NICE Ltd

IThe business

NICE is an Israeli software company, founded in 1986, with its American base in Hoboken, New Jersey. It has two businesses. Customer Engagement sells CXone, the cloud software that large companies use to run a contact center: the phone lines, the chat, the scheduling of agents and the recording and scoring of every call. Financial Crime and Compliance, sold as Actimize, sells software that banks use to catch money laundering and fraud. In the twelve months to 30 June 2026 revenue was $3.07B and net income was $425M C (calculated).²³⁴ The fiscal year ends on 31 December, so TTM on this page means the four quarters to 30 June. NICE has not reported Q3, and I expect that in mid-November, though no date has been announced.

Read that table from the right. Customer Engagement is five sixths of revenue and earns a 27% margin. Actimize is a sixth of revenue and earns 34%. The part that gets the attention, and the part the AI argument is about, is the big one. The unallocated line of $186M is a cost the segments are not charged, and the 20-F does not say what is in it.²

NICE reports revenue by how it is sold. Cloud revenue, which is subscription and usage fees, was $2,238M in 2025, up 12.8%, and 76% of the total C (calculated). Services, mostly support and maintenance on older installed software, fell 6.0% to $560M. Product, the licences, fell 5.2% to $147M.² The two older lines are shrinking because customers are moving to cloud, which is the plan, and the shrinkage took about 1.6 points off total growth in 2025 C (calculated). In 2016, cloud was 8% of revenue C (calculated).⁹

Americas supplied $2,466M of 2025 revenue, or 84%. EMEA supplied $325M, up 16.9%, and Asia-Pacific $155M, up 14.0%.² International is the faster part: in Q2 2026 international revenue rose 22%, and in Q1, 30%.³¹⁰ NICE says it has more than 25,000 customers in over 150 countries, including 85 of the Fortune 100, and 9,626 employees.² Revenue per employee has gone from about $206,000 in 2016 to about $306,000 in 2025 C (calculated).⁹²

Q2, and what 2026 looks like

Q2 revenue was $782.3M, up 7.6%. Cloud revenue was $609.0M, up 12.6%. Operating income on the accounting rules (GAAP) was $104.0M, a 13.3% margin against 22.1% a year earlier, and operating income as NICE adjusts it (non-GAAP) was $198.0M, 25.3% against 30.2%. GAAP earnings were $1.40 a share against $2.96, and the adjusted figure was $2.70 against $3.01.³ Revenue beat the top of the guided range. The costs did not behave: operating expenses on the GAAP basis rose 22% on 7.6% of revenue growth C (calculated), and general and administrative cost alone rose 44%, to $93.7M.³

For the full year NICE guides to revenue of $3,170M to $3,190M, up 8.0% at the middle, with cloud up 13% to 15%, and to adjusted earnings of $11.06 to $11.26 a share.³ Last year it earned $12.30 on that measure.¹¹ So the guide has adjusted earnings per share falling about 9% in a year when the share count fell too C (calculated). NICE is growing and earning less per share at the same time, and that is what the stock fell on.

That chart carries most of the story, and it needs reading with care. Cloud growth ran 20% to 27% from late 2022 through 2024, then dropped to 12.4% in Q1 2025 and has stayed between 12% and 15%. The timing points to the LiveVox purchase. NICE closed that deal at the end of 2023 and said it added nothing to the 2023 income statement, so 2024 had LiveVox in it and Q1 2025 was the first quarter without the lift.¹² NICE does not say how much LiveVox added. If the jump from 19.5% growth in Q4 2023 to 27.4% in Q1 2024 was mostly LiveVox, it was about 8 points on a base of $368M, or about $117M a year E (estimate). Against $447M paid, that is about 3.8 times sales, which is where my assumption of four times sales for bought revenue comes from E (estimate).⁹ On that reading organic growth went from the high teens to the low teens and did not halve. It is my reading and not NICE’s. The releases do not give the cause.¹¹⁶ One thing flatters the recent quarters: NICE bought Cognigy on 8 September 2025, so quarters since then include revenue the year-earlier quarters did not.² NICE does not say how much. If bought revenue is two or three points of the 12.6%, organic growth is nearer 9% or 10%. That is my estimate and not a reported number E (estimate). It also gives a signpost. Cognigy is in the base from Q4 2026, so reported cloud growth should step down then, even if nothing else changes.

Cognigy, and what AI means here

Cognigy is a German company that builds software agents that can talk to a customer, look up an account and fix a problem without a human. NICE paid $887.4M in cash and put it inside CXone.² The goodwill account rose $579M.² NICE reports a measure called AI annual recurring revenue, which was $328M at the end of 2025, up 66%, and $362M at the end of June, which it says is 15% of cloud revenue.¹¹³ The company reports no profit or margin for it.

The same technology that makes this a product is the threat to the rest of the business. CXone has historically earned money from the number of human agents a company runs on it. The 20-F says this itself, in the risk factors: a shift from seat-based revenue to consumption, or lower demand for solutions priced on the number of human agents, could hurt.² If a company’s AI handles half its calls, it needs fewer agent seats. NICE hopes to be paid for the AI instead. Whether the AI revenue replaces the seat revenue dollar for dollar is the question the whole stock rides on, and nobody knows yet.

