ServiceNow (NOW): a company that is hard to leave

☕ The short version (for those with 60 seconds)

• 🏢 What it does: gives big companies one place to handle internal business: from a broken laptop, through leave requests, to a customer complaint.
• 💰 How it earns: subscriptions. About 97% of revenue is recurring.
• 🔒 Why customers stay: replacing such a system means months of work, big costs and even bigger migration risk, so swallowing a price increase is easier.
• 🧮 My valuation: 163 USD per share in the base case, against a price of 137.87 USD. But if you count share-based pay as a full cost, you get about 115 USD. I explain the gap below.
• ⚠️ Next test: Q3 results on 28 October, after the market closes.
• 🤝 Full disclosure: I own shares in this company, and for years before that I used its system every day.

History: where this company came from

📉 The man who lost everything just before fifty

This story starts with a failure, not a success.

In the early 2000s Fred Luddy was chief technology officer of Peregrine Systems. The company collapsed under an accounting scandal, and his entire fortune went with it, because Luddy kept it in its shares. As he later recalled, 35 million dollars turned into zero in an instant.

Most people in that spot would look for a quiet job. Luddy did the opposite. In 2004, two weeks before his fiftieth birthday, he started a new company. He explained later that after fifty a fresh start would have seemed unrealistic. He called it Glidesoft (formally the company may date from a year earlier).

💡 A broken laptop as a business idea

The idea did not sound like a revolution. Just a system for handling IT requests. The devil was in the detail: it ran in a browser, not on servers the customer had to install and look after. For those times that was cheeky.

You know this from work. Your laptop breaks and instead of hunting for "that computer guy", you file a request in one place. It lands with the right people on its own, you can see what is happening to it, and they know what to do. Simple? That is exactly why it worked.

In 2006 the company took the name we know today: Service-Now.com.

🎲 The best decision is the one not taken

At the end of 2009 the Sequoia fund believed in the company and led a 41 million dollar funding round. A year later an offer arrived that the industry still talks about: VMware wanted to buy the company for about 2.5 billion dollars.

A few years after Luddy lost everything, billions were on the table. The company was reportedly willing to accept. Then Doug Leone of Sequoia stepped in and convinced the board that ServiceNow would be worth ten times more.

The market proved him right faster than anyone could expect. In 2011 Frank Slootman took the helm, and on 29 June 2012 the company debuted on the New York Stock Exchange. Shares were sold at 18 dollars, above the range the company itself had set. By the end of that year ServiceNow was worth 3.7 billion dollars, more than VMware had wanted to pay.

👔 From help desk to company command centre

Slootman left in 2017. After a short spell under John Donahoe, Bill McDermott, formerly head of Germany's SAP, took over at the end of 2019. Under him a tool for the IT department grew into something much bigger: the Now Platform, one system in which a company runs IT, security, HR, customer service and risk.

When the world went mad for artificial intelligence, ServiceNow did not stand aside. In July 2023 it added generative AI to the platform (Now Assist), and then went shopping: in March 2025 it bought Moveworks for about 3 billion dollars, and in December it announced the purchase of Armis for 7.75 billion. A company that started with tickets for broken laptops sold over 13 billion dollars of services in 2025.

☕ Why this company ended up in my portfolio

I have to admit something here. Before I started analysing ServiceNow, I simply used it. Every day. And honestly, I could not imagine how my corporation would function without it.

So when the "SaaSpocalypse" came, the big sell-off of companies selling subscription software, and ServiceNow's share price dived, buying was a formality for me. I had seen from the inside how deep this system sits in the daily work of thousands of people.

It was similar at my previous firm with CrowdStrike, after its flash crash. But that is a story for another coffee.

So I have a stake in this story, and I say so openly. All the more reason to make sure the numbers keep talking, not my fondness for a tool I used every day.

🔒 And here it gets interesting

Customers who have deployed such a platform usually do not leave. I know that from my own desk. And the company has just decided to test how much that loyalty is worth.

In April 2026, according to market reports, ServiceNow replaced its five licence tiers with three packages (Foundation, Advanced and Prime), with AI included in each. From 1 July the old packages can no longer be bought. The CFO spoke of expected price increases of 20-30%.

Will the moat hold when its owner raises the toll for crossing the bridge?

Products, services and the revenue model

🏢 Picture a company with 10,000 employees

Every day somebody needs something:

• a new team member needs a laptop, accounts in several systems and an access card,
• someone else takes leave,
• a customer files a complaint,
• the security team spots a suspicious login.

In many companies all of that happens in emails, spreadsheets, phone calls and in the style of "I'll remind you tomorrow".

Example (an illustration, not a description of a real customer): Ania is joining the company.

