Two things happened within 24 hours this week. Together they say more about where B2B is going than most of the “SaaS is dead” takes from the past 18 months.
On Thursday, shares of ServiceNow fell 26.9%, while the S&P 500 slipped 0.3%, after its October 28 Q3 report. On Friday morning, Bret Taylor and Clay Bavor posted Sierra’s Q3. Sierra added over $150M of ARR in the quarter and passed $400M in total ARR. That one quarter brought in more ARR than all of last year.
Some will read this as “SaaS is dead, agents won.” ServiceNow is still a great business, so I don’t think that’s right. SaaS changed into something else, and the same B2B budget is now moving from the old version to the new one.
ServiceNow Grew 20% at $16B and Still Lost a Quarter of Its Value
ServiceNow’s quarter was good:
- Subscription revenues of $3,988 million in Q3 2026, representing 21% year-over-year growth.
- 19 transactions over $5 million in net new ACV, over 135% year-over-year growth, and 680 customers with more than $5 million in ACV at quarter end.
- Management raised its year-end AI target for the second time this year, from $1 billion at the start of the year to $1.7 billion.
- Full-year free cash flow margin guidance went to 36% from 35%.
Most B2B companies would take 20% growth at this scale with 36% free cash flow margins in a heartbeat.
Even so, ServiceNow now trades 41.3% below its highest close of the past year. By at least one account, Thursday was its worst single day since its 2012 IPO.
The headline miss was small. Adjusted EPS was $1.00, below the $1.02 analysts expected, while revenue reached $4.11 billion, slightly ahead of the $4.09 billion forecast. A two-cent miss doesn’t erase $25B+ of market cap on its own. Three other things did:
- Backlog growth is slowing. Management expects growth in its current backlog to slow to 18.5%, down from 21% in the third quarter. CFO Gina Mastantuono also said about half of Q3’s cRPO beat came from early renewals pulled forward from Q4, triggered by a July 1 pricing change.
- Margins are compressing. Reported operating margin fell to 9.4% in fiscal Q3 2026 from 16.8% in fiscal Q3 2025. Gross margin dropped to 70.5% from 77.3% over the same year. Part of that comes from acquisitions like Armis, and part from the cost of serving AI.
- The AI payoff keeps slipping to next year. Citi’s Tyler Radke says investors may need to wait until next year to see a clearer inflection point.
The market’s verdict: ServiceNow is a 20% grower that will stay a 20% grower, and AI is costing it margin faster than it adds growth.
Sierra Doubled From $200M to $400M ARR in Five Months
Sierra’s numbers from its Q3 post:
- $100M to $200M ARR took six months.
- $200M to $400M ARR took five months.
- $150M+ added in Q3 alone, more than all of last year.
- Almost 50% of the Fortune 50 are customers.
- A US retailer with $40B+ in revenue went from first meeting to a live agent in 27 days.
- The company launched about two and a half years ago.
At about 20.5% growth, ServiceNow takes roughly 3.7 years to double. Sierra did it in five months.
Net new dollars make the gap clearer than percentages. ServiceNow’s quarterly subscription revenue rose about $689M year over year, from $3,299M to $3,988M. That works out to about $2.76B of new annualized run-rate over twelve months, or roughly $690M a quarter. Sierra added $150M in Q3, about 22% of ServiceNow’s quarterly net new recurring revenue, at 2.5 years old.
On AI alone, ServiceNow’s target goes from $1B to $1.7B across all of 2026. That’s about $700M of incremental AI ACV for the full year, from a base of 680 customers each paying over $5M. Sierra added $150M in a single quarter, and had zero customers 30 months ago.
It’s the Same B2B Budget: Software Grows 15% in 2026 While Agentic Spend Grows 141%
Sierra isn’t being paid out of new money. It’s the same B2B budget enterprises already spend every year on customer service, CX, IT and operations. The same CIOs and CX leaders sign the same kinds of checks. The vendors getting those checks are changing.
The Gartner and IDC data shows it:
- The total software budget grows at a normal pace. Gartner’s April forecast puts software spending at $1,443,621 million in 2026, 15.1% growth.
- The agentic slice of that budget more than doubles. Agentic AI spending grows 141% in 2026 to $201.9 billion.
- Agents pass the last generation of AI next year. In Gartner’s forecast, agentic AI spending overtakes chatbot/assistant spending by 2027. Chatbots peak at $264.7B then decline. Agentic AI grows at 119% CAGR to $752.7B by 2029.
- Then the old software shrinks. By 2028, software with agentic capabilities crosses 50% of total application software spend. Non-agentic software spending starts declining in 2027.
- IDC lands in the same place. It forecasts agentic AI at over 26% of worldwide IT spending, and $1.3 trillion in 2029.
