ServiceNow Analysis

As the 'SaaS-pocalypse' phenomenon advances unabated, and software stocks continue their free-fall, rereading Michael Burry's early writings during the original dot-com collapse has felt less like a history lesson and more like a field-guide.

Understanding how he navigated the turmoil and placed his bets offers a sharper lens on the current sell-off than most of what's being published today.

ServiceNow is one of the more interesting cases caught in the crossfire. It's down 50% from its highs on fears that AI agents will cannibalise its business. Yet it sits directly on the fault line the market is trying to price - is enterprise software the victim of AI disruption, or the vehicle for it? The business tells a different story. And Burry's framework, applied honestly, helps explain why the sell-off is both more justified and less justified than either side admits.

ServiceNow - (NYSE:NOW)

ServiceNow is the workflow layer of the enterprise. Its platform sits on top of an organisation's existing systems, the likes of HR, finance, IT, security, customer service etc. It connects them, and automates the work that flows between them. If SAP is the system of record and Salesforce is the system of engagement, ServiceNow can be thought of as the system of 'action', the plumbing that turns decisions into executed tasks across departments.

The numbers reflect that positioning. Everything's lit green with very little to flag. FY2025 revenue hit $13.3 billion, up 21% year-over-year, with 97% coming from subscriptions. Gross margins have held at 77-78% for a decade. The company generates strong cash flow and carries $10 billion in cash and securities against a mere $1.5 billion in debt. Contracted but unrecognised revenue stands at $28.2 billion; more than two full years of current revenue already locked in. By most operational measures, the business has never been better. The stock, trading at roughly $105 (at time of writing) and down 50% from its July 2025 high of $211, disagrees.

What the Market is Pricing

The simplest version of the bear case boils down to this:

"every company that doesn't touch physical reality is just expensive middleware between a problem and a solution".

AI offers pure digital output at near-zero marginal cost, around the clock. So if your product is organised thinking, and thinking can now be replicated at near-zero cost, then the moats that software companies built over the past two decades, e.g. near-zero marginal cost, strong pricing power, minimal capital requirements, are straight-up eroding. Under this framework, ServiceNow is a workflow tool that will be disintermediated by AI agents that can orchestrate tasks directly.

The trigger was concrete. For instance, when Anthropic launched Claude Cowork, $285 billion vanished from SaaS valuations in roughly 48 hours. The market concluded that AI agents could replace entire categories of per-seat knowledge work software. And ServiceNow, which sells workflow automation on a subscription basis, was squarely in the crosshairs. Its trailing P/E compressed from ~155x to ~63x in under a year. It wasn't alone either, its peers Salesforce, Workday, and the broader BVP Emerging Cloud Index all sold off in sympathy, but few such companies of this scale and quality have been marked down this hard on narrative alone.

There's a contradiction worth noting here too. As Bank of America pointed out earlier in February, the sell-off is paradoxically pricing in two scenarios that cannot simultaneously be true: AI capex deteriorating because the returns aren't there, and AI being so pervasive it renders established software obsolete. If AI is powerful enough to disrupt ServiceNow, the infrastructure spending behind it can't simultaneously be collapsing. And the same strategists warning that software is being commoditised are, in the same breath, identifying the categories that I believe will actually go on to thrive.

The Qualitative Edge

Putting myself in Burry's shoes for a moment, I'd acknowledge that he'd probably leap frog the qualitative question entirely (i.e. whether ServiceNow is positioned to benefit from AI rather than be displaced by it) and instead jump straight into the 10-K to prove the stock is overvalued on any honest reading of the financials.

That's understandable, his edge was forensic accounting, not technology. Back in the Dot-com Era, he could easily tell you Cisco's profits were an illusion, but his biggest winners during the bust were actually manufactured housing and building products, not tech. His returns came from playing defence, not from picking what would thrive.

The question facing ServiceNow now though, isn't whether the stock is cheap enough on a financial buyer's basis; it's whether the market has fundamentally miscategorised the business. That's a technology question, and one I feel fits squarely in my own circle of competence.

