NICE Ltd. NICE

Key data

The business

NICE sells the software that runs enterprise contact centers. Its CXone platform in the cloud handles the routing, recording, workforce scheduling and quality scoring behind the phone call, chat or email you get when you contact a big bank or airline. A second, smaller business, Actimize, sells financial-crime and compliance software to banks: sanctions screening, anti-money-laundering, know-your-customer. The customer-experience side is the larger business, though the exact split is not separately disclosed in recent filings. The moat is switching cost: once a 10,000-seat contact center runs its agents, data and workflows on CXone, ripping it out means retraining a workforce and re-plumbing every integration, so customers rarely leave. The thing a customer holds is a per-seat subscription, an agent's login, and that is exactly what the market now fears: agentic AI that resolves calls without a human means fewer seats. NICE's answer is to sell the AI itself, CXone plus the Cognigy voice-AI it bought, priced on outcomes rather than logins, and to bank the customers who consolidate onto one platform. Banco do Brasil, Sopra Steria and a resold RingCentral partnership are the recent proof points.

The numbers

The five-year record is a clean compounder. Revenue grew about 11% a year and operating income 25% a year, the gap being cloud gross margins scaling. Then 2026 turned.

The June quarter is where the story lives. Revenue kept climbing; the profit from operations dropped by more than a third.

Read the operating line, not the per-share one: Q2 2025 net income was flattered by non-operating gains, so the EPS drop overstates the damage. The honest figure is operating income down 35% on 7.6% more revenue. That is deliberate spend, Cognigy integration and AI buildout, not a demand collapse: new cloud ACV bookings set a second-quarter record and AI bookings an all-time high. AI ARR reached $362 million, 15% of cloud revenue, implying cloud is running near a $2.4 billion annual rate, about three-quarters of the company. Cash generation stayed real: $698 million of free cash flow in 2025, a 24% margin, stock-based comp only 5% of sales, $355 million of cash and no debt. What this memo believes that the market does not: the margin dip is an investment year, and the AI dollars are additive to seat dollars, not a swap at a loss. The print that settles it is cloud revenue growth holding double digits while operating margin turns back up over the next two quarters.

Management

The record allocates capital well. Buybacks climbed every year to $489 million in 2025, another $58 million last quarter, and every dollar now retires stock at about 11 times earnings and nine times free cash flow, which is the right time to be buying. There is essentially no insider selling to read: one officer sold $50,000 in August, plan status not disclosed, a rounding error. Executive leadership details are not separately disclosed in the most recent 6-K filings reviewed for this analysis. The test is whether the AI-first pivot the company is funding converts to margin recovery rather than permanent reinvestment. The pay-versus-profit and guidance record are not in this run's filings to judge.

How it fails or surprises you

Seats erode faster than AI fills the gap. The base is per-seat cloud subscriptions. If agentic AI deflects human agents faster than NICE sells AI to replace them, cloud growth slips below 10% and $362 million of AI ARR cannot cover the hole. First tell: cloud revenue growth and net retention over the next two prints.

The margin drop is structural, not investment. Operating margin fell 880 basis points year on year. If it does not recover as Cognigy integrates, the earnings base that makes 11 times look cheap shrinks to meet the price. Watch operating margin through the back half of 2026.

AI inflection (right tail). AI ARR is $362 million and compounding off record bookings. If AI reaches a quarter to a third of cloud within 18 months at higher prices per outcome, revenue re-accelerates and a stock at a twelve-year-low multiple re-rates toward its 28-times norm. The market pays nothing for this today. First tell: the AI ARR growth rate and cloud ACV bookings.

Closing thoughts

The next two quarterly prints settle whether the margin drop was spending to build AI or permanent damage. At 11 times trailing earnings, the lowest multiple in twelve years, nine times free cash flow with an 11% yield, no debt and aggressive buybacks, the market has already priced the seat-model fear. The last quarter showed record AI bookings and AI ARR at 15% of cloud revenue, meaning the AI pricing is additive so far, not a swap at a loss. The fatter tail is up: from a twelve-year-low multiple, margin recovery plus AI compounding re-rates the stock hard. The real risk is margins stay compressed while seat-based revenue erodes, and cheap stays cheap because the business model is breaking.

The bet is still that big enterprises keep running their contact centers on NICE and pay more for AI as they pay for fewer human agents. It breaks if cloud growth decelerates while operating margin stays compressed. The one pair of numbers that tells you first: cloud revenue growth and operating margin, together, over the next two quarters.