Financial Crime and Compliance

Actimize is the quiet business. It sells to banks and brokers, who must by law watch transactions for laundering and fraud, and who change vendors rarely. Revenue was $485.4M in 2025, up 7.0%, at a 34.4% operating margin, and its cloud platform X-Sight is growing.² The 20-F names SAS, FICO and Feedzai as competitors in fraud.² It is steady and earns well. It is about 16% of revenue, so it can’t carry the group if the big business stumbles.

NICE is a low-risk-looking software company with a sharp question inside it: do AI agents turn its customers’ seats into usage, or into nothing?

IIThe moat

A software business protects itself when customers find it painful to leave. Contact-center software is among the stickiest. It is wired into the phone system, the customer database, the training of thousands of agents and the recordings that a bank must keep for years. Companies do move, but they plan it for a year and dread it. NICE has that, and it has been selling cloud versions of it to its old on-premises customers for a decade.

What holds

The number of customers is the first piece of evidence. More than 25,000 organizations, and 85 of the Fortune 100, run NICE software.² That is a lot of installed base to sell AI to, and NICE says AI was part of every new seven-figure CXone deal in 2025.¹¹ The second is the compliance record. NICE says more than 30 of its applications carry FedRAMP authorization, which US government buyers require, and the platform is built to handle card data natively.² A start-up has to earn those. The third is the backlog. Contracts longer than a year hold $3.68B of revenue that has not yet been recognized, most of it due within 24 months, which is about 1.2 times a year’s revenue C (calculated).² Cloud backlog grew 25% in the year to December, ahead of the 13% growth in cloud revenue.¹¹

The fourth is Actimize. Banks do not rip out their anti-money-laundering system, because the regulator has already approved it. A 34% margin and slow growth is what that looks like.²

What does not

The cloud growth rate is the evidence against. Growth fell from 24.5% in Q4 2024 to 12.4% in Q1 2025, and a product with real lock-in should not lose that much speed. Part of the fall is probably a purchase lapping, as I show in the financials. I can’t prove it, and the part that is not a purchase is what I would ask NICE about: it could be a pricing change, a lost large customer or a sign of competition.⁶

Cloud gross margin is the second. On the accounting basis, cloud revenue earned a 64.7% gross margin in 2024, 65.6% in 2025 and 63.8% in the first half of 2026, against 65.8% in the first half of 2025 C (calculated).²³⁴ Part of that is the amortization of the bought Cognigy technology, which is not a cash cost. Part is the cost of running AI models for customers, which is. NICE does not split them. Lock-in shows up as prices that stay up and costs that fall, and the second half of that sentence is not happening yet.

Third, the competition is large and well funded. In contact-center software NICE names Amazon Connect, Avaya, Cisco, Five9, Genesys and Talkdesk. In conversational AI it names Kore.ai, Sierra, Cresta and Salesforce. It also says it faces Zoom, 8x8 and CRM vendors that offer parts of the same job.² Amazon, Cisco and Salesforce can afford to give contact-center software away to protect a larger sale. I can’t see how much of that is happening.

Where it could fail

Agents instead of seats. If a customer’s AI does the work of 40% of its agents, the software priced per agent shrinks. NICE’s answer is Cognigy and usage pricing. The 20-F says a move to consumption could change how its revenue behaves.² The first sign would be seat counts or net retention, and NICE publishes neither. That is a gap I would like filled.

The model vendors. The 20-F says AI companies could train on NICE’s data and outputs and replicate its features.² That is a risk the company wrote down itself. The customer interactions NICE records are valuable, and the people with the best models do not need to own the phone system to use what the recordings teach.

Israel. Much of the engineering and the tax structure sits in Israel, which brings war, currency and political risk. About 0.1% of revenue is booked there, so customers are not exposed, but the company is.² I handle this in the discount rate.

The lock-in is real, and the growth rate says it is less than the 85 of the Fortune 100 suggests.

IIIThe competitive advantage

NICE’s advantage is breadth in a job that is hard to do whole. A bank can buy a cloud phone system from one vendor, scoring from a second and an AI agent from a third. Putting them together is work, and customers pay to avoid it. NICE sells the phone system, the workforce scheduling, the quality scoring, the analytics and, since Cognigy, the agent, on one platform and one set of data. In the 20-F that is the "complete set of must-have AI assets" argument.²

Against the closest page on this site

The nearest comparison is ServiceNow. The two are different companies. ServiceNow sells workflow software that runs IT, human resources and customer cases for 8,700 large organizations, and grew subscription revenue 21% in Q3 2026. NICE grew 7.6% in Q2.⁷³ But the question that decides both stocks is the same: whether software agents add to the seats customers pay for or take them away. They are one bet and not two. A reader who owns both has the same view twice, so I would not count them as diversification.

ServiceNow has the higher multiple and the faster growth. On my page its base value returns 11.2% a year at its price against the 10% I require, and NICE returns 9.6% against the same 10% E (estimate).⁷ The two are close, which surprised me. NICE’s earnings are better covered by cash: 73% of its free cash flow is left after stock pay, against about half at ServiceNow. NICE’s growth is slower and its margin is falling, and ServiceNow’s growth is quicker with less margin evidence. Neither is an obvious choice, which is itself a result.