Working in software gives me a different vantage point on this sell-off. I've seen trends come and go with the same fear cycle playing out. Whether IoT, containerisation or cloud adoption. The narrative is always substitutive, and the outcome is almost always additive. The new paradigm arrives, the market prices in wholesale replacement, and then the incumbents integrate it and expand their moats. That recognition is where I think the pure-quant Burry framework has a blind spot.

Augmentation, Not Substitution

I remember reading Clive Thompson's book, 'Smarter Than You Think' years ago. But the first chapter about 'centaur chess' has stayed with me in particular.

Thompson's central argument is that technology tends to augment human capability at the frontier rather than displace it wholesale. The famous case being a grandmaster paired with a mediocre chess engine consistently beat the world's best engines playing alone. The asymmetry came from the human setting the strategy and the engine optimising the tactics. Neither was sufficient alone, but the pairing was stronger than either part.

The SaaS sell-off is pricing in a like-for-like substitution model. The AI agent does the work, fewer seats are needed, and revenue collapses. Historically, though, that's not how tool-layer disruptions typically resolve. Thompson documents this pattern across calculators, search engines, and even writing itself (Socrates being the famous example - in warning that the written word would ruin memory).

Each time, the feared tool became infrastructure and the skill floor rose. AI will likely raise expectations of what a knowledge worker produces, making the baseline higher, not eliminating the worker.

Thompson's framework surfaces a useful filter: does this software gain leverage from AI augmentation, or does AI simply execute the task the software was wrapping? For narrow, well-defined tasks like document review or report generation, the answer is substitution; the AI executes the whole task and the human is out of the loop.

AI agents are brilliant at processing information, but they're only as useful as the data they can access and work with, they still need a filing cabinet. And it's the companies that own those filing cabinets, with decades of proprietary data, identity governance, audit trails, cross-system permissions and workflow context, that hold the real value. This is 'data gravity' in practice. It's far more practical to bring the AI to where the data already lives than to migrate it somewhere new.

So for platforms that provide that infrastructure - deep data integrations, compliance requirements and cross-system dependencies for example, the answer is augmentation. AI operates through them, not in place of them.

I'd place ServiceNow firmly in the second category. You can swap models underneath, and you can change cloud providers. But migrating your entire agent orchestration, identity governance, and asset visibility infrastructure? That's ripping out the foundations. ServiceNow isn't competing with Claude. It's the platform Claude will be licensed into.

That technical lock-in is reinforced by something the sell-off narrative overlooks entirely - the sheer psychological entrenchment of status quo bias. Enterprises that have spent years embedding ServiceNow across HR, IT, security and finance aren't evaluating alternatives from a blank slate. They're anchored to their existing investment. The switching costs are real, but the psychological entrenchment runs deeper than the technical lock-in. It's the same compromise effect I've written about before. Once an organisation has committed to a platform, the cumulative investment generates its own justification. Every new workflow built on ServiceNow makes the next one more likely to be built there too.

Meanwhile, the market is caught in its own feedback loop running in the opposite direction. The classic 'price drives narrative' reflexivity trap. When a stock drops, people reverse-engineer reasons why it should be down, and the "AI kills SaaS" thesis has provided a conveniently tidy explanation. The sell-off validates the thesis, which attracts more sellers, which further validates it, and the consensus hardens not because the analysis has deepened but because the price action keeps confirming the story. It's hard to avoid the conclusion that much of what passes for bear commentary on ServiceNow right now is post-hoc rationalisation of recent price movement.

The Moat

So if ServiceNow isn't the kind of software AI replaces, how is it positioning itself for what comes next? The abstract arguments I've laid out - augmentation over substitution, data gravity and centaur roles, only really matter if the company is actually executing on them.

To understand why its AI strategy has credibility, we first have to appreciate the depth of its existing footprint.

ServiceNow functions as the system of record for IT service management, HR workflows, security operations, and customer service across almost 8,400 enterprises, including 85-90% of the Fortune 500. As I've already touched on, its moat is rooted primarily in switching costs, and the 98% renewal rate tells you everything you need to know about how much it costs to leave.

