The $125 Billion Question

Look at the chart, and it reads like a parable of the software age. A company goes public in June 2012 at roughly a $2 billion market cap. Within five years it has 10x'd to $20 billion. By January 2025 it peaks near $220 billion — a hundredfold return from IPO in thirteen years, briefly one of the largest pure-play enterprise software companies on the planet. And then, on Wednesday, April 23rd, it reports its Q1 2026 earnings and the stock falls 18% in a single session, the biggest one-day decline in its entire history.

By that Friday's close, the market capitalization sat at $93 billion — down 43% year to date and 57% from the peak. More than $125 billion in value had changed hands from those who held to those who took the other side. The company is ServiceNow, and in this episode of The Liam Highland portfolio, the host sets out to answer a single, expensive question: has the market correctly identified ServiceNow as a casualty of the AI transition, or has the fear simply run ahead of the arithmetic?

It is a jarring setup, because the same quarter that cratered the stock also delivered numbers most companies would kill for. ServiceNow grew subscription revenue 22%. It beat its guidance and raised the full year. It is sitting on roughly $4 billion in cash, runs a renewal rate of 97%, and is — the host notes — the fastest enterprise software company in history to reach $15 billion in revenue. Nineteen months into its AI strategy, it is already booking $1.5 billion a year in pure AI revenue. "You'd probably think that I was reading off the wrong stock," the host admits. "But I'm not."

The layer that sits above everything

To understand why a beat-and-raise can lose a fifth of its value in an afternoon, you first have to understand what ServiceNow actually does. In 2022, the average large enterprise ran 367 different software applications. More recent studies put that figure at 897, and as high as 928 once you fold in the shadow-IT sprawl of tools bought quietly by individual teams. The problem is brutally simple: none of those tools talk to each other.

When something complicated has to happen — onboarding a new employee, resolving a complaint that touches billing and tech support, investigating a security incident — the work does not live inside a single app. It threads through many. Historically that thread was stitched together by humans copying data, sending emails, opening tickets, and chasing approvals. ServiceNow positions itself as the layer that orchestrates all of it: it sits above the 897 applications and runs the workflows that connect them.

The host frames enterprise data architecture as a seven-layer stack, and where ServiceNow lands within it is the whole thesis. At the bottom is data origination — ERP, CRM, HR systems, IoT. Above that, storage (Snowflake, Databricks). Then compute (AWS, Azure, GCP). Then the semantic layer, where raw data becomes structured meaning — the layer Palantir's ontology occupies. Above that, reasoning: the AI models from Anthropic, OpenAI and Google, where costs have fallen 99% in two and a half years and commoditization is fastest. Above that, action and orchestration, where agents click buttons and trigger processes — UiPath's territory. And wrapping around everything, governance: identity, permissions, audit trails, compliance — the layer that knows who is allowed to do what.

ServiceNow lives at the top — across layer six, workflow orchestration, and layer seven, the governance wrapper. And here is the insight that matters: the lower you go in the stack, the more commoditized the resource. Data is effectively free. Storage is cheap. Compute is cheap. Intelligence is getting cheap. But governance — the institutional knowledge of how a Fortune 500 company actually runs, who approves what, which compliance rules apply in which jurisdiction, how incidents get resolved across a dozen systems — compounds over twenty years. "You cannot replicate it with compute," the host argues. "You cannot download it from an API."

The numbers behind that moat are staggering: 80 billion workflows running across ServiceNow's customer base each year, 6.5 trillion transactions processed, 85% of the Fortune 500 as customers, and a renewal rate at or above 97% for five straight quarters. Yes, the host concedes, there are plenty of complaints about the UI — but not from the executives who decide what software to procure, so those complaints don't move the needle. What keeps customers is not affection. It is a switching cost.

Why a good quarter crashed the stock

If the fundamentals held, why did the stock fall off a cliff? The simplest answer, the host argues, is that the market stopped looking at ServiceNow as ServiceNow and started looking at it as a category. He lays out a "software repricing matrix" that sorts companies into four quadrants by two questions: how much coordination does the software manage, and how directly does AI drive its revenue?