Where NICE wins

It wins on being the only company that does all of it for a large regulated customer. The Public Safety and Justice business, which handles digital evidence for police and courts, is a niche that few large software vendors work in.² It wins on a profit record that few software companies match: GAAP operating margin rose from 13.2% in 2016 to 21.9% in 2025 C (calculated).⁹² It wins on partners. In the months before this page NICE announced a tie-up with ServiceNow (May 2026), a deal under which it can resell RingCentral’s phone product (August), and a deployment on the AWS European Sovereign Cloud (August).¹⁴

Where it loses

It loses on speed. Cloud growth is 12.6% with some bought revenue in it, against more than 20% two years ago. Sierra and Cresta, named in the 20-F, are young companies built for AI-first service, and they do not carry a base of customers who pay for seats.² It loses on cost. Operating expenses grew 22% in Q2 against revenue growth of 7.6%, and non-GAAP operating margin has fallen 4.9 points in a year C (calculated).³ And it may lose on capital. A privately funded rival can run at a loss for years to take accounts.

NICE has the widest product and the oldest customers, and it is spending to keep both.

IVManagement

Scott Russell has been chief executive since 1 January 2025. He spent 14 years at SAP and was its global chief revenue officer and a member of its executive board.² NICE’s board chose him after Barak Eilam, who had run the company for ten years, said on 16 May 2024 that he would step down at the end of that year. The stock fell 12.9% that day C (calculated).¹⁵¹ Under Eilam, by the board’s own account, revenue tripled, cloud revenue went from almost nothing to $1.6B, and earnings per share more than tripled.¹⁵ Beth Gaspich has been chief financial officer since October 2016.² She is the continuity in the room.

How they are really paid

Mr. Russell’s base salary is $900,000. His target bonus is 120% of that, $1.08M, with a maximum of 200%. The bonus is 80% financial and 20% personal goals, which for him are about cloud and AI. The financial part is 56.25% revenue and 43.75% non-GAAP operating income.⁸ His equity is the bigger number. His starting package was worth about $17.8M: 55,000 units that vest over four years, 27,500 units that vest on financial results, and 27,500 that vest on NICE’s share price return against the Software and Services index after three years.⁸ The proxy gives no weights or targets for the financial units, only that they follow year-over-year financial growth.⁸ The chief financial officer’s bonus uses the same two numbers: 53.33% revenue growth and 46.67% non-GAAP operating income, within a 75% financial weight.⁸ The 2025 accounting cost of his pay was $924,000 of salary, $1.24M of bonus and $10.5M of equity.² From 2025 half of an executive’s equity is performance-based, up from 40%.⁸

Now ask what that pays him to do. It pays him to grow revenue and to grow operating income before stock pay and before the amortization of bought intangibles. Both measures ignore two things I count: the cost of the shares he is paid in, and the cash NICE spends to buy revenue. A deal that adds $50M of revenue for $200M of cash improves both of his scores. The relative-return units pull the other way, since they only pay if NICE’s shares beat a software index, but they pay at the 25th percentile, which is a low bar. None of this is unusual. It is the standard package. It just does not reward owner cash.

The board also had to change its first draft. On 18 May 2026 NICE replaced the proposed compensation policy with a revised one that it said reflected feedback from investors.¹⁶ Shareholders approved the policy and the CEO’s terms on 9 June.¹⁷ I do not know what changed, because the filing says only that the revision responds to investors, and I did not compare the two texts. That the first draft needed revising tells me some holders were not happy with it.

Their own money

NICE is a foreign private issuer, so its officers do not file the Form 4 reports that show US insiders buying and selling. I can only see what the 20-F says. On 18 February 2026 directors and officers held 407,795 shares in total, including options and units that vest within 60 days, or 0.7% of the company. No one held 1% or more.² At $113.76 that is about $46M for all of them, before the cost of the options C (calculated). Mr. Russell’s $17.8M grant is bigger than most of that. The people running NICE have little of their own money at stake beyond what the company gives them.

The largest outside holders in the filings are Harel, an Israeli insurer, at 5.36%, per a report dated 5 October, and Principal Global Investors, which held 7.9% in February and told NICE on 19 October that it no longer holds 5%.⁵² I can’t say why Principal sold, and one fund’s sale tells me little.

The board

The board has eight members, all independent under Nasdaq and Israeli rules. Seven of the eight have sat for more than ten years. David Kostman has been chairman since 2013 and a director since 2001, apart from a year; Rimon Ben-Shaoul and Dan Falk have been directors since 2001, and Mr. Ben-Shaoul and Mr. Falk are both 81.² Caroline Tsay, a director of Coca-Cola and Morningstar, joined in September 2025 and is the only new face.² Long service gives knowledge. It can also leave a board slow to challenge a plan, and a plan to spend $887M on one deal is the kind that tests it.