As more departments adopt the platform, the value of the shared data model compounds, making the next adoption both easier to justify and harder to unwind. For example, if a security incident flows from Security Operations into IT Service Management, this in turn triggers a HR workflow if employee access is compromised. That cross-departmental data flow is something siloed competitors can't easily replicate. 603 customers now have annual contract values exceeding $5M, up 20% year-over-year, and at that scale these are enterprise-wide platform commitments, not simply individual departmental software subscriptions.

And crucially, this moat is widening, not eroding. Current remaining performance obligations (cRPO) grew 27% year-over-year, outpacing revenue growth of 21%, meaning that forward commitments are accelerating faster than the business is already growing. Free cash flow margins have expanded from 30% to 35% as scale economics compound. The platform is deepening its hold on existing customers while simultaneously becoming more profitable.

This isn't to say the moat is invulnerable. The most credible competitive threat is Microsoft bundling a workflow engine directly into M365 at zero marginal cost. And with Power Automate already in the market, this isn't hypothetical. But enterprise workflow management requires a depth of functionality that a general-purpose tool hasn't replicated - tracking every change across an organisation, maintaining a live map of every asset and dependency, enforcing service-level commitments, and satisfying compliance requirements. Microsoft has had years to displace ServiceNow and hasn't. Coexistence remains more likely than substitution.

This gilded moat is the cornerstone of ServiceNow's AI strategy. A decade's worth of accumulated proprietary workflow data, e.g. approval chains, incident resolution patterns and HR case outcomes. It's the kind of training dataset no competitor could feasibly acquire by outspending them. On its own, the moat neatly explains why ServiceNow won't be displaced. More interesting to me is how management is deciding to capitalise on it, and they've already committed $11.6B to the effort so far.

The Orchestration Layer

The model layer is an arms race with deep-pocketed competitors on every side, and enterprises have little incentive to lock themselves into any single one. What isn't commoditising is 'context', the proprietary data, system integrations, identity permissions, and institutional knowledge that an AI agent needs before it can do anything productive inside a real enterprise. You can't delegate a task to an agent without giving it the context to execute. That makes context the competitive battleground, and the moat I've previously described is that context. What looks like switching cost from a defensive standpoint is the exact raw material AI agents need to function. ServiceNow already occupies this ground.

Owning the context is one thing, but orchestrating what happens with it is another, and that's where the aforementioned $11.6B in acquisitions comes in. The company has been building the stack to govern how AI agents access that context, what they're permitted to do with it, and how their actions are coordinated across systems. The platform is model-agnostic by design, with ServiceNow recently adopting Claude as the default for its Build product while maintaining partnerships with OpenAI, Google, and Microsoft.

The orchestration layer sits above the model competition entirely, so you can swap the model underneath without touching the governance infrastructure. And unlike the model layer or the infrastructure layer, where multiple well-funded players can coexist, orchestration has natural winner-take-most economics. Once an enterprise standardises on ServiceNow for agent governance, the switching costs become self-reinforcing in a way that model selection never will.

It's still early days, and projecting a straight line from here would be naïve. That said, the early financial signals are encouraging. Now Assist has crossed $600M in annual contract value, roughly doubling year-over-year, and is tracking toward $1B by the end of 2026. When analysts describe this traction as "healthy but not enough," they're measuring it against the scale of the existing $13.3B subscription base. Whether that pace holds remains to be seen, but it's a credible start for a product that's only been in the market for eighteen months.

The Quantitative Caveat

That's the case for the business, but a case for the business is not a case for the stock. Burry would likely remind us that a strong business and a good investment are not necessarily equivalent. A stock down 50% from an absurd peak is not automatically cheap. For example, during the original dot-com bust, many of the biggest tech names fell only to what Burry called "no fire sale in a fundamental sense at all". The danger is anchoring to the high. When ServiceNow trades at $105, the exuberance to call it a bargain because it was once $211 is measuring a markdown, not a margin of safety.

Burry's analytical framework was built on a single premise - reported earnings lie, cash doesn't. He focused almost exclusively on free cash flow and enterprise value, ignored price-earnings ratios outright, and called return on equity "deceptive and dangerous". The discipline was in stripping a business back to what it actually generates for shareholders versus what the income statement suggests it does. Applied to ServiceNow, three areas in particular warrant that scrutiny.

