GM. This is Milk Road PRO, turning last week's SaaS thesis into the first real position.
Today's edition features our latest PRO report, the first real buy off last week's SaaS thesis. ServiceNow is down about 50% in a year, yet it still grows 22%, renews 98% of its revenue, and now trades at its cheapest multiple in a decade. You'll get the opening below, with the rest on the site.
Here's what we've got for you today:
- 📉 Why a genuinely great company got dumped 50% and now trades at a PEG of 1, the cheapest it's been in ten years.
- 🏢 The land-and-expand engine behind the moat: 98% renewal, 600+ customers paying over $5M a year, and 91% of new business coming from accounts already running five or more products.
- 🤖 The "AI Control Tower" bet, with Now Assist crossing $1B+ in contracted value and management targeting $1.5B in 2026.
- ⚠️ The bear case worth checking: compute costs squeezing gross margin to around 80%, plus the exact tripwires that get this position cut.
Squid now powers instant cross-chain access for RLUSD across XRPL, Ethereum, Base, Optimism and beyond. Stop bridging. Start Squidding.
Prices as of 2:00 p.m. ET. Powered by CoinGecko.
THIS STOCK IS DOWN 50%. I JUST BOUGHT IT
ServiceNow is down about 50% in a year.
It still grows 22% YoY, renews 98% of its revenue, and now trades at a PEG of 1 suggesting fair valuation looking at PE divided by expected earnings growth. That is historically cheap for a company that the market never used to let you touch at a fair price. The selloff treated it as if AI were about to hollow it out. I think the market read that backwards.
Last week, I argued SaaS isn't dead. ServiceNow is my first buy on that thesis, and the cleanest version of it.
The bet in one line: when AI agents do the actual work, the systems that own the records, enforce the rules, and keep the receipt are where that work has to land. ServiceNow owns that layer for IT, HR, customer service, and security inside the world's biggest companies.
I put the first position at around $106. Here's why this name, and the one number that decides whether the bet is right.
And before we get into it - you can see my full portfolio right here, including my move on this stock and which crypto tokens I hold (and why). It’s all in Milk Road PRO and will only cost you a buck for a quick trial (you can cancel anytime).
WHAT IS SERVICENOW
Let’s start with what it actually is, because the company’s name doesn’t really tell you anything.
ServiceNow provides the ticket routing and infrastructure for large companies: the IT requests, HR cases, customer-service tickets, and security incidents that move work through a big organization. Something happens ("my laptop's dead," "onboard the new hire," "this customer is furious," "we've been breached"), and ServiceNow turns it into a job that gets tracked, routed to the right person, approved, logged, and actually closed. Not a chat about the work. The work itself, with a record of who did it and whether they were allowed to.
It does this for one department after another, all on a single system. An enterprise customer usually comes in through their IT needs, then switches the same machine on for HR, then customer service, then security. Different jobs, same system underneath.
ServiceNow owns the record, enforces the rules, and keeps the receipt. That makes it different from a simple information interface, the kind of software an agent eats the moment it can read the database directly. While basic interfaces are vulnerable to being bypassed by AI, ServiceNow’s role in securing and routing complex workflows creates a critical layer of oversight and trust that simple automation cannot easily replace.
They already have a great system with tons of customers… but like all SaaS companies, their fate will be decided by how well they can incorporate AI into their product and fend off competitors.
RLUSD x SQUID ANNOUNCEMENT
Since 2023, Squid has been seamlessly connecting it all - moving $6B+ across 100+ chains and powering 1,000+ apps you probably already use.
And today, Squid also powers instant cross-chain access for RLUSD across XRPL, Ethereum, Base, Optimism and beyond.
Here's what Squid lets you do:
- Buy, swap, bridge & send 20k+ tokens across 100+ chains
- Tap into 130+ liquidity sources in a single transaction
- Skip the chain-hopping, wallet-juggling headaches
Across chains, assets, and apps - wherever you go in crypto, Squid already exists.
Stop bridging. Start Squidding.
HOW DO THEY MAKE MONEY?
ServiceNow was founded in 2004 (as Glidesoft), went public in 2012, and is run by CEO Bill McDermott out of Santa Clara. It is not small and not a startup: revenue is north of $15B a year (and growing).
Worth remembering, because the market is filing companies like this under "walking dead," and obviously their books say otherwise.
Here's how it actually earns. ServiceNow sells subscriptions, and almost nothing else. Customers don't buy the software and install it; they pay every year for access to the Now Platform and whatever suites they've switched on.
In Q1 2026, 97% of its $3.77B in revenue was subscription revenue… and that keeps growing at a 20% clip every quarter.
Contracts run for years, get billed annually, and can't be canceled mid-term. That's about as predictable as revenue gets.
But predictable isn't a moat. The moat is what happens after a customer signs: they expand, and they don't leave.
In 2025, roughly 91% of new business came from customers already running five or more products. That repeat-buying is the moat.
Expand first. ServiceNow lands a company in one place (usually IT), proves it works, then sells the same machine into other departments like HR, customer service, security and more.