Capital over a full cycle

This is the part that matters most to an owner, and the ten years of data are plain. Between 2016 and 2025 NICE generated $4.9B of operating cash flow, spent $0.7B on equipment and capitalized software, and paid $1.2B in stock pay C (calculated).⁹²¹⁸ That leaves $3.0B of owner cash over the decade. Acquisitions took $3.0B of cash: $1.16B in 2016 for inContact and Nexidia, $0.45B for LiveVox in December 2023 and $0.86B in 2025 for Cognigy, with about $0.53B across the six smaller years C (calculated).⁹²¹⁸ Buybacks took $1.6B.

Put the acquisitions beside the owner cash and they are nearly the same number. Over ten years NICE kept almost none of what it earned after stock pay: it bought businesses with it. Revenue rose from $1.0B to $2.9B, a 12.6% annual rate, and a good part of that growth was bought C (calculated).⁹² Buybacks of $1.6B paid for the stock pay of $1.2B, plus a little. The share count was 59.5M at the end of 2015 and 60.4M at the end of 2025 C (calculated).⁹² So an owner of a share at the start of the decade has the same slice of a company that is three times the size, built with the owner cash that might otherwise have been paid out. That is not a bad result. It is not the result the headline numbers suggest.

The newest chapter is different. Shares outstanding fell from 63.25M at the end of 2024 to 60.43M at the end of 2025 and to 57.36M on 27 October 2026, down 9.3% in under two years C (calculated).²⁵ NICE bought 3.64M shares in 2025 at an average of $134.43, spending $489M, and $311M more in the first half of 2026.²³ The stock is 15% below that 2025 average now. At the time those were not unreasonable prices, but they were higher than today’s. Sound capital allocation is judged at the price paid, and the price paid in 2025 is not yet looking good.

The incentives point one way and the record the other. The pay plan rewards growth in revenue and adjusted profit, and the record shows growth bought at roughly the rate cash is earned. The new chief executive has one deal behind him, and it is a big one. I give him credit for a clean balance sheet, a new $300M revolver he has not drawn, and a buyback that is finally shrinking the count.¹¹³ I hold back on judging Cognigy, because the number that settles it, how much cash it earns, is not disclosed.

The managers are paid to grow revenue, they own little, and ten years of cash went into acquisitions and into replacing stock pay.

VFinancials

NICE’s accounts look better the further you are from them. Read this section as three questions: is the growth real, is the margin holding, and what does an owner actually get?

Growth is real, and a lot of it was bought. Revenue rose from $1.0B in 2016 to $2.9B in 2025 C (calculated), and the TTM is $3.1B.⁹²³⁴ GAAP operating margin rose from 13.2% to 21.9% and is 18.5% TTM C (calculated).⁹²³⁴ The rise to 2025 is real and reflects cloud scale. The fall since then is the reason for this page.

The margin that went backwards

Non-GAAP operating margin was 31.0% in Q4 2025, 26.0% in Q1 2026 and 25.3% in Q2.¹¹¹⁰³ The company adds back stock pay and amortization of bought intangibles to get there. I do one thing differently: I take stock pay back out, because it is pay, and use the stock pay line from the cash flow statement. On that basis the margin was 27.2% in Q4 2025, 21.4% in Q1 2026 and 18.6% in Q2 C (calculated).¹¹¹⁰³

Two things moved. Costs rose, as the Cognigy staff came on and as NICE spent to sell AI: GAAP research and development rose 14.5% in Q2, sales and marketing 18% and general and administrative 44%.³ And stock pay, which had fallen to $29.6M in Q4 2025, which I think reflects forfeited awards when executives left although NICE does not say, rose back to $52.7M in Q2 2026, 6.7% of revenue C (calculated).¹¹³ The second effect is likely to be the more permanent of the two, because the unvested stock units outstanding at the end of 2025 carried $168M of cost still to be booked.² On my base revenue path, a margin after stock pay held at 18% is worth about $81 a share, and one at 23% about $115 E (estimate). I come back to this in Valuation.

Tax

Taxes are lumpy and they matter. The effective rate was 26.1% in 2023, 26.8% in 2024, 13.1% in 2025 and 48.7% in the first half of 2026 C (calculated).⁹²³ The 2025 rate was low because a preferred-regime benefit worth 8 points, against 2 in 2024, and a 2-point release of uncertain tax positions outweighed 4 points of tax on capital gains from moving intellectual property between its own subsidiaries.² The first half of 2026 is high for a related reason: the Q1 release says its non-GAAP figures now leave out "the tax rate impact resulting from the non-U.S. intercompany transaction", a phrase that is not in the 2024 and 2025 results releases I checked C (calculated).¹⁰¹¹ The Q1 2026 tax charge was $99.3M on $146.1M of pre-tax income, 68%, and Q2 was 22.7% C (calculated).¹⁰³ The Q1 bridge to non-GAAP shows a tax adjustment of minus $58.0M, against plus $10.1M a year earlier, and it does not separate the intercompany item from the ordinary tax effect of the other adjustments.¹⁰ At my 26% rate Q1 tax would have been about $38M, so up to about $61M is unexplained E (estimate). If all of it were still owed, it would take about $1 a share off my base E (estimate). Cash taxes paid were $192M in 2023, $110M in 2024 and $214M in 2025, about 26% of non-GAAP operating income after stock pay over the three years C (calculated).² The current tax charge, which was $183M, $202M and $81M, is a poor guide because it moves with the intercompany deals. I use 26% in my base case.