Cheap relative to What?

There are two ways a stock can be cheap. It can be cheap relative to something, e.g. its own history, its peers, the market. Or it can be cheap in an absolute sense, meaning it returns enough cash to its owners to justify the price on its own terms, without needing the story to get better.

The first kind of cheapness is what most investors are measuring when they look at ServiceNow's 50% drawdown and call it a buying opportunity. The second is what a value framework demands before it will agree. They aren't mutually exclusive. A stock can be relatively cheap and absolutely cheap at the same time, but assuming one implies the other is where valuation mistakes begin.

The most relevant metric is enterprise value to free cash flow (ie. EV/FCF); what you'd actually pay to own the entire business, measured against the cash it actually produces. Not earnings, which can be manufactured through accounting discretion. And not revenue multiples either, which say nothing about profitability. Free cash flow is what the business deposits into the bank after everything is paid. The ratio between the two answers the basic question, how many years of cash generation are you actually paying for?

ServiceNow's market capitalisation is roughly $108 billion today. Subtract $10 billion in cash and securities, add the $1.5 billion in debt, and the enterprise value, lands in around $99 billion. Against $4.6 billion in free cash flow, that's approximately 22x.

Inverting the multiple makes the implied return concrete. At 22x free cash flow, an owner is buying a ~4.5% cash yield on their investment. A ten-year Treasury currently offers roughly 4.3% with no business risk attached. Burry's threshold was typically closer to a 10% free cash flow yield (around 10x EV/FCF) before he'd consider a stock genuinely cheap in an absolute sense. By that standard, ServiceNow isn't even in the same postcode.

The obvious rebuttal is that a static yield ignores growth. ServiceNow's free cash flow has compounded at roughly 25% annually over the past five years, and if anything close to that trajectory holds, today's 22x will look very different in hindsight. A business growing this quickly arguably deserves to be valued on what it will earn, not just what it earns today. That's a reasonable position, and it's the core disagreement between a value framework and a growth framework. Neither is wrong in the abstract, rather the question is which one the evidence supports.

But before that debate can be settled, there's a more fundamental question to be answered - is the $4.6 billion itself even an honest number?

A Private Company in the Public Domain

In March 2001, Burry dissected Adobe's 10-K and discovered that $357 million in stock option compensation, recognised by the IRS as real expense, never appeared on the income statement. Adjusting for it dropped operating income from $408 million to $51 million. He called companies like these "private companies in the public domain - existing for themselves, not for their shareholder owners." Adobe wasn't alone either. "Cisco, Intel, Microsoft and many of the greatest tech 'wealth creators' of the last decade" were reporting headline profits that masked the dilution shareholders were actually absorbing.

The accounting rules have changed since then. Stock-based compensation now appears on the income statement under GAAP. But the instinct to look past it hasn't. The entire SaaS industry still reports non-GAAP metrics that strip it back out, and most investors still oblige.

ServiceNow's income statement tells a version of the same story. In FY2025, GAAP operating income was $1.8 billion, with stock-based compensation totalling $2.0 billion. The company handed its employees more in equity than it earned for its shareholders. To put that another way: ServiceNow's employees were compensated more in stock than the entire business earned from operations. That doesn't mean the business is being deceptive - after all, SBC is how the entire software industry competes for talent, and ServiceNow is no outlier among its peers. But it does mean that the free cash flow figures flatter the underlying economics, and on that point, the parallel with Burry's era feels undeniable.

Where the parallel diverges is growth. Adobe, Cisco, and Intel were indeed profitable companies when Burry was writing, but their growth engines were stalling while dilution continued. The distinction to make is that ServiceNow today is still compounding revenue at 21% with accelerating forward commitments. The SBC is arguably buying something real, which is the talent that is scaling the business. And there are signs of improving discipline too. SBC as a percentage of revenue has compressed from 17.9% in FY2023 to 14.7% in FY2025, suggesting the company is outgrowing its dilution rather than being consumed by it.