The big accounts keep getting bigger: more than 600 customers now pay over $5M a year, up from about 420 two years earlier.
That's the entire growth engine; it's called land-and-expand.
And customers don’t even really leave. ServiceNow renews 98% of its business every year.
Once a large company runs its operations on this thing, walking away means tearing out the operational nervous system: the integrations, the permissions, the audit trails, the procurement, years of encoded process.
Pulling it out is open-heart surgery. That's why "own the records and you survive AI" fits ServiceNow better than almost any name on the board.
And here's the piece that matters most for my thesis.
The bear case says AI kills per-seat pricing: one human plus AI does the work of three, so the vendor sells fewer seats.
So ServiceNow's answer is to stop selling seats.
It's shifting toward charging for usage and outcomes instead of logins, with roughly half of new business already on non-seat pricing and its newer AI bundles priced to lift the average deal 20-30%.
The company is trying to reprice itself from "humans × subscription" to "work done × value," which is exactly what’s needed to survive.
Whether that new pricing captures as many dollars as the old one is the open question. And it runs straight into the AI product.
CONTROL TOWER OR CASUALTY?
Ok, so here’s the big SaaSpocalypse-driven question: when agents can do the work themselves, does ServiceNow become the thing they plug into, or the thing they replace?
ServiceNow's answer is the most ambitious version of "plug into" you'll hear from any incumbent. It now calls itself the AI Control Tower, the "AI of the AIs," an air-traffic-control layer for every AI tool running inside a company.
The pitch: don't build your own model to fight the big LLMs. Welcome all of them in, and be the place where a business points them, governs them, and watches what they do. At its 2026 investor day, the company leaned all the way into this, name-checking partnerships with OpenAI, Anthropic, Google, NVIDIA, Microsoft, and Amazon, and its CFO called ServiceNow "the AI operating system for the enterprise."
Marketing? Partly. But the pitch is credible, for the same reason as sections one and two: an agent that wants to actually do something inside a big company needs the records, the context, the permissions, and the audit trail to do it. A model can draft the change. It can't be trusted to push a menu update to 4,000 stores unless something underneath knows who's allowed, what it touches, and how to undo it.
ServiceNow is betting that raw intelligence stays cheap and the safe place to use it stays scarce, and that it owns the safe place. That's the bull answer to the scariest line in the bear case (the model makers climbing the stack). Let them climb. They still have to land somewhere.
The AI product is Now Assist, and it's where the whole pitch gets graded. It crossed $600M in contracted annual value in 2025, reached roughly $750M early in 2026, and is now at $1B+. That's fast. It's also still small: even at $1B, you're talking a few percent of a $15B+ business.
Management has told investors to expect $1.5B from Now Assist in 2026, a target it raised from $1B.
So the honest read: The control-tower story is the right story, and the early numbers are real, not vapor.
Gartner still expects 40% of agentic AI projects to be canceled by 2027. ServiceNow counters with a 100-day guarantee of clear, measurable ROI, tied to results rather than AI promises. That guarantee is the strongest piece of the bull case.
ServiceNow is also testing Autonomous Workers, starting with an L1 Service Desk Specialist: six live pilots and 50-plus early adopters in flight.
In ServiceNow's own model, a customer cuts service-desk costs by 65% while the revenue ServiceNow earns on that account rises 5x.
That's labor replacement priced for both sides. If it scales, it's the clearest version of selling work done instead of seats.
THE BEAR CASE
These are the risks specific to ServiceNow, the ones that could make this exact name the wrong call, even if the broad thesis is right.
The flagship product is expensive to run.
This is the big one. Traditional software is beautiful because it scales for almost nothing: write the code once, sell it a million times. AI doesn't work that way. Every Now Assist query burns compute, so the better the AI product does, the more it costs to deliver. ServiceNow's subscription gross margin slipped about two points in 2025, to roughly 80%, as those compute costs landed on the income statement, and its multibillion-dollar Armis acquisition adds another ~25 basis points of gross-margin drag in 2026. Management swears it all normalizes in 2027 on internal AI efficiencies. Maybe. But you'd be underwriting a margin recovery that hasn't happened yet, in a model that is structurally more expensive than the one it's replacing.
Success could shrink the toll.
Few people think ServiceNow disappears. The deeper fear behind the near-50% drawdown is that AI commoditizes the workflows underneath it, leaving the company a gatekeeper collecting a smaller toll on each process. There's an early tell: charging extra for AI (the old "Pro Plus" tier, usage-based fees) reportedly nudged some IT leaders to ask whether they could just build the agent themselves. April's move to fold AI into the base price reads partly as a fix for that friction, but it cuts both ways, because giving the AI away in the base makes it harder to charge a premium for the very thing that's supposed to drive the next leg of growth.
Challengers can chip the expansion engine.
Section two's whole bull case was land-and-expand: customers buying the fifth and sixth products. But AI-native startups are aiming straight at those individual modules, a sharper service desk here, a smarter HR agent there. No challenger has to rip out the entire nervous system to do damage. It just has to win enough of the next-module decisions to slow the expansion the growth (and the valuation) depends on.