The balance sheet is a hint of the 2026 effect. Between December and June, deferred tax liabilities fell $92M to $18M and accrued expenses and other liabilities rose $166M to $635M.¹¹³ That is consistent with a tax bill that has been booked and not yet paid. It is not proof. H1 operating cash flow of $302M includes a $93M rise in accrued liabilities.³ If the unpaid part is tax, cash flow in the second half will be lower. I can’t see which.

What an owner can take out

The screens show NICE with about $561M of free cash flow in the TTM, which is NICE’s own measure: operating cash flow less equipment and capitalized software.³¹¹⁴ That would put the stock at 12 times free cash flow. That is not the number an owner gets, for two reasons. Stock pay of $153M is a real cost. And the free cash flow figure does not deduct what NICE paid for companies, which was $0.82B in the TTM, nearly all of it Cognigy.

So I get a range. Owner cash after stock pay was $408M in the TTM, or $7.04 a share C (calculated). If the $93M rise in accrued liabilities turns out to be unpaid tax, it is nearer $315M, or $5.44 a share E (estimate). And after acquisitions the TTM is a loss of $412M. Three years of the same measure are $301M in 2023, $551M in 2024 and $477M in 2025 C (calculated).⁹²¹⁸ The TTM of $408M sits inside that range. Net income, on the accounting rules, was $425M. Adjusted net income was $724M, and the gap between the two is mostly stock pay, the amortization of intangibles and a tax item C (calculated).³¹¹⁴

The gap between owner cash and the reported free cash flow is stock pay. The gap between owner cash and what the company really needs to keep going is the one I can’t settle. Capital spending and capitalized software cost $111M, 3.6% of revenue, in the TTM, and the depreciation in the cost base was about $98M in 2025, so the machinery of the business can be kept up for roughly what it costs to run C (calculated).³² That is a good fact. A business that needs to spend 3.6% of revenue to stand still is a very good one. What it can’t do is buy its growth for free. If I treat the past decade as the guide, acquisitions of about $207M a year, the average of the nine years since inContact, belong in the stand-still cost, and owner cash is about $201M C (calculated).⁹¹⁸ I show both because I can’t tell which way NICE will go.

The balance sheet

Cash and short-term investments were $354.7M at 30 June, and there was no debt. NICE had $1.62B of cash and investments at the end of 2024 and in 2025 used $0.86B for acquisitions and $0.46B to repay convertible notes.¹¹³ On 20 February 2026 it signed a $300M revolving credit facility with JPMorgan as agent, to 2029, with a cap on net debt of 3 times EBITDA.¹¹² Goodwill is $2.44B, 48% of total assets, and bought intangibles are $516M.³ Trade receivables rose from $738M at December to $837M at June, up 13% against revenue growth of 7.6%, which is the sort of thing I watch C (calculated).¹¹³ Days of sales outstanding rose from about 86 to about 98 C (calculated).¹¹³ It may be timing around a quarter-end. It may be longer payment terms for large customers.

Dilution from stock pay is the other cost. The company reported 1,434,753 options, mostly at par value, and 1,339,760 restricted units outstanding at 31 December, about 4.6% of the shares then outstanding C (calculated).² Those are the claims that stock pay keeps creating, and the buyback exists largely to absorb them.

Screens show $561M of free cash flow. An owner got $408M before buying anything, and spent more than that on Cognigy.

VIValuation

How this goes badly wrong

Before any numbers I ask what would most likely wreck this, and work back. I see three ways.

1. The seats go and the usage does not replace them. Customers run more of their service through AI agents, need fewer human agents, and pay NICE per resolved issue at a lower price than they paid per seat. Cloud revenue stops growing around 2029 and falls after that. My bear case is built on this. It is the way the 20-F itself says the business could go wrong.² It would first show up as cloud growth under 6%, which is not far from where organic growth may already be E (estimate).
2. The margin does not come back. Cognigy staff, AI compute and a stock pay line that is back to 6.7% of revenue keep the margin after stock pay near Q2’s 18.6%. Revenue keeps growing, and an owner gets little more than today. This is the most likely of the three, and it needs no AI disaster, only that the extra cost turns out to be permanent.
3. NICE buys its way out again, at a full price. It spent $887M on Cognigy and $447M on LiveVox. If it does another deal of that size twice in ten years, and the revenue it buys is no more profitable than the revenue it has, about $7 a share goes from my base. That is what the pay plan rewards, as the Management section says.⁸

The three share one feature. None shows in this year’s accounts, and each makes the others worse, because a business whose seats are shrinking has more reason to buy growth. The second and third are within management’s control. The first is not.

What I can’t judge

I can’t tell you NICE’s organic cloud growth, because the company does not say how much of the 12.6% is Cognigy. I can’t tell you how its revenue is split between seats and usage, because it does not publish it, or how many customers it keeps, because the 20-F has no retention figure. I can’t judge the cash tax on the 2026 intercompany transaction, because the releases name the item but do not size it or say how much is paid in cash.¹⁰³ I can’t judge whether NICE will keep buying companies at its old pace of about $207M a year, or stop. And I can’t judge how much cash Cognigy makes, because it is inside CXone. Where I can’t know a number I show the range: acquisitions of none, $100M, $200M and $300M a year give a base value of $136, $121, $107 and $93 E (estimate).

