Are SaaS stocks AI roadkill or AI winners? A look at ServiceNow (NOW), Adobe (ADBE) and the SaaSpocalypse

Software was supposed to be the safest seat in tech. Instead it has been 2026's ugliest trade. Roughly $2 trillion in software market value has evaporated over the past year, and for the first time in the cloud era, the software sector now trades at a lower forward earnings multiple than the S&P 500. The question tearing Wall Street in half is simple to ask and hard to answer: in the AI game, are SaaS companies roadkill or essential infrastructure? Here is what the big names actually do, how far they have fallen, and which side of that question the evidence lands on.

What broke: the "SaaSpocalypse"

SaaS stands for software-as-a-service, the subscription model where you rent software by the user, by the year, instead of buying it once. It built some of the most reliable businesses in the market. Then AI agents showed up.

The selloff has a nickname now: the SaaSpocalypse. It kicked off in late January and February, when Anthropic rolled out Claude Cowork, a desktop AI agent that automates the kind of multi-step office work, drafting, compliance, document handling, that people used to do inside paid software. A single Claude legal plug-in in early February helped wipe about $285 billion off tech stocks in 24 to 48 hours. A second leg hit on April 9 when a new Anthropic agent dropped, taking Cloudflare down 12 percent, Snowflake down 9 percent, and ServiceNow down 7 percent in a day. The research firm Forrester published a note literally titled "SaaS as we know it is dead." The software-heavy IGV ETF fell more than 24 percent in the first quarter, its worst quarter since 2008.

The fear boils down to one sentence: if the software charges by the seat, and an AI agent does the work of ten people, the customer needs one seat instead of ten. Goldman Sachs strategist Ben Snider compared software's path to newspapers. Orlando Bravo, who runs the software-focused buyout firm Thoma Bravo, said the valuation cuts on AI-exposed software are "very warranted." When the biggest software buyer on earth says the discount is deserved, people listen.

Lately, though, the tape has been turning. Software just had its best month since 2001 in May as the SaaSpocalypse panic cooled, and the group is green again today: the software ETF, IGV, is up about 3.4 percent, with ServiceNow up more than 8 percent and Salesforce, Snowflake, and Adobe all higher. Before reading too much into it, these names are still down 30 to 50 percent on the year. This looks like a leg up inside a brutal stretch, not a confirmed bottom.

ServiceNow (NOW): the textbook round trip

ServiceNow (NOW) runs the "Now Platform," enterprise software that automates workflows for big organizations: IT service management, HR cases, customer service, security operations. If a Fortune 500 company files an IT ticket, odds are it runs through ServiceNow. It is also the cleanest example of the round trip you saw in the price.

ServiceNow did a 5-for-1 stock split in December 2025, so these are all split-adjusted. The stock started the year around $150. The February SaaSpocalypse knocked it down toward $100. It kept sliding to a 52-week low of $81.24 on April 10. Then earnings on April 22 made it worse: despite beating on basically every line, revenue of $3.8 billion was up 22 percent, the stock fell about 15 percent the next day, its worst single session ever. The problem was not the quarter, it was the read-through: bookings growth (cRPO) decelerated, the $7.75 billion Armis security acquisition dinged margins, and some big Middle East deals slipped on regional conflict. From there it staged a real recovery, climbing back to roughly $135 by early June on a wave of AI product launches, then rolled over again into the low $90s by late June. As of June 26 it is around $97, up about 8.5 percent on the day, but still down more than half from its 52-week high of $211. Dropped big, went up, came back down. The whole AI-software argument in one chart.

So is ServiceNow needed in the AI game? Management's answer is to make ServiceNow the place AI agents do their work. Its Now Assist AI line hit about $750 million in annual contract value, and the company raised its 2026 AI target to $1.5 billion, on the way to a $30 billion total revenue goal by 2030. At its Knowledge 2026 conference in May it launched Action Fabric, which opens its workflow engine to outside AI agents, including Anthropic's Claude, and an AI Control Tower to govern all the AI running across a company, not just its own (see its May 5 announcement). The bull case is that ServiceNow becomes the control layer for enterprise AI. The bear case, and the reason UBS downgraded it in April, is that its seat-based revenue is exactly what agents erode, and that margins compress as the mix shifts.

Adobe (ADBE): the one the AI bears love to hate

Adobe (ADBE) makes Creative Cloud (Photoshop, Illustrator, Premiere), Document Cloud (Acrobat and PDFs), and Experience Cloud (enterprise marketing tools). It is the company the "AI kills software" thesis points to first, because generative AI hits its core product directly. If a free tool can make a professional image, why pay for Photoshop?

The stock reflects that fear better than any fundamental does. Adobe is down about 43 percent year to date, trades around $201, roughly 49 percent below its 52-week high of $392 and its lowest level in eight years, with a market cap near $80 billion and a forward earnings multiple that has fallen into the single digits. The bears point at Anthropic's Claude Design, launched in April, plus