• ❌ Without a shared platform: HR emails IT, the email gets lost, and Ania sits for three days without a computer or access to systems.
• ✅ With ServiceNow: HR ticks "new employee" in the system, and the system creates tasks for IT (laptop), security (access) and administration (card) by itself. Everyone sees what they have to do and by when.

That is what ServiceNow sells: order in the flow of work. The modules cover IT service and operations management, security, HR, customer service and risk.

Recently AI agents have been added: programs that carry out part of the steps in such a workflow themselves. In our example it is as if some of the steps for onboarding Ania were done not by a person but by a program.

💸 Why the customer pays

Because they get order where there was chaos, and they save the time of expensive people.

The company says its AI products passed 1 billion dollars of annual contract value (ACV, meaning how much customers pay for them per year) in Q2 2026.

🔁 Where the money comes from

Think of a phone plan or a streaming subscription, only for companies and for millions of dollars.

• The customer pays every year for access to the platform.
• In 2025 subscriptions brought in 12.88 billion of 13.28 billion dollars of revenue, about 97%.
• That makes revenue predictable. The company largely knows how much it will earn next year before that year begins.

You can see it well in cRPO. That is money from contracts already signed that the company will show as revenue within the next 12 months. In Q2 2026 it was 13.20 billion dollars, 21% more than a year earlier.

🥊 Who can take that money away

1. Competitors. Microsoft and cloud providers build AI into tools the customer already has and already pays for.
2. The customers themselves. They can build their own tools on AI models instead of buying ready-made ones.

ServiceNow's protection is switching cost. Replacing a platform that runs the processes of a whole company is like replacing the wiring in an office tower that is in use. It can be done, but few people do it to save on the electricity bill.

Because the cost of switching is far more than the price of a new licence:

• 🗄️ Data migration. The system holds years of history: requests, changes, access rights, contracts. Moving it to another tool always carries the risk that something gets lost, breaks or loses its links. And in a large company, especially a regulated one, lost history is not a trifle but an audit problem.
• 🔌 Integrations. The platform is plugged into HR, accounting, email and security systems. Every such connection has to be built again and tested.
• 👩‍🏫 People. Thousands of employees know their way of working by heart. A new tool means training, new habits and a period when everything runs slower and requests get lost because "it was different in the old one".
• ⏳ Transition period. For months a company often runs both systems at once and pays twice, while the IT team deals with the move instead of developing the business.

All of this together means a price rise of several dozen percent can still come out cheaper than moving house. And the whole thesis rests on exactly that.

📊 What profitability depends on

Two things:

1. How much it costs to run AI. AI models use a lot of computing power. Non-GAAP subscription gross margin (how much of every 100 dollars of subscription is left after the cost of delivering it) fell in Q2 2026 to 80.5 dollars from 83 a year earlier. For the whole of 2026 the company assumes 81%.
2. How much of pay is in shares (SBC). The company reports a non-GAAP operating margin of 31% for 2025, but once pay in shares is counted (GAAP) it is only 13.5%. That gap is the heart of the whole valuation, so it gets its own chapter below.

The thesis in 3 sentences

1. 🔒 ServiceNow is a company that is hard for customers to leave, it grows by over 20% a year and its balance sheet carries no excessive debt.
2. 🧮 Under my assumptions, where I treat pay in shares as a non-cash cost, fair value is about 163 dollars against a price of 137.87 dollars, so in my opinion the stock is cheap today.
3. ⚠️ That value depends heavily on how SBC is treated and on whether growth after the price rises is lasting, and the first hard test is the Q3 results.

Financial position

📖 How to read this table without a dictionary:

• GAAP is the official US accounting standard. It counts everything, including pay in shares.
• Non-GAAP is a result "cleaned" by the company itself, mainly of the cost of share-based pay. It usually looks nicer.
• FCF (free cash flow) is the cash that really stays in the company after paying for operations and investment. It is what the company can use to repay debt, buy other companies or buy back shares.

🔭 What the company guides for 2026:

• subscription revenue of 15,760-15,780 million dollars (+22.5%),
• non-GAAP operating margin of 31.5%,
• FCF margin of 35%,
• in Q3 alone, subscription growth of 20.5% y/y.

In Q2 the company did not buy back shares (in the first half it bought back 2,225 million dollars).

🏦 Debt. At 30 June 2026 net debt was 810 million dollars (debt 7,517 million, cash and securities 6,707 million). As recently as December 2025 the company had net cash of about 8.6 billion dollars. Acquisitions and debt issues made the difference. The balance sheet is still healthy, but the direction of change is clear.

🏷️ What the whole company costs. The market values ServiceNow at 142.6 billion dollars (price 137.87 USD at the 7 October 2026 close), with a P/E of 83.6 (TTM, GAAP) and a 52-week range of 81.24-192.97 USD.