What the price implies

At $113.76 NICE’s market value is $6.59B on 57.97M shares C (calculated).¹⁵² The close is confirmed on a second feed.¹¹⁹ Take off $0.35B of net cash and the price puts a value of $6.24B on the business C (calculated).³ That is 16 times TTM net income and 9 times non-GAAP net income C (calculated). On the 2026 adjusted earnings guide of $11.06 to $11.26 it is about 10 times.³ The stock looks very low on a screen. The screen’s number is a stack of adjustments, and the chart below peels them off.

From non-GAAP net income, the price needs cash flow to fall 4% a year for ten years. From company free cash flow, it needs almost no growth. That is the screen’s reading. From owner cash after stock pay, it needs 3.5% a year, which is about what NICE does now. Take off the tax that may still be due and it needs 7.0%. And if I treat $207M of acquisitions as the cost of growth, it needs 13% a year E (estimate). The honest range of what the market is paying for is wide, and it depends on a question the filings cannot answer: how much of NICE’s growth costs real money.

Another way to see it. Hold my base revenue path and set the margin after stock pay at one level from 2027 to 2036. The price equals my base value at 22.8% E (estimate). That is just under the TTM level of 23.4% and well above the 18.6% of Q2 C (calculated).³ So the price is not assuming a collapse. It assumes the margin recovers most of the way.

The market is paying for a margin after stock pay of about 22.8%. NICE earned 27% in Q4 and 18.6% in Q2, and that gap is what the stock is about.

The discount rate

I use 10.0%, the same rate I use for the other large software company on this site. The capital asset pricing model starts from the 4.94% ten-year Treasury yield and adds beta times Damodaran’s 4.20% implied equity risk premium for 1 October 2026.²⁰²¹ NICE’s own beta is no help: 0.75 over 60 months, which would give 8.1%, but the fit has an R-squared of 0.15, and the 36-month beta is 0.51 with an R-squared of 0.04, so it is mostly noise C (calculated).²²¹ Betas use month-end price returns against SPY. A US index is the wrong yardstick for a company that trades in New York and Tel Aviv. I take a beta of 1.0 for a software company, which gives 9.1% E (estimate). Then I add 0.9 points for Israel. Damodaran puts Israel’s country premium at 2.07% and the United States’ at 0.23%.²³ Nearly all NICE’s revenue is outside Israel, so I add about half the 1.84-point gap, because the engineering, the tax structure and the board are inside it. That is my judgment and not a formula.

Three cases

Each case is built by year, 2027 to 2036, valued at 1 January 2027. Cloud revenue grows at an organic rate, and each year’s acquisitions add the revenue they buy, which I take as cash divided by four. Services and product shrink. Non-GAAP operating margin applies to all revenue. Stock pay comes off as a share of revenue, then tax. Acquisitions are a cash cost. I treat the amortization of bought intangibles as non-cash, so the cash that bought them appears as the acquisitions line.

The base case does not take management at its word. It has NICE doing a bit less than its cloud guide.

• Organic cloud growth is 8.5% in 2027 and fades to 3% by 2034, and the purchases add about 2.0 points in 2027 E (estimate). Reported cloud growth in 2027 is about 10%, against 12.6% now and the 13% to 15% NICE guides for 2026.³
• Services and product shrink 8% a year, a little faster than the 6% and 5% of 2025.²
• Non-GAAP operating margin is 26.5% in 2027 and 27.5% from 2030, below the 30.8% of 2025 and above Q2’s 25.3% E (estimate).¹¹³
• Stock pay is 6.0% of revenue in 2027 and 5.5% from 2030. It was 5.0% in 2025 and 6.7% in Q2 E (estimate).²³
• Cash tax is 26%, about what taxes paid have run at over three years.² Acquisitions are $200M a year. Growth after 2036 is 3%.

My 2027 owner cash comes to about $276M, after $200M of acquisitions. That is $4.77 a share E (estimate). Before acquisitions it is $476M, in line with the TTM of $408M E (estimate).

The bear case is the first way, with some of the second: organic cloud growth falls from 4% to −3% a year, services and product shrink 12% a year, the non-GAAP margin ends at 21% and stock pay stays near 6%. NICE keeps buying at $200M a year to no great effect, and growth after 2036 is only 1%. The bull case is the platform story coming true: organic cloud growth 12% fading to 5%, margin up to 31% and stock pay down to 4.5% of revenue, so that the margin after stock pay reaches 26.5% E (estimate).

Weighted 25%, 50% and 25%, the cases give $104 E (estimate). At 20%, 50% and 30% it is $112, and at 30%, 50% and 20% it is $97 E (estimate). The value after 2036 is 52% of my base C (calculated). That is about the usual split for a business growing 3% to 5% at the end of the ten years.

In the stress case the bear case gets a $890M deal in 2030 that earns nothing, a 29% tax rate and no growth after 2036. NICE survives it, because it has no debt and low capital needs, but $19 a share is a 83% loss from here E (estimate). I put no probability on it. It is on the page because the stock has already fallen 64% from its $315.02 close of 15 November 2021.¹

The one number

Most valuation debates have five moving parts. This one has one: how much of each revenue dollar an owner keeps after stock pay.

