In August 2025, Henry Schuck, the founder and CEO of ZoomInfo, sat down for a podcast episode with a title he presumably approved. The title was “How ZoomInfo’s CEO Rewired a 3,500-Person Company to Be AI-First.” The promotional copy was even more direct. It promised the audience would learn how Schuck and his team “reduced teams from 26 to 2 people using AI agents” and why “two-thirds of employees now use AI daily.”
It was an evangelism tour. Agents replacing humans. Teams compressed by an order of magnitude. The future is here. Look at me lead it.
Nine months later, on May 11, 2026, Schuck sent a different message. This one was an internal email announcing that ZoomInfo was eliminating roughly 600 jobs — about 20% of the company — and shutting its entire Israel operation. The framing had changed. Now AI was a source of “confusion.” Customers were “pausing.” The macro had turned “less constructive.”
“These are the hardest decisions I have to make as a CEO,” Schuck wrote. The market read the email, did the math, and dropped the stock 28.6% in a single session.
The story you are supposed to take from this is a story about a tough quarter, a difficult pivot, and a leader doing what leaders have to do. That is not the story. The story is that the CEO who spent the last eighteen months publicly cheering the AI agent era is now being eaten by the AI agent era, and the customers reducing their seat counts are running the exact playbook he was selling at conferences.
This is what happens when the moat turns out to be sand.
What actually happened
Let’s get the facts in one place, because the wire-service version was thin and the press release was a fog of corporate verbs.
On May 5, 2026, ZoomInfo’s board approved what the company is calling the 2026 Restructuring Program. On May 11, employees found out. About 340 people in the U.S., India, and the U.K. were notified the same day. About 260 more in Israel were told via a Sunday-morning emergency meeting led by Chad Herring, the Chief Human Resources Officer, who flew to Israel specifically for the announcement. By the end of 2026, the company’s development center in Ra’anana — the heart of its engineering operation — will be gone.
The financial sketch: pre-tax restructuring charges of $45 to $60 million, mostly cash severance. Expected annual run-rate savings of $60 million. Revenue guidance for fiscal 2026 cut from $1.247–1.267 billion to $1.185–1.205 billion. That midpoint represents a 4% year-over-year decline. The company was previously guiding for slight growth.
ZoomInfo did not miss the quarter. That is what makes this strange. Q1 2026 revenue came in at $310.2 million, slightly above expectations, with adjusted EPS of $0.28 against a $0.26 consensus. They beat. Then they cut a fifth of the company and slashed forward guidance, and the stock fell off a cliff to a fifty-two-week low of $4.01.
Beating the quarter while announcing twenty percent layoffs is not a tactical move. It is an admission that the business model is structurally broken and you have decided to stop pretending.
The $575 million Israeli bet
To understand why closing Ra’anana matters, you have to rewind to July 2021, the absolute peak of the conversation-intelligence hype cycle.
Back then, ZoomInfo was a freshly public darling with a $30 billion market cap. Schuck announced the acquisition of Chorus.ai for $575 million in cash. Chorus was a conversation-intelligence company that recorded sales calls, transcribed them, and used machine learning to surface insights. It was Israeli at its core — founded by Roy Raanani, headquartered in San Francisco for the U.S. money, but with the engineering muscle in Tel Aviv. At the time of the deal, Chorus’s last private valuation had been around $150 million. ZoomInfo paid almost four times that. They paid for the patent portfolio. They paid for the talent. Mostly, they paid for the narrative that conversation intelligence was about to be the next great B2B category.
It made some sense. Gong, the category leader, was raising at a $7.25 billion valuation around the same time. Conversation intelligence was the hot ticket. Buy the number two player, bolt them onto your contact database, and the bull case writes itself.
It didn’t happen. Today, Gong has roughly 45% of the conversation-intelligence market with more than 4,000 customers. Chorus has roughly 20% with about 1,500 customers. Industry reviewers now openly describe Chorus as “significantly declined since ZoomInfo’s acquisition” and “considered inferior to newer alternatives.” The Israeli engineering center that was supposed to be the innovation engine became the engineering center of a stagnating asset.
Then 2025 happened. AI-native conversation tools — Avoma, Momentum, Fireflies, Otter, a half-dozen others — started doing the same thing Chorus does, at one-fifth the price, with better summarization. The product moat dissolved. The customer base shrank. The Israeli team, which once numbered around 400 people at its peak, became a cost center sitting on a depreciating asset.