But outgrowing dilution is not the same as eliminating it, and the free cash flow still reflects the gap. The $4.6 billion figure is as high as it is precisely because SBC isn't a cash outflow. The cost is being borne by shareholders through dilution, not by the business through the bank account. Adjust for it and the picture shifts materially.

At 38x adjusted free cash flow, the implied yield drops to 2.6%, which is roughly half the risk-free rate. Burry's threshold of a 10% annual return after dilution required something closer to 10x. By that yardstick, the gap between what ServiceNow appears to cost and what it actually costs is substantial.

The Buyback Treadmill

The natural follow-up to the SBC question is whether buybacks are offsetting the dilution. Management certainly frames it that way. On the Q4 FY2025 earnings call, CEO McDermott announced a $5 billion share repurchase authorisation with an immediate $2 billion accelerated buyback, positioning it as a sign of confidence: "We are one of the few companies totally in control of our own destiny".

The 10-K tells a more nuanced story though. In FY2025, ServiceNow issued 11.9 million shares to employees through stock option exercises, RSU vestings, and its employee stock purchase plan. It repurchased 10.3 million shares for $1.8 billion. That's a net increase of 1.6 million shares from compensation alone, before accounting for the 13.2 million shares issued to acquire Moveworks. Total shares outstanding rose from 1,032 million to 1,047 million. The company nearly tripled its buyback spend from the prior year's $696 million and the share count still went up.

Burry documented the identical pattern at Intel in April 2001, "$23.2 billion that Intel has spent since 1990 buying back shares. Pretty impressive. Unfortunately, there is roughly the same number of shares, adjusted for splits, outstanding now as back then". The mechanics here are the same, even if the intent is different.

ServiceNow isn't at that scale of abuse, and it's worth separating the signal from the noise. Excluding the Moveworks acquisition, organic dilution from SBC was just 1.6 million shares, so the buyback program offset 87% of employee issuance. The annual dilution rate on an organic basis is roughly 0.15%, which is modest by software industry standards. The buybacks are doing their job managing dilution, but that is precisely the point; they're just managing it, not returning capital. And as SBC grows, the buyback budget has to grow with it just to keep the share count flat.

The Lottery Ticket Test

Taken altogether, these three observations - a multiple that barely clears the risk-free rate, free cash flow inflated by stock-based compensation, and a buyback program running to stand still - paint a picture that Burry's framework would recognise immediately.

He wouldn't need to understand the AI thesis or the orchestration layer. The numbers alone would tell him to move on. But Burry has been in this position before, weighing a strong business against a price he couldn't justify, and the way he navigated it is instructive.

For example, in the summer of 2000, with the Nasdaq in freefall, a fellow investor on the Silicon Investor message board posted that he'd bought Amazon. Burry's response was characteristically blunt:

"Jim, that takes guts. Not to buy it. To post that you bought it. I won't buy it at all. I don't see a margin of safety, and I do believe margin of safety should apply in all cases."

He didn't necessarily think Amazon was worthless. He acknowledged that "visionaries can get rewarded." His objection was more so structural. Amazon's valuation depended on projections with "no fundamental basis in past reality", the price required a growth story to play out, and the balance sheet, debt-laden and pre-profit, offered no protection if it didn't. In Burry's framework, it was a lottery ticket. Perhaps one with favourable odds, but a lottery ticket nonetheless.

What's revealing to me is what Burry did buy during the same period. In May 1999, he took a position in Apple at around $34, calling it "a bona fide value stock on an enterprise value/ratio basis, generating tons of cash" with "an extremely strong balance sheet, and little downside". He saw it as "a huge contrarian play because a generation of security analysts has been trained to think that whatever is wrong with this world, AAPL is a part of it". He bought Oracle at 30x earnings around the same time, after consulting with "some techie friends and family" and doing his own due diligence. He filed both under what he called "Buffett-like tech stock long-term holds".

The distinction wasn't quality. It was price. Apple had the margin of safety. It was cash-rich, out of favour, and trading below what the business was demonstrably worth. Amazon might have been the superior business in hindsight to us now, but at the prices on offer at the time, Burry couldn't see a floor beneath him.