Move non-GAAP margin by two points and the base moves about $14. Move cloud growth by two points a year and it moves about $17. Move stock pay by one point of revenue and it moves about $7, which is the cost of the pay plan E (estimate). Acquisitions are the swing item: none gives $136 and $300M a year gives $93. Every other input is small next to those. Margin is the cleanest of them to watch, because NICE publishes it every quarter.

What about the odds? My base needs four things to go roughly right at once: organic cloud growth fading from 8.5% as I show (I say 65%), a margin after stock pay holding between 20% and 22% (60%), acquisitions staying near $200M with revenue bought at about four times sales (65%), and cash tax near 26% (75%). Multiplied, that is 19%, about one in five E (estimate). These are my judgments and not measurements. They are not independent either, which is why I do not read 19% as exact. The point of the multiplication is that a case needing four things to go right is a weaker case than it looks. The price asks for much the same four things.

The margin of safety

What does the same dollar earn elsewhere? My base returns 9.6% a year at $113.76, against 4.94% on a ten-year Treasury with no business risk and the 10% I require E (estimate).²⁰ The closest choice on this site is ServiceNow. My page has it 11% below its base value, returning 11.2% a year.⁷ On those pages ServiceNow offers a little more return for a little more growth and a good deal more price. The two stocks fell on the same fear, and they are one view of AI and seats held twice.

How good the business is and how good the price is are two questions. How good is the business? Good, and getting harder. It has customers who rarely leave, a compliance franchise with a 34% margin, a clean balance sheet and a platform that makes real money. But its cloud growth fell from 25% to 12%, its margin after stock pay fell by a third in two quarters, and over ten years it spent almost every dollar of owner cash on other companies. How good is the price? About fair. It is 6% above my base and returns a little under the 10% I require, and below the margin-of-safety price of $75 I would be paid for the questions I can’t answer. A good business at a fair price is not a bad purchase, but it is not an obvious one either.

• Price$113.76Close, Q4 2026 [1]
• Base value$107My estimate E (estimate)
• Margin of safety$7530% below base E (estimate)

My rule for a discount to base value is 25% for a business with net cash and no customer above a tenth of revenue, 30% when one question or one cycle drives results, and 35% when fixed charges sit ahead of the shares. NICE has net cash and no large customer, but the question of seats against agents drives the stock, and the bear case is 74% below the price. So I use 30%: $75. At that price the base returns 13.1% a year and the bear case 3.4% E (estimate).

The price, and how it got here

NICE closed at $66.47 on 31 October 2016, at $315.02 on 15 November 2021 and at $113.76 on 30 October 2026.¹ That is 1.7 times in ten years, and 64% below the peak C (calculated). The last twelve months were more violent than the ten years: the stock fell from $136.66 to $83.15 and rose 37% from there.¹

The 6 May drop is the one to read. In one day NICE lost 22.5% of its value on a quarter in which revenue and adjusted earnings beat the top of the guided range.¹⁰¹ What changed was the margin (26.0% against 30.5%) and the cloud guide, cut from 14.5%-15.0% to 13%-15%.¹⁰¹¹ Investors are not paying for the revenue. They are paying for what the revenue costs. The rise since the 17 June low has no event I could tie to it. NICE is up 37% since then and SPY 2.0%, so the rebound is NICE’s own C (calculated).¹²² Q2 results on 5 August beat the guided range and raised the earnings guide, though the stock closed 3.8% lower that day.³¹

Why the market may be right

I am above the market on some points and I may simply be wrong. Here is the honest case against my $107. Cloud growth fell from 24.5% to 12.4% in a quarter in 2025, and the part of it that is not a purchase lapping has no explanation. The guide is for adjusted earnings per share to fall 9% this year. Margin after stock pay fell 8.6 points in two quarters. The cost of the AI move shows up now, and the revenue shows up later or not at all. The 20-F tells you in its own words that AI could change how customers pay. Cognigy, bought for $887M, has no disclosed profit. Principal Global Investors, which held 7.9% in February, no longer holds 5%.⁵² And ten years of cash went into acquisitions, so management’s record is better at buying growth than at keeping it.

The case for the market being wrong is simpler. NICE has $3.68B of contracted revenue, 76% of sales are cloud, the margin is a choice that can be reversed, and the share count is falling about 5% a year. If AI adds to seats, the price is low. If it takes them away, the bear case is not harsh enough. I can’t tell you which, and the stock is priced as though nobody can.

Where this could be in 5 and 10 years

In my base, NICE is worth $122 a share at the end of 2031 and $143 at the end of 2036 E (estimate). Against $113.76, $143 is 2.3% a year of price growth. The rest of the return is cash, $76 a share paid out by 2036, which brings it to 9.1% a year. The returns in the table above price the business in 2036 at the rate it earns. Here I price it at my 10%, so the two sets differ.