Gartner’s software and AI forecasts are separate models, so treat this as a rough estimate. If the whole software budget grows about 15% and the agentic part grows 141%, everything else grows about 6% this year. Per Gartner, that part starts declining in 2027. The pie isn’t getting much bigger. It’s being re-split.
Gartner’s agentic number includes agents built into existing apps, so ServiceNow’s $1.7B AI target sits in that bucket too. Incumbents can win some of this. What matters is share. In the same quarter ServiceNow guided backlog growth down from 21% to 18.5%, Sierra added $150M of ARR.
Sierra Charges for a Loan Refinanced. ServiceNow Charges for a Seat
Pricing is how the budget moves, and both companies were explicit about theirs this week.
Sierra prices on outcomes. Its post says customers pay for outcomes, not tokens. The outcome might be a loan refinanced, an insurance claim paid, a sale closed or a customer retained.
ServiceNow took the other side on its call. Bill McDermott said customers like the predictability of seat-based pricing for budgeting, with consumption layered on top. Mastantuono said customers are “very, very happy with the hybrid model” and aren’t asking for outcome-based pricing. ServiceNow has moved some: 50% of net new business now sourced from non-seat-based pricing, including tokens and connectors. That’s still software pricing with a usage meter attached.
The two models:
- Old SaaS sells a tool to the people who do the work. It’s priced per person, and every efficiency gain you deliver to the customer eventually shrinks your seat count. ServiceNow is seeing this inside its own company. Management said total headcount has slightly declined and internal AI efficiency gains top $1B a year. Its customers are running the same numbers.
- Agent-native SaaS sells the work itself. It’s priced per outcome, so the buyer pays for results instead of logins. When a CX leader decides where this year’s budget goes, the vendor that can show claims paid and customers retained gets the line item.
This is why I no longer invest in agentic B2B companies that can’t show deal sizes at least 2x what pre-AI competitors get. With a fixed budget, the winners have to take a much bigger share of it. An agent product landing deals the same size as the incumbent’s is splitting the budget, not taking it. Sierra adding $150M in a quarter shows what taking it looks like.
Sierra Is Still SaaS: Recurring, Multi-Tenant, Sold to the Enterprise
Sierra didn’t kill SaaS. Sierra is SaaS: recurring revenue, multi-tenant software sold to large enterprises, with a product suite that looks like any mature platform (Agent Studio, Ghostwriter, Insights, Explorer, Channels and more). It’s also building an ecosystem with Poppy, an open protocol for personal agents, with design partners including Meta, Shopify, Stripe and Walmart.
What changed is what the software does:
- It runs the conversation instead of helping a human run it.
- Its newer agents handle interactions that run for weeks or months and drive revenue, beyond one-off support tickets.
- It builds context with every interaction. Sierra calls its customer relationships its competitive advantage.
ServiceNow can do much of this. McDermott calls ServiceNow “the execution layer,” and its AI numbers are real: AI net new ACV up 65%, AI deals over $1M up 4.5x, and agentic consumption up 13x year to date. But ServiceNow has to protect a $16B seat-based base while it builds. Sierra has no base to protect.
Five Things Founders Should Take From This Week
- Plan as if no new budget is coming. Your agent product competes for dollars your customer already spends. Show the buyer which existing line item you replace and by how much.
- Price by outcome if you can measure the outcome. Sierra ties price to a closed sale or a paid claim, and that’s how it pulls budget away from seat-based incumbents. If your AI product is priced per seat, every efficiency win for the customer sets up a future downsell for you.
- Track net new dollars, not just growth rate. Sierra adding about 22% of ServiceNow’s quarterly net new run-rate tells you more than any percentage. If your AI product’s net new ARR isn’t rising every quarter, you’re shipping a feature without taking share.
- Shorten time to live. Sierra took a $40B retailer from first meeting to a live agent in 27 days. If deploying your product takes a services team and two quarters, the budget goes to someone faster before price comes up.
- Watch gross margin as AI ships. ServiceNow’s gross margin fell from 77.3% to 70.5% in a year. Investors will accept margin compression when growth accelerates, and punish it when growth stays flat.
ServiceNow at 20% and Sierra at 2x in Five Months Are Drawing From the Same Budget
ServiceNow will be fine. It’s a $16B subscription business with 36% free cash flow margins and $29.7B in total RPO.
But the B2B budget that built ServiceNow is being re-split. Software that helps people do work, priced per person, now gets a multiple that assumes 20% growth for good. Outcome-based agents that do the work, priced per result, took Sierra to $400M in ARR in 30 months, with the last doubling taking five.
SaaS isn’t dead. SaaS before this is mostly dead.
