TBLA · Taboola.com Ltd. · Stock research

Current Thesis

Taboola sells the recommendation widgets beneath news articles a business the market has valued as declining open-web plumbing for four years. The repricing underway is about DeeperDive, its generative answer engine, and specifically what happened to it on 2026-06-16: Taboola opened the monetization layer behind DeeperDive to third-party LLMs, chatbots, virtual assistants, utility apps and agents. Taboola keeps roughly 35% of ad spend and passes 65% to the publisher or distribution partner, billing advertisers per click against its existing pool of ~20,000 performance advertisers. CEO Adam Singolda's framing "the economic layer of the AI internet" is aggressive, but the structural claim underneath it is not absurd: every conversational surface that isn't Google or Meta needs a way to make money, and almost none of them have built one.

The tape has followed. Shares traded around $5.41 on 2026-07-17, up ~18.6% YTD, after setting a new 52-week high near $5.25 at the start of the month. The 2026-07-01 Wikitree announcement a Korean-language publisher adopting DeeperDive moved the stock 4.7% on what is, in isolation, a small logo. That reaction size on a minor customer win is the interesting signal: the market has started paying for DeeperDive datapoints as a category rather than pricing each contract on its own merits. Sell-side has re-engaged, with TD Cowen reiterating a $6.00 target on 2026-07-14 and Zacks moving to a Rank #2 Buy on consensus estimates that have climbed 20.6% over three months.

The counterweight is unchanged and it is large. FY26 revenue guidance of $2.006–2.062B implies roughly 6% growth. That is the actual business. Everything else is optionality on a revenue line that has never been broken out. The 2026-08-05 Q2 print is where those two facts collide.

Bullish and bearish views on Taboola.com Ltd.

The model's bull view on Taboola.com Ltd. (TBLA), in brief: DeeperDive became an ad network on 2026-06-16, not just a product. The bear view: The AI line has no number attached to it. Both cases follow in full.

Bull Case

• DeeperDive became an ad network on 2026-06-16, not just a product. Third-party AI applications can now insert CPC ads sourced from Taboola's Realize demand pool. The 35%/65% split is a defined take rate on a surface that currently has no incumbent monetizer outside the walled gardens.
• Adoption is measurable, unlike most AI narratives. ~7M monthly active users and 10M+ reader questions per month as of April 2026, eight months after launch. Named publishers include USA TODAY, Ynet, Ouest-France, El Nacional, India Today and BuzzFeed Asia; Wikitree joined 2026-07-01 following the April 2026 Korean-language rollout.
• Unit economics inside the existing base already work. Publishers running DeeperDive report ~10% monetization lift on top of existing revenue, and Singolda has stated it carries the highest conversion rate across Taboola's entire platform. That is an upsell into 9,000 existing publisher integrations before any third-party AI customer signs.
• Estimate revisions have turned, dated 2026-07. The Zacks consensus estimate is up 20.6% over three months and the rating moved to Buy; TD Cowen reaffirmed $6.00 on 2026-07-14. The prior read on this name flagged stale May-dated sell-side that objection has expired.
• Float is shrinking under the tape. ~6.9M shares repurchased in Q1 at an average $3.41 against a ~$240M authorization, with cumulative 2025-YTD buybacks near 19% of float. A standing bid plus mechanical EPS accretion, and management has been buying materially below the current print.
• Q1 operating trend was fine. Revenue $466.4M (+9.1% YoY) and ex-TAC gross profit $168.1M (+10.8% YoY) on 2026-05-07, with guidance raised.

Bear Case

• The AI line has no number attached to it. No disclosed third-party customer count, no revenue contribution, no pipeline figure. The entire re-rate rests on a take-rate structure and a "~10% lift" anecdote. A story trading on optionality gets marked down fast when the first quantification disappoints.
• ~6% growth does not support a narrative multiple indefinitely. The FY26 midpoint on a ~$2B base is single-digit. Q2 guidance of $492–505M revenue and $49–55M EBITDA sets a bar the company should clear, but clearing a low bar is not the same as inflecting.
• Q1 earnings quality was poor. GAAP net income of $59.1M included a ~$77M one-time legal settlement. Anyone anchoring to trailing GAAP earnings or a trailing P/E on this name is anchoring to a number that will not repeat.
• Price is at 52-week highs into a print, 12 trading days out. The move from roughly $4.76 in late June to $5.41 happened on announcements, not results. Entering fresh at the highs ahead of 2026-08-05 is taking binary risk on a thesis that is not itself an earnings thesis.
• The mechanical flow is gone. Russell 2000/3000 inclusion executed at the 2026-06-26 close. Passive positions are established; the forced buyer has finished, and post-reconstitution drift is the standing base-rate risk into July and August.
• Competitive position is not defensible on paper. If monetizing AI answer surfaces proves lucrative, Google, Microsoft and the model labs themselves are better placed to capture it. Taboola's advantage is that it is early and already integrated with 9,000 publishers a distribution lead, not a moat.