For $122 to hold in 2031, three things must go right. Cloud growth fades from 8.5% to 4.5% without a seat decline. Margin after stock pay climbs from 18.6% to 22%. Acquisitions stay near $200M a year. For $143 to hold in 2036, that must last, because the 2036 value is a price for a business growing 3% a year after that, at 14.7 times its cash. The bear differs on growth and margin. Cloud revenue shrinks from 2029, margin after stock pay slides to 15%, and owner cash falls from $201M to $92M. Its value stays near $18 and it still pays $23 a share. The bull has cloud growth near 12% fading to 5% and a margin after stock pay of 26.5%.

The most likely way my base is wrong is the margin. If margin after stock pay stays at the 18.6% of Q2, my value today is $85, and it is $95 in 2031 and $112 in 2036 E (estimate). That needs no AI disaster, only that the extra cost is permanent. A discount rate of 11% instead of 10% gives $107 and $125.

These are estimates of value, not forecasts of the share price. The price can sit above or below value for years.

VIIThe case for the price

What most people miss

NICE needs very little capital to stand still. Equipment and capitalized software cost 3.6% of revenue in the TTM, about what it depreciates C (calculated).³² So each point of margin is almost all cash: one point on my 2027 revenue is about $25M of owner cash a year after tax E (estimate).

Its contracted revenue is growing faster than its sales. Contracts over a year carry $3.68B of revenue not yet recognized, most of it within 24 months, and cloud backlog grew 25% in 2025 against 13% for cloud revenue.²¹¹

The share count is shrinking at pace. It fell from 63.25M to 57.36M in 22 months, 9.3% C (calculated).²⁵

The bull case

In plain words: NICE owns the system that large companies run their service on, it has bought Cognigy, which the 20-F calls a global leader in conversational and agentic AI, and its customers will want one vendor for both. Scott Russell said on 5 August that Q2 brought a record second quarter for new cloud bookings and an all-time record for AI bookings, with AI ARR at $362M.³ The board backed that with a new $600M buyback in February and $311M spent in the first half.¹¹³ Margin, on this view, is a bill NICE pays now to grow later. I checked the filings for the two 5% shareholders and found only Harel and Principal, with no published view from either. I did not review analyst notes or fund letters, so this case rests on management, a biased witness.

Against reality

The price needs about my base. It needs a margin after stock pay near 22.8% and organic cloud growth that stays near 8.5% in 2027. If NICE reaches non-GAAP margin of 26.5% in 2027 with stock pay at 6% of revenue, keeps organic cloud growth above 8.5%, and keeps acquisitions near $200M a year, then $107 is possible. Today the margin after stock pay is 18.6% and organic growth is unreported and probably 9% to 10%. The bull case of $174 needs 12% organic growth and 31% non-GAAP margin, as in 2024, when it was 31.1%.¹¹

What a 10x would take. Ten times $113.76 is about $1,138, or $66B of equity on today’s share count C (calculated). By 2036 that is about 26% a year C (calculated). NICE is worth about 16 times owner cash now and earns a 13.9% net margin on the accounting rules C (calculated). At 25 times and a 25% net margin, a tenfold price needs about $2.6B of net income and $11B of revenue, against $3.07B now. That is 13% a year for ten years, a hair above NICE’s best decade, 12.6% a year from 2016 to 2025 with acquisitions in it C (calculated).⁹² At today’s margin it needs 20% a year. So it needs a higher multiple, a margin NICE has never had and its best decade of growth, all at once. NICE itself rose 1.7 times in ten years.¹ From here a tenfold rise is close to impossible, and I base nothing on it.

What they have to hit

Where I come down

The slide of cloud growth from 25% to 12% moved me, and so did the fall in margin after stock pay. The backlog and the shrinking share count moved me the other way. AI ARR growth of 66% did not, because NICE gives no profit or retention for it. I would raise my values if organic cloud growth stays above 11% for two quarters, if margin after stock pay returns above 21%, and if NICE passes a year without a deal over $300M. At a 23% cash tax rate my base would be $113 E (estimate).

VIIIRisks, and what would change my view

1. Seats and agents. The 20-F says a move from seat-based to consumption revenue could hurt, and Sierra, Cresta, Kore.ai and Salesforce are named as AI rivals.² NICE’s results would show it first as slower cloud growth.
2. Margin. Non-GAAP operating margin fell from 30.2% to 25.3% in a year, with stock pay back at 6.7% of revenue.³
3. Acquisitions. Cognigy cost $887M. A second deal of that size at a poorer price would take about $7 a share off my base E (estimate).²
4. Tax. A named intercompany item, an effective rate of 48.7% in H1, and a booked liability that may not yet be paid.³¹¹
5. Israel. War, currency and a tax structure that depends on Israeli preferred rates of 6% on qualifying income.²
6. AI vendors. The 20-F says model providers could use NICE’s data and outputs to copy its features.²
7. One idea, many stocks. NICE, ServiceNow and other software stocks that fell on the same AI fear share one view. If that view is wrong in one direction they all rise, and if it is wrong in the other they all fall.⁷

What would raise my base value: organic cloud growth above 11%, margin after stock pay back above 21%, acquisition spending that stays under $200M a year, or a disclosed retention figure that shows seats holding. What would lower it: cloud growth under 6%, margin after stock pay below 17%, a deal over $500M, or cash tax above 30%.