarter basis; 97 customers above $1M ARR, +59% YoY; Enterprise Plus at 59% of revenue versus 55% a year earlier.
• Cash conversion (2026-08-06): record free cash flow of $53.8M at a 33% margin; gross margin 83.2% versus 83.1% a year prior.
• The agent-distribution surface is real and named (2026-08-06 release): a JFrog plugin built with Anthropic, and a Cursor coding-agent plugin the company describes as reaching over 1M AI developers.
• swampUP has the counterparties on stage (2026-08-12 release): Anthropic Deputy CISO Jason Clinton, Cursor Field CISO David Pan, plus NVIDIA, Microsoft, Google Cloud, Broadcom Tanzu, AT&T and Chainguard.

Bear Case

• The stock has not made a new high on strictly better information. The $98.10 52-week closing high was set 2026-07-06, before the guidance raise. Six sessions after the print and twelve target raises, the 2026-08-14 close of $96.17 was still 2.0% short of it.
• Sequential growth guided flat-to-thin. Q3 revenue guidance of $164M–$166M sits barely above the $163.772M just reported; the raise is carried by the full-year number, not by the next quarter's step.
• The multiple absorbed the raise. $96.17 divided by $1.00, the top of the guided FY26 non-GAAP EPS range, is roughly 96x — arithmetic on two stated figures, and richer than the level that drew Morgan Stanley's 2026-07-21 downgrade to Equal-Weight on valuation.
• GAAP still negative. Q2 2026 GAAP net loss was $(4.165)M against $0.27 non-GAAP diluted EPS.
• The July CVE question was not answered with a number. The Q2 release disclosed no security-specific ARR or RPO breakdown, so the 10%-of-ARR / 16%-of-RPO security markers from Q1 2026 have no Q2 update against which the eight Artifactory zero-days patched on 2026-07-28 could be scored.
• Scheduled insider supply kept selling into the recovery. The CEO sold 15,000 ordinary shares on 2026-08-07 at weighted-average prices between $88.83 and $94.45 under the 10b5-1 plan adopted 2026-02-19; Form 4 filed 2026-08-11.
• Two post-print marks sit below spot. Canaccord Genuity's $95 (2026-08-04) and Piper Sandler's Neutral $90 (2026-08-07) are both under the 2026-08-14 close.

Setup & Price Structure

The narrative is accelerating, dated by the 2026-08-06 beat-and-raise, the eleven-firm target reset on 2026-08-07 that moved the leading cluster from $100–$105 to $110–$120, and the 2026-08-12 swampUP announcement putting Anthropic, Cursor, NVIDIA and Microsoft on a JFrog stage on 2026-09-01. What argues against the label is the structure: the round trip from $79.79 (2026-07-31) to $96.17 (2026-08-14) has recovered the July break without yet clearing the $98.10 high made on worse fundamentals.

Levels that matter. The 2026-07-06 closing high of $98.10 is the line the new information has to take out; the pre-print range top sat in the low $80s where the old shelf broke. The 2026-08-07 gap zone brackets roughly $88.83–$94.45, the weighted-average price band disclosed in that day's Form 4 — losing it on a weekly closing basis puts price back inside the pre-print $80s. RSI(14) was 66.9 on 2026-08-14, three-month return +45.4%, distance from the 52-week high −2.0%.

Crowding observables, stated as observables: twelve target raises inside two sessions with three at the same $120 figure; an options-flow screen listing FROG on 2026-08-13; a 45.4% three-month advance into a reading near 67; and a continuing 10b5-1 disposition stream from the CEO at prices in the high $80s and low $90s. Working the other way, there is no earnings date inside the next 30 days — Q3 results are a November event — so the near-term catalyst is a user conference rather than a print, which is a lower-variance event than the one just cleared.

Catalyst Calendar (next 30 days)

• 2026-09-01 → 2026-09-03 — swampUP 2026, The Glasshouse, New York. The 2026-08-12 release states new product introductions in the JFrog executive keynotes, themed on the "Agentic Software Supply Chain," with Anthropic, Cursor, NVIDIA, Microsoft, Google Cloud, Broadcom Tanzu, AT&T and Chainguard named.
• ~2026-09-08 (est.) — post-conference analyst notes. Whether the $110–$120 cluster is defended with new product detail or the marks drift toward the $90–$95 laggards (Canaccord $95, Piper Sandler $90).
• ~2026-11-05 (est.) — Q3 2026 results. Tests the $164M–$166M guide and whether the raised FY26 $648M–$652M range gets a second lift.

Elapsed catalysts

• Ongoing — Form 4 filings under the 2026-02-19 10b5-1 plan. Each filing dates the price band at which scheduled supply met the bid; the 2026-08-11 filing covered 2026-08-07 sales at $88.83–$94.45. (passed 15d ago)

What Would Change Our Mind

The structure that would have to fail first is the post-print gap. Price re-entering the pre-print $80s would say the 2026-08-06 raise was fully discounted inside a week and that the $110–$120 target cluster is again failing to lead price, the exact mechanism that broke down between 2026-07-17 and 2026-07-31 when four notes at $100–$105 landed while the stock lost 19% from its 2026-07-06 close. The gradeable version: a weekly close below $88.

Secondary conditions, none of them price. SwampUP running 2026-09-01 to 09-03 with product introductions that draw no estimate revisions in the following fortnight would date the narrative as maturing rather than accelerating — the conference is the company's own stage, and a keynote that cannot move a number on its own stage is a hard read to argue with. A Q3 print that reaffirms rather than raises the $648M–$652M FY26 range would put the acceleration back in question. And a security metric that appears at the November print showing security's share of ARR or RPO flat against Q1's 10% and 16% would undercut the attach half of the thesis independent of the top line.

What would deepen conviction instead: a weekly close above $98.10 on volume, a disclosed security ARR figure above the Q1 10% marker, or a named self-hosted enterprise renewal cohort quantified after the July CVEs.

Correlation Notes

• Anthropic sits on both sides of the ledger. The 2026-02-20 session fell 23.1% on Anthropic's Claude Code Security launch with no JFrog-specific news; the 2026-08-06 release cites a JFrog plugin built with Anthropic, and Anthropic's Deputy CISO is a named swampUP speaker for 2026-09-01. Platform announcements from model vendors have moved this name more than its own numbers have, in both directions.
• The upstream variable is agent-generated code volume, not IT budgets. The Cursor plugin reach cited on 2026-08-06 — over 1M AI developers — is the mechanism; a slowdown in coding-agent adoption reaches JFrog before it reaches DevOps spend generally.
• Duration exposure is structural at roughly 96x the top of guided FY26 non-GAAP EPS. Moves in long-duration infrastructure software and in rate expectations transmit into this multiple with no earnings cushion beneath them.
• The CVE-exposed base is the self-hosted 47%. Cloud was 53% of Q2 revenue, so the July zero-days touched the smaller — but not small — half of the revenue line.
• Israeli incorporation with operations split between Sunnyvale and Netanya is an operational exposure US-domiciled DevSecOps peers do not carry.