AI-native infrastructure — Snowflake, Datadog, Cloudflare — is roughly flat on the year. Domain defenders like CrowdStrike, Palo Alto and Zscaler are down about 18%: painful, not existential. The disruption zone — Asana, Monday, Atlassian, Figma, GitLab — is down 51% on average, the market betting those business models might genuinely break. And the platform transformers — Salesforce, HubSpot, Adobe and ServiceNow — sit somewhere in between, with Salesforce down 26%, Adobe down 23%, and ServiceNow down 43%.

Here is the tell: ServiceNow is a platform transformer being priced like the disruption zone. The market, the host says, simply isn't differentiating. Palo Alto Networks CEO Nikesh Arora made the same point on X after the earnings, arguing the market ought to distinguish "SaaS that is impacted by AI, SaaS that needs to evolve and SaaS that benefits from AI," and predicting that "the constant paranoid reaction of the market will continue to create buying opportunities for the discerning."

But the deepest explanation came from ServiceNow's own CEO. Asked on Bloomberg about the bear case, Bill McDermott replied: "I think the bare case is really just the terminal value of software as a service companies. It has nothing to do with just ServiceNow." That phrase — terminal value — is, the host insists, the single most important concept for understanding what happened to the stock, and most viewers scrolled right past it.

The tyranny of terminal value

When institutional investors value a company like ServiceNow, the number comes from two pieces. The first is the discrete forecast — cash flows over the next five to seven years, the part you can actually model: revenue, growth, margins, free cash flow. For ServiceNow, those near-term numbers are fine. Q1 was fine. The guide raise was fine. Nobody is arguing the company stops generating billions in cash flow over the next three years.

The second piece is terminal value — the present value of all cash flows from the end of the forecast to infinity. For a company growing 20%, terminal value typically represents 60 to 70% of the entire enterprise value. "So the market is not repricing the next four quarters," the host says. "It is repricing the next 40." Will ServiceNow still grow 15% in year eight, 10% in year twelve? Will the workflow layer still command premium margins when AI agents can do the coordination themselves? If the answer shifts even slightly — from probably yes to genuinely uncertain — the terminal value collapses, and 60 to 70% of the company's value comes down with it. That alone can take 40 to 50% off the stock while the near-term cash flows never move. It is precisely how a company can beat earnings, raise guidance, and drop 18% in a single session.

Four signals, run one at a time

The four signals the host walks through, then, are not quarterly health checks. They are instruments for testing whether the terminal-value repricing is justified. If the signals hold, the market panicked and the math is mispriced. If they break, the repricing is rational.

Signal 1 — Is the demand real or engineered?

The bear argument: strip out 300 basis points of FX tailwind, 125 from the Armis acquisition, and 75 from Middle East delays, and 22% reported growth becomes 18–19% organic. The host's counter is constant-currency current remaining performance obligations (cRPO) — the dollar value of contracts already signed for the next 12 months but not yet booked as revenue, and far harder to inflate with acquisitions. ServiceNow's constant-currency cRPO grew 21% in Q1, above the bears' adjusted revenue figure — and it has held at roughly 21% for five straight quarters (22, 21, 20, 21, 21). "That's not deceleration," he says. "That's stability."

Signal 2 — Is the profit real?

GAAP net income was $469 million, up less than 2% on 22% revenue growth — the bears aren't inventing that. The culprit is stock-based compensation: $558 million in Q1, inflated by Move Works, Visa and Armis retention equity. Is that a real cost? "Yes, absolutely. Anyone who tells you otherwise is selling something." But ServiceNow is buying back stock to offset it — share count up just 1.5% year over year — the board just authorized $5 billion more in buybacks, McDermott personally bought shares, and the C-suite halted automated selling. "That's the strongest insider signal you can get."

Signal 3 — Can it charge for agents like it charged for seats?

The traditional SaaS model is per-seat. If an AI agent does the work of five seats, do you still pay for five? That question is what's breaking the disruption-zone names. The host leans on an essay from Lionus Capital, "The AI Threshold Effect," whose line he calls "the entire bull case": "For enterprises, control is not friction. It is the product." ServiceNow isn't selling AI capability — intelligence is commoditizing — it's selling the identity resolution, permissions, audit trails and orchestration that make an agent "not a brilliant liability." The evidence it's working: creator-and-other share of trailing-12-month new annual contract value has risen from 17% to 21% in a year.