ServiceNow today sits uncomfortably between the two. It has Apple's balance sheet, the $10 billion in cash against $1.5 billion in debt, generating $4.6 billion in free cash flow, with a 98% renewal rate and $28 billion in contracted future revenue. But it trades at the same impasse Amazon presented; a multiple that requires the growth story to hold. At 22x reported free cash flow (or 38x if you adjust for stock-based compensation), the margin of safety that Burry demanded simply isn't there. The business has a floor, whereas the valuation doesn't.

For context, Burry was willing to pay 30x earnings for Oracle because he understood the business and could see the value. ServiceNow today trades at ~62x trailing GAAP earnings, more than double that, even after a 50% drawdown.

Burry's framework would stop here. The numbers don't clear the bar. But his framework was also built for a market where strong businesses were being exposed as illusory ones. ServiceNow is the opposite case, a business the market is treating as impaired, despite every operational metric suggesting otherwise.

A Wonderful Company at a Fair Price?

Burry's framework excels at stripping a business back to what it actually earns for shareholders. Applied to ServiceNow, it reveals real tensions beneath the surface, and on a strict reading of his criteria, the stock simply doesn't pass at present.

But there is a different school in value investing, one that evolved precisely because buying average businesses at bargain prices has a ceiling. Charlie Munger comes to mind. His most famous inversion of Graham’s teaching was that "it is far better to buy a wonderful company at a fair price than a fair company at a wonderful price." The question he would ask of ServiceNow isn’t whether the stock is cheap enough. It’s whether the business is good enough that the price is secondary.

On its own terms, ServiceNow has never been cheaper. The trailing P/E of roughly 62x is the lowest in the company’s public history, and well below the five-year average of ~218x and the ten-year average of ~221x. The EV/FCF of 22x is similarly at an all-time low. Against its peers though, ServiceNow still commands a premium.

Salesforce at 14x free cash flow and Workday at 12x have both been marked down further, and on a pure multiple basis either would sit more comfortably within a value framework. But neither is growing at ServiceNow’s pace, and among the three, ServiceNow arguably offers the strongest combination of scale, growth, and platform breadth. A stock can be cheaper than it’s ever been and still not be cheap enough, but it bears reiterating that is the first time the market has offered this business at a multiple where the growth-versus-price debate is even worth having.

By most measures of business quality, ServiceNow passes that test comfortably. A 98% renewal rate, 77-78% gross margins held for a decade, revenue compounding at 24% annually over five years with free cash flow margins expanding from 30% to 35%. The cash-on-cash return on tangible capital is roughly 50%. These are not the characteristics of a business you want to own only if the price is perfect. These are the characteristics of a compounder.

But perhaps the most telling number is the one the market is implicitly providing. A rudimentary reverse DCF (working backwards from the current price rather than forward from growth assumptions) reveals how the market actually expects this business to perform.

The logic is straightforward. ServiceNow’s enterprise value is $99 billion. It currently generates $4.6 billion in free cash flow. If you assume an investor demands a 10% annual return (the standard hurdle for owning an equity over a risk-free bond), that the business eventually matures to grow at roughly the rate of inflation (~3%), and that at maturity it would trade at 25x its free cash flow, then the question becomes: how fast does the $4.6 billion need to grow over the next decade for the maths to work at today’s price?

The answer is roughly 8-9% per year.

For a business that has compounded free cash flow at 28% annually over the past five years, with 27% cRPO growth signalling continued acceleration and a platform still expanding into new product categories and geographies, 8-9% is an extraordinarily low bar. The market is pricing ServiceNow as though its growth engine is about to stall to a third of its historical pace. The stock has moved from being priced for perfection to being priced for mere competence. At today’s price, you are paying almost nothing for future growth.

As we've established, the margin of safety here isn't in the current-year multiple. On that basis alone, the stock is neither fair nor cheap. But the market has embedded a margin of error so wide it starts to look like a margin of safety in reverse. Burry would recognise the concept even if he'd apply it differently. I can't call ServiceNow at today's price a bargain. But equally, I can't ignore that it may be a historic entry point for a business of this calibre. The risk of not buying may be greater than the risk of overpaying. In Munger's language, that's the opportunity cost of inaction. And for a compounder of this quality, it compounds too.

When the quantitative evidence can appear ambiguous, being too expensive for a value framework and too cheap for what the business is actually doing, I find it useful to look at what the people with the most at stake are doing with their own money. Frameworks can disagree, but capital commitments rarely do.

The Signal from the Inside

So who's committing capital? The executives who know this business better than any analyst, framework, or model ever could, and it's obvious they're not hedging.

CEO Bill McDermott has committed approximately $23 million of personal capital to purchasing ServiceNow shares. $20 million was already pledged in late January, alongside a commitment to lead the company through 2030, and a further $3 million purchased on February 27 at prevailing market prices, the earliest date he could buy without triggering short-swing profit rules under Section 16. These are open-market purchases with after-tax personal funds, not stock grants or options exercises. McDermott called it "a once-in-a-generation moment with ServiceNow", the kind of statement that's easy to dismiss from any CEO, less so from one who's already put the money behind it.

But it goes beyond just the CEO. On February 13, five senior executives - the CFO, the Vice Chairman, the Chief People Officer, and the company's Special Counsel among them, all simultaneously terminated their Rule 10b5-1 trading plans, cancelling all future pre-planned stock sales.

For the record, these are pre-scheduled, automated selling programmes that executives set up months in advance to avoid any appearance of insider trading. Cancelling them is not something done lightly; it forfeits future liquidity and invites scrutiny. Cancelling them en masse, during a 50% drawdown, while the CEO is simultaneously buying with personal capital, is as close to a coordinated signal of conviction as securities law permits. Whatever the quantitative framework says about the multiple, the people with the deepest understanding of this business are betting that it's wrong.

The institutional filings tell a similar story. Q4 2025 13F data shows 315 new institutional positions opened against 213 closed, a net addition of over 100 holders during the worst quarter of the drawdown. Among hedge funds specifically, 69 new positions were initiated versus 51 closed, with aggregate 13F shares up 9% quarter-over-quarter. Perhaps most telling is the options market. The put/call ratio dropped from 1.24 to 0.94 in a single quarter, flipping from bearish to bullish for the first time since the sell-off began.

The insiders aren't alone in their conviction. And these aren't passive funds mechanically rebalancing either, it's active managers that are making concentrated bets. Bain Capital Venture Investors opened a new position of over 1.1 million shares in Q4, making ServiceNow nearly 69% of its portfolio. Lightspeed Management initiated a 1.8 million share position at roughly 20% of its portfolio. When high-conviction active capital is flowing in during a narrative-driven sell-off, it suggests the money doing the deepest work views this as mispricing, not fundamental deterioration.

Conclusion

This analysis started with a qualitative conviction and then tested it against the most unforgiving quantitative framework I could find. Burry's lens is the most honest place to stress-test any investment thesis - strip the story away, ignore the narrative, and ask what the numbers actually say. Applied to ServiceNow, they say the stock isn't cheap; not on an absolute basis, not after adjusting for stock-based compensation, and certainly not by the standards Burry himself would apply.

But the same numbers, read differently, tell another story. The market is pricing in 8-9% growth for a business compounding at 28%, and the insiders are buying, not selling.

Markets tend to price first-order effects quickly and second-order effects slowly. The first-order narrative driving this sell-off is straightforward. AI can automate workflows, therefore ServiceNow is disrupted. But first-order thinking stops at the obvious consequence and misses what follows from it. The second-order reality, the one the market hasn't fully processed, is that AI makes what ServiceNow does more valuable, not less. The more AI agents an enterprise deploys, the more it needs governance, identity permissions, audit trails, and cross-system orchestration. That is the core of ServiceNow's business. The market in my mind, is pricing the threat while overlooking the tailwind.

I don't know whether ServiceNow at $105 will look cheap in hindsight. No one does. But the gap between what the market expects and what the business is delivering is wider than I've ever seen for a company of this quality, and the people closest to the business are signalling the same thing with their own capital. Burry I imagine would pass. Munger, I suspect, would not.