So in May 2026, five years after Schuck called ZoomInfo “arguably the largest public company in Israel” in a triumphant interview, the company sent its HR head on a one-way trip to tell 300 engineers on a Sunday morning that the office was closing.
That is the arc. Five years from peak to wind-down. $30 billion market cap to $1.9 billion. A 94% destruction of value. And it is not because the Israeli team was bad. It is because the asset they were tending was already dead and the company kept watering it.
“I genuinely believe” — the spin vs the receipts
Schuck’s public statements on May 11 follow a familiar template. The internal email to employees and the prepared remarks on the earnings call use the language of strategic clarity. The filings, the analyst notes, and the customer behavior tell a different story. Both are technically true. Only one is causally accurate.
Walk through it line by line.
The spin, on the strategic pivot: “We are simplifying our operations, accelerating our move upmarket and reducing the resources we allocate downmarket.”
The receipt: Fiscal 2026 guidance was cut by $62 million, a 5% reduction at the midpoint. The stated cause in the filing was a “regression in positive upmarket trends” combined with an “elevated pace of contraction in the downmarket segment.” Translation: ZoomInfo is losing customers at both ends of the market, not strategically choosing to retreat from one of them.
The spin, on pricing: “The industry is moving toward consumption-based pricing.”
The receipt: Management told the earnings call that customers are “increasingly confused about the capabilities of AI agents and internal tools, leading to slower purchasing activity, increased downsell pressure, and seat compression.” The shift to consumption pricing is not a strategic embrace of where the puck is going. It is a forced response to a per-seat model that is collapsing under their feet.
The spin, on enterprise demand: “Our largest enterprise customers are asking more of us, including a deeper, forward-deployed engineering motion.”
The receipt: Per-seat SaaS adoption across customers has dropped from 21% to 15% in twelve months. Outside analysts describe ZoomInfo’s core product — a B2B contact database — as “the category most directly threatened by AI agents that can assemble the same data on the fly.” Enterprise customers asking for more forward-deployed engineering is what vendors call it when their data alone is no longer enough to justify the price.
The spin, on confidence: “I genuinely believe in where this company is going, and I am more confident than ever in the team we have to take it there.”
The receipt: Five separate Wall Street firms downgraded the stock in a single day. BTIG cut to Neutral. Piper Sandler cut to Underweight with a $4 price target. Stifel cut to Hold at $4. Canaccord cut to Hold at $5. Citizens went to Market Underperform with a $2.50 target. When five analysts independently downgrade the same stock on the same day, what they are pricing in is not the CEO’s confidence.
The spin, on compassion: “These are the hardest decisions I have to make as a CEO… behind every role is a teammate, a friend.”
The receipt: This is the third layoff round in three years. About 100 cut in January 2023. Another 120 in June 2023, with a $6 million restructuring charge. Now 600, with $45 to $60 million in charges. Each round was framed as “hard” and “necessary” and “focused on the future.” Hard decisions are supposed to be infrequent. When they happen on a schedule, the language stops landing.
The spin, on Israel, then: In 2021, Schuck called ZoomInfo “arguably the largest public company in Israel,” with a market cap then north of $30 billion.
The receipt, now: In May 2026, the company is shutting the entire Israel R&D center. The market cap sits around $1.9 billion. Roughly 94% of the company’s value has evaporated in five years. The Israel office that was once the engineering heart of the company is being closed with notice delivered by a flown-in HR executive on a Sunday morning.
Read those pairs back to back and the pattern becomes clear. Schuck is describing a deliberate strategic pivot. The market is describing a forced retreat. The financial filing the same day used the word “regression” in three separate places to describe the customer base. Companies that are choosing their pivots do not describe their customer trends as regressions.
The 26-to-2 problem
Now back to the podcast.
“We explore how they reduced teams from 26 to 2 people using AI agents, why 2/3 of employees now use AI daily, and the critical role of data infrastructure in AI success.”
That is the literal pitch from the show notes of an August 2025 podcast where Schuck was the guest. Twenty-six people, replaced by AI agents, leaving two human operators. A 92% headcount reduction in a specific function. The CEO of ZoomInfo went on a podcast and bragged about it. He was telling a story about leadership. Look how aggressively I am riding this wave.
Now read the May 2026 earnings call language back to back with that. Customers are “confused about the capabilities of AI agents.” There is “seat compression.” Demand is “paused.” The “downmarket segment” is in “regression.”