Signal 4 — Is the moat cracking?

Marc Benioff has claimed customers are defecting from ServiceNow to Salesforce Agentforce. The host points to the cleanest data in the deck: renewal rates of 98, 98, 97, 98, 98, then 97 — the single dip attributable to a US federal agency closing. If displacement were happening at scale, "this number would break first. It hasn't budged." Customers spending over $5 million in annual contract value grew from 516 to 630 in a year — 22% growth — with average contract value among them rising from $14.2 million to $14.9 million. And the 2011 customer cohort now sits at 228% of its initial ACV, fifteen years in. "Customers are not leaving," he says. "They're expanding."

The math: bear, base and bull

With a $93 billion market cap, $4 billion in cash, roughly $89 billion enterprise value, and forward free cash flow around $5.5 billion, ServiceNow at about $90 trades at 16 times forward free cash flow on a company growing 20% — where the median 20% grower fetches 30 to 40 times. "ServiceNow is trading at the multiple of a slow-growing mature company." The three cases:

At $90, the host notes, you're sitting at the low end of the base range: the bull case isn't paid for, the bear case is partly priced in, and the asymmetry has flipped from twelve months ago when the stock was $211. He is careful to reject false precision — "I don't believe in probability-weighted expected returns … it's just thumb-sucking dressed up as precision" — but even a one-in-four shot at the bear case produces a positive expected value, "because the downside is already partially priced in at $90 a share."

Four dates that settle the argument

Rather than assign probabilities, the host prefers to run all three cases and track which one the data confirms, quarter by quarter. His dated checkpoints:

• July 21 · Q2 earnings — The pivotal print. cRPO below 19% and the bears are winning; above 21% and the bulls are. Operating margin around 26.5%.
• October · Q3 earnings — Two consecutive quarters establish a trend into 2026.
• January 2027 · Q4 + FY27 guide — Above an 18% organic guide is bull confirmation; below 16% flags the bear case.
• April 2027 · Full thesis test — The complete picture on demand, margin, monetization and moat.

The 2022 rhyme — and the honest doubt

There is a precedent that haunts the setup. In late 2022, Salesforce traded at $125, down 47% from its peak, wrapped in a word-for-word identical narrative: death of SaaS, can't grow, activists piling in. Eighteen months later it had nearly tripled. The host is careful not to predict a repeat — ServiceNow entered this drawdown growing 20% versus Salesforce's 12%, with stronger renewal data and a clearer AI story — but the pattern rhymes: large platforms get derated on category fears that are simultaneously real and overblown, and a multi-quarter window opens where fundamentals are intact and price is depressed.

Zoom out, and the argument becomes almost philosophical. Steel became cheap and value accrued to the architects; compute became cheap and value accrued to the software companies; now intelligence is becoming cheap. AutoGPT went viral, the host observes, but Perplexity captured the revenue — "raw power goes viral, constrained products capture durable value." If AI commoditizes the workflow layer and agents orchestrate themselves without governance, then ServiceNow's terminal value is broken and it's a 15-year compounder, not a 30-year one. But if agents need more governance, identity and audit trails than humans ever did, the terminal value hasn't shrunk — it's grown. NVIDIA's Jensen Huang, the host notes, calls ServiceNow "the operating system of our company."

So is it a buy at $90? The host declines certainty. He believes the governance bull case but holds it "with the appropriate uncertainty," because "the honest answer is nobody knows. We are all watching the same experiment unfold in real time." What he does know: at $90 you're paying 16 times forward free cash flow for a 20% grower with a 97% renewal rate — and you have July 21, October and January to tell you whether the base case holds. His closing disclosure is worth repeating: at the time of recording he held no position in ServiceNow, and none of this, he stresses, is financial advice. "You don't need heroic execution here," he says. "You need the base case not to break. And you know exactly how to test whether it's breaking."