What does a P/E of 83.6 mean? For every dollar of annual accounting profit you pay 83.6 dollars. If profit never changed, the company would "earn its own price" in over 80 years. This is not a cheap stock on simple multiples, so the whole thesis rests on the cash the company will generate over a decade, not on today's profit.

💡 Pay in shares (SBC), in plain language

This is the most important chapter for understanding the valuation, so let us take it in pieces.

What SBC is

Technology companies pay part of their staff not only a salary but also shares. A programmer, say, gets a transfer every month and on top of that a package of company shares every quarter.

At ServiceNow that is a lot: 14.7% of revenue in 2025 and 16.4% in Q2 2026. So out of every 100 dollars customers pay, about 15-16 dollars is "spent" on pay in shares.

🍕 Dilution, or a pizza cut into more slices

Imagine a pizza cut into 8 slices. You hold one, 12.5% of the whole.

To pay its staff the company adds new shares. That is like cutting the same pizza into 9 slices. Your slice is now smaller (11.1%), though nobody took anything off your plate.

That is shareholder dilution: the company's results are split across more shares, so each share gets less.

Companies often try to offset it by buying shares back from the market. ServiceNow bought back 2.2 billion dollars of shares in the first half of 2026. Only then the cost is paid in real cash.

📄 Paper profit versus 💵 cash

Here is the catch that makes two people looking at the same company reach completely different conclusions:

• In the income statement (GAAP) SBC is a cost. Hence an operating margin of only 13.5%.
• In the cash flow statement SBC is added back, because no money left the account. Hence the impressive 4.6 billion dollars of free cash and a 35% FCF margin.

Both numbers are true. The question is: is pay in shares a cost or not? Critics of SBC answer: it is a cost, just paid differently. Not by the company from its account, but by you, the shareholder, with a slice of your pizza.

⚖️ Two valuation variants, side by side

Variant A is kinder to the company because it ignores dilution. Variant B is harsher because it assumes somebody does pay for those shares in the end.

I show both on purpose. My value of 163 USD is not a fact, it is the result of a choice that you can challenge. If you prefer to count conservatively, variant B is your reference point.

Valuation

🧮 How I calculate it (no formulas)

I use a DCF model. It sounds scary, but the idea is simple: how much all the money the company will earn in the future is worth today.

The catch is that 100 dollars in 10 years is worth less than 100 dollars today: there is a risk it will not be there, and today you can put it to work somewhere. So I "discount" future money by the cost of capital, the minimum annual return I expect for the risk.

What exactly I assume in the base case:

• 📈 Revenue growth averaging about 18.6% a year over 2026-2030. That gives about 31 billion dollars of revenue in 2030, including over 30 billion from subscriptions, which is what the company itself guides to.
• 💪 EBITDA margin before SBC (operating profit before depreciation and share-based pay) rising from 37.5% to 40%.
• 🎯 Cost of capital of 9.5% a year.
• ♾️ Perpetual growth of 4.0%, the pace at which I assume the company will keep growing for ever after the ten-year forecast.
• 🔄 Negative working capital: customers pay up front, so the company receives cash before it delivers the service.

The model covers 10 years: five years of detailed forecast and five years of gradually slowing growth. I run three scenarios weighted 25/50/25. You can download the model and replace my assumptions with yours.

🎭 Three scenarios

The weighted average is about 162 USD, which is 17% above the price.

🔄 The same thing from the other side

You can also ask the reverse: what assumptions "sit" in today's price? A price of 137.87 USD corresponds, in the base scenario, to a cost of capital of about 10.5%. So the market demands one percentage point more for this risk than I do.

🎚️ What moves this number most

1. SBC. Described above: changing the approach moves the base value from 163 to about 115 USD. At its Financial Analyst Day the company committed to cut SBC below 10% of revenue by 2029, and I took that direction for variant B.
2. Perpetual growth. At 3.5% the value falls to about 153 USD, at 5% it rises to about 189 USD.

For the curious: the implied terminal EV/EBITDA multiple is 13.7 in the base scenario, and terminal value makes up 66% of enterprise value. Details are in the model.

The result is a fair value following from the assumptions I made, not a price target or a recommendation.