Translate that. ZoomInfo’s downmarket segment is small and mid-sized companies. Those companies have sales teams. Those sales teams were the buyers of ZoomInfo seats. Those companies are now doing exactly what Schuck did in 2025 — they are replacing sales development reps with AI agents. When a fifteen-person SDR team becomes a three-person team plus an agent stack, you do not buy fifteen ZoomInfo seats anymore. You buy three. Or you buy data credits and skip the seats entirely.
This is not customer confusion. This is customer clarity. The customers learned the lesson Schuck was teaching, and the lesson cost him a fifth of his company.
There is a name for this pattern in software. We used to call it eating your own cooking, and it was supposed to be a compliment. The problem is when your customers also start eating it, and you sell ingredients.
Why ZoomInfo specifically
Not every SaaS company is in this much trouble. Some categories are agent-resistant. CRMs sit on top of proprietary workflow data. Vertical software has compliance and integration moats. Even Salesforce, whose Agentforce pricing pivot is the closest analog to what ZoomInfo is now attempting, has a database other people built their workflows around.
ZoomInfo doesn’t have that. The core product is a database of business contacts and firmographic data. Names, titles, email addresses, phone numbers, company size, technology stack, funding history. That data is, almost by definition, the most automatable thing in the B2B stack. It is information arbitrage. The moat was always that ZoomInfo had a bigger, cleaner database than what a human could assemble with a few hours of LinkedIn and Crunchbase. The moat is now that ZoomInfo has a bigger, cleaner database than an agent can assemble with thirty seconds of web requests.
That moat is not zero. It is something. ZoomInfo’s data is genuinely more complete than what a casual agent can pull. But the gap has collapsed. And the price the market is willing to pay for that gap, when it is sold on a per-seat basis, has cratered. Per-seat adoption in SaaS dropped from 21% to 15% in twelve months. That is a category-level repricing, and ZoomInfo is on the bleeding edge of it because their product is the most cleanly automatable.
The competitive landscape made it worse. Apollo.io, Clay, Common Room, Ocean.io, RB2B, a dozen smaller players — every one of them has either better AI tooling, lower price points, or both. Some are giving away in their free tier what ZoomInfo charges five figures a year for.
The bigger story
Pull the camera back and ZoomInfo becomes a data point in a much larger industry shift that has been gathering speed since late 2024. Per-seat SaaS pricing is dying. Not theoretically dying, financially dying.
Salesforce moved to per-conversation pricing with Agentforce. Intercom Fin charges $0.99 per AI-resolved conversation instead of seats. Zendesk now has two pricing models running in parallel: seats for humans, resolved tickets for agents. Microsoft is folding usage-based AI credits into nearly 60% of its service customer base. IDC is projecting that by 2028, pure seat-based pricing will be obsolete and that 70% of vendors will be forced to refactor their pricing model. Bloomberg is projecting that subscription pricing falls from 60% of software pricing today to 30% over the next decade.
The mechanism is simple. AI agents do the work of seats. When one agent absorbs the work of ten reps, you don’t need ten reps, and you don’t need ten of anything you used to buy for those reps. The vendors that had outcome-based or consumption-based pricing rails already laid down survive. The vendors that built their entire valuation on per-seat ARR get repriced. ZoomInfo is in the second bucket. So are a lot of others. Watch the next four quarters.
Be careful what you evangelize
The most useful thing about the ZoomInfo story is that it is small enough to be legible. The company has roughly 3,000 employees, one product line that matters, and a CEO who has been on the record for years. You can hold the whole thing in your head. The company’s collapse is a controlled demonstration of a force that is going to flatten a lot of bigger, less legible companies over the next two years.
Per-seat SaaS does not survive contact with agents. Information-arbitrage businesses do not survive contact with agents that can run the arbitrage themselves. And executives who spent the last two years touring podcasts to brag about AI-driven headcount reduction should not be surprised when their customers, who took notes, return the favor.
Henry Schuck got on a podcast and explained that ZoomInfo had compressed a 26-person team down to 2. The customers heard him. The customers are doing the same thing. And the company that taught them the trick is now down 94% from its peak, closing its largest engineering office, on its third round of layoffs, with five analyst downgrades in one day.
There is a lesson somewhere in all of this. It probably has something to do with not selling shovels in a mine you also dug for yourself.
Gal Ratner
Software architect, founder, and Chief Architect at Prana Entertainment. Writes about technology, capital, and the gap between what executives say and what the filings say.
In my experience their data quality was poor in comparison to other platforms, and certainly the cost of other platforms vs ZI was a no-brainer to pick the other. ZI is painfully difficult to integrate and set up. It's a dated product that's overpriced. Every client I had who had it, I made effort to get rid of it due to the pain and cost it caused.