Setup & Price Structure

Price near $5.41 (2026-07-17) sits above the July breakout at ~$5.25, which was itself a clean break of the May–June consolidation that capped around $4.70–4.80. Structure is constructive: a base, a news-driven breakout on 2026-06-16, a hold of the breakout through Russell reconstitution on 2026-06-26, and a second leg on the 2026-07-01 Wikitree headline. Higher lows through June and July.

Two things temper the read. Price is now approaching the $5.79 consensus average target (median $6.25, range $3.75–$7.00), so the sell-side cushion is thinner than it was at $4.76 upside from here increasingly requires target raises rather than target convergence. And an ~18.6% YTD move that came in two headline-driven gaps rather than a steady advance leaves the gaps themselves as the reference levels. The $4.80 shelf is where the June breakout would be fully retraced.

Nothing in the structure suggests a blowoff. Volume-driven gaps into a print are a different risk from parabolic exhaustion, and the appropriate handling differs: this wants position sizing discipline into 2026-08-05, not a fade.

Catalyst Calendar (next 30 days)

• 2026-08-05 (confirmed, announced 2026-07-15) Q2 2026 results. Guidance to beat: revenue $492–505M, Adj EBITDA $49–55M. The number that actually matters is any DeeperDive disclosure third-party AI customer count, ad-network revenue contribution, or MAU update past the ~7M April figure.
• ~2026-08-05, on the call FY26 guidance revision. Current $2.006–2.062B revenue / $222–240M Adj EBITDA. A raise driven by core Realize demand reads differently from a raise attributed to DeeperDive; the latter is what supports the current multiple.
• Rolling through August buyback disclosure in the Q2 filing. Q1 pace was ~6.9M shares at $3.41 average. Continued repurchase at $5+ would signal management conviction; a pause would signal the opposite.

Elapsed catalysts

• Undated, ongoing further DeeperDive publisher or third-party AI partner announcements. The Wikitree reaction on 2026-07-01 established that the market currently pays ~4-5% for these, which makes each one a live tape event. _(passed 18d ago)_

What Would Change Our Mind

The thesis breaks on a weekly close below $4.80. That level erases the 2026-06-16 breakout and returns price to the May–June base, which would mean the DeeperDive ad-network announcement produced no durable repricing.

Beyond price, three specific developments would end the read:

• 2026-08-05 passes with no DeeperDive quantification. If management discusses the AI answer engine qualitatively for a fourth consecutive quarter while guiding core revenue at ~6%, the optionality premium has no support and the open-web adtech theme should be treated as saturated.
• FY26 guidance cut or held flat with soft Q3 commentary. Q2 revenue below $492M or EBITDA below $49M against a guide issued only three months earlier would indicate the core network is deteriorating faster than DeeperDive can offset.
• A major model provider launches native publisher-side ad monetization. OpenAI, Anthropic or Google shipping an equivalent revenue-share layer for third-party AI surfaces would compress the window Taboola is currently trying to occupy.

Conversely, the read strengthens materially on a disclosed third-party AI customer with named economics, or a DeeperDive revenue line broken out at any scale.

Correlation Notes

TBLA trades with open-web adtech comparables Outbrain/Teads, Criteo, PubMatic, Magnite which share exposure to programmatic ad budgets, cookie deprecation and publisher-side pricing power. Relative strength against that group is the cleanest confirmation that the DeeperDive story is being priced as company-specific rather than as a sector beta bounce; if the whole cohort is rallying, the AI premium is not real.

Secondary sensitivities: digital advertising spend broadly, which makes the name pro-cyclical and vulnerable to any consumer or macro softening that hits performance budgets first; and the AI-application layer complex, where sentiment toward monetization-stage AI businesses drives the multiple. The Israel-domiciled corporate structure adds a geopolitical risk premium that is idiosyncratic to the name and unrelated to the operating business.

Small-cap positioning matters more than usual post-2026-06-26. Russell inclusion means TBLA now moves with small-cap index flows it was previously insulated from, adding beta that was not part of the pre-June trading profile.