Catalysts (what could help the company's earnings)

• 📅 Q3 results (28 October, after the market closes). The company guides to subscription growth of 20.5% y/y. If cRPO and gross margin hold up after the pricing change, that will be a strong argument that growth is lasting.
• 🤖 Making money from AI. ACV of AI products has passed 1 billion dollars, and in September 2026 the CFO spoke of raising the 2026 target to 1.5 billion. In Q2 the company had 123 deals worth over 1 million dollars of ACV (almost +40% y/y) and 658 customers with ACV above 5 million dollars (+23%).
• 💵 Returns to shareholders. In January 2026 the board approved an additional 5 billion dollar buyback programme, and in the first half of 2026 the company bought back 2.2 billion dollars of shares.
• 🎯 Company targets for 2030 (Financial Analyst Day, 4 May 2026): over 30 billion dollars of subscription revenue, 30% of ACV from AI and "Rule of 60+" (growth rate plus free cash margin above 60%), plus cutting SBC below 10% of revenue by 2029. These are management's declarations, not forecasts.

Risks (what could prove me wrong)

• 🤖 AI as a competitor, not an add-on. AI agents may replace work that is paid for per seat today. Fewer people in the IT department can mean fewer licences. On top of that, customers may build their own tools on AI models. This risk is long-term and hard to measure in a single quarter.
• ⏩ Demand "borrowed" by the price rise. Analysts worry that some customers renewed early to avoid the new price list. The CFO spoke of early renewals as part of the consumption model. If the worries are justified, Q3 and Q4 will show weaker growth than the last report suggests.
• 📉 Margin. Non-GAAP subscription gross margin fell in Q2 to 80.5% from 83%. If AI costs grew faster than prices, my assumed rise of EBITDA margin to 40% would be too optimistic.
• 🍕 Pay in shares and dilution. Described above: with a different treatment of SBC the base value falls by about a third.
• 🏷️ Valuation. A P/E of about 83 does not forgive wobbles. After the Q2 results the share price fell despite beating expectations on profit and revenue, because the second-half guidance turned out weaker.
• 🧩 Integration and debt. The Moveworks and Armis acquisitions turned net cash into net debt.
• 🔐 Security. In June 2026 a critical vulnerability, CVE-2026-6875 (CVSS 9.5; remote code execution without authentication), was patched. The company saw no evidence that it affected instances it hosts itself. This is a reputational risk, not proof of worse results.
• 🧪 Model limitations. Terminal value is two thirds of the total. Historical data from before 2025 are incomplete and do not affect the result, because the model starts from 2025 revenue.

Conditions that invalidate the thesis

This is the most important paragraph of the text, so I will write it honestly.

We are at a moment when everything converges. The Q3 results are the first report after the pricing change and after AI was written into the basic packages. Everyone will be looking for one thing: whether the outlook is what management promises.

📈 What the chart says. A moving average is simply the average price over the last X days or weeks. It helps you see the trend without daily noise.

The figures below are a reading from 2 October 2026. The chart above runs live, so the levels will gradually drift away from it.

• On the daily chart the 50-day average (SMA50, 128.34 USD) sat about 10.8% above the 200-day average (SMA200, 115.86 USD). That is the so-called golden cross, which has already formed.
• On the weekly chart the price stood right on the EMA200 line (137.74 USD), just below the SMA200 (143.36 USD) and the resistance zone of 137.7-143.4 USD.

Such boundaries can turn either way, and once they move, the move tends to be decisive.

✋ When I will admit I was wrong. One weaker quarter does not overturn the thesis of a company that is hard to leave. Only a break in the moat itself would. So I will call the thesis out of date only when something happens that shows customers really are leaving, or that AI is eating the profits:

1. cRPO growth falls below 15% y/y in two consecutive reports (in Q2 it was 21%);
2. non-GAAP subscription gross margin falls below 75% in two consecutive reports (in Q2 it was 80.5%);
3. the company cuts its 2026 subscription revenue guidance below 15.46 billion dollars, i.e. to growth under 20% (today it guides to 15.76-15.78 billion, +22.5%);
4. the price closes a week below the lowest level of the last 52 weeks (81.24 USD).

🟡 Yellow light. There are also weaker signals that do not overturn the thesis but tell me to go back to the model and check the assumptions: cRPO growth below 18% y/y, subscription gross margin below 79%, any cut to the guidance for 2026, or a weekly close clearly under the weekly EMA200 (137.74 USD) with no return within two weeks. If any of them lights up, I will write about it openly in a follow-up.

The thresholds are mine, chosen in advance so I do not move them after the fact. After the results I will be back with a follow-up text, whether or not I was right.

And you? How do you count pay in shares: as a cost or not? Tell me in the comments.

Disclosure of interests

My cup: I hold a position in the company. That means I have a stake in this thesis turning out right, and that is exactly why the assumptions, scenarios and model are open and the invalidation condition was written down in advance.

Disclaimer

Content on Kawiarnia Rynkowa is for informational and educational purposes only and is not investment, tax or legal advice. Investing involves risk, including loss of capital. The author holds positions in the instruments discussed — see disclosure in the text.