The pricing pivot could catch it mid-jump.
ServiceNow is deliberately moving off seats and onto consumption and outcomes. Smart long-term, risky in transition. If the old seat-based revenue softens faster than the new work-based revenue ramps, you get an air pocket, the same trap I flagged for Salesforce. And it's harder to spot here, because once AI is baked into every product, the clean "AI revenue" line the thesis leans on goes blurry. I'm partly flying on a number whose definition the company controls.
In fairness, the bear case has real holes. Nobody is actually leaving (that 98% renewal again). And free cash flow margin expanded last year, from 31% to over 34%, even as gross margin dipped, which means the cash machine absorbed the AI costs and still grew. None of these risks is a cliff. They're slow leaks. But slow leaks are exactly what a near-50% repricing should make you check for.
The honest summary: the machine keeps working while the economics quietly get worse. A slow leak, not a cliff.
Which sets up the one question that actually decides this: after a 50% haircut, am I being paid enough to take that risk?
CURRENT PRICE
The thematic piece made the macro version of this point: after the SaaSpocalypse, SaaS trades around 25x forward earnings, down from over 80x at the 2021 peak.
ServiceNow is the purest example of what that repricing did to a genuinely great name.
For a decade, NOW was the stock you wanted and could never quite stomach paying for. It compounded north of 20% a year, rarely lost a customer, and the market made you pay through the nose for it: a forward earnings multiple up in the 40s and 50s, sometimes higher. The quality was never the question. The price was.
That's what changed. The stock is down roughly 50% over the past year (it ran a 5-for-1 split in Dec. 2025, which is why it now trades near $100 rather than near $1,000).
At about $96 today, it changes hands for somewhere in the mid-to-high 20s times forward earnings.
Read that again: a business growing 20%+, renewing 98% of its revenue, and rebuilding itself around AI now trades at only a slim premium to the broad software group, next to the enormous premium it used to command.
Its PEG ratio, the multiple measured against its growth, sits right around 1. For ServiceNow, that is historically cheap.
This is my entire thesis in a single name. When the market sells everything for the same reason, some companies that don’t fit that narrative go on sale by accident. NOW got dumped as if AI were about to hollow it out, while it kept printing 22% growth and rebuilding its product around the very shift the bears fear. Prices say structural decline. The bookings say record backlog.
Now, the honest caveat, because mid-20s forward earnings are reasonable, not a fire sale. On a trailing basis, the multiple still looks rich (up in the 60s), because earnings are still small against the price while the company spends heavily on AI and acquisitions.
So the valuation only works if the growth holds and the margins normalize the way management promises for 2027.
At their last investor day, the CEO opened with "promises made, promises delivered." The track record backs it up. That's a real part of why I trust this team to hit the 2027 margin recovery, the whole valuation rests on.
I'm buying a great company at a fair price, for the first time in years, on the condition that it keeps executing.
Which is the whole point. At around $96, where this position went on, I stopped paying for perfection. I'm paying for continued execution. And continued execution is something you can actually measure.
MY POSITION AND EXIT
Every thesis needs an exit written before you need it. So here are mine, narrowed from the thematic piece to this one name. Clear the bull bars and the bet is working. Hit a bear tripwire and I don't argue with it, I act.
My bull bars are the continued growth of the backlog and the delivery of the $1.5B Now Assist target, while my bear tripwire is if a major enterprise adopts a raw model stack for its operational nervous system and doesn't return to ServiceNow.
Where I land:
ServiceNow is the cleanest version of my thesis: the company that most clearly owns the process, runs the workflow, and gets paid year after year to keep doing it. The old machine is compounding (22% growth, 98% renewal). The AI machine is real but unproven (Now Assist at $1B+, on the road to a $1.5B test). And for the first time in a decade, I can buy it at a fair price instead of a fantasy one.
The position went on at around $106, roughly where it trades today.
The plan from here is the boring kind: add if Now Assist stays on track for $1.5B and the backlog keeps growing north of 15%, and cut if either breaks.
And I don’t think it will. But to keep up with my moves here, you’ll need to get into Milk Road PRO, which is just a buck to test out. You can even get in to ask me about NOW, NU, or any of the other 11 positions I currently hold.
BITE-SIZED COOKIES FOR THE ROAD 🍪
Is this one of crypto's most underrated real-adoption plays? XDC sits right where RWAs, cross-border payments, and institutional money are all colliding.*
Kevin Warsh's logic behind killing forward guidance: "Financial market prices are probably the most important source of information to guide central bankers."
Kyle Reidhead: "Buy compute, sell Bitcoin…. The capital is just going into the AI CapEx right now and it's not going into Bitcoin."
The Fed's new playbook: Strategic ambiguity is back.
DeFi Saver turns Aave v4 into a trading bot for your loans: Stop losses, take profits, and one-click loan migration - we broke down the pros, cons, and fees so you don't have to.
*this is sponsored content.
Stop bridging. Start Squidding.
MILKY MEMES 🤣
ROADIE REVIEW OF THE DAY 🥛
VITALIK PIC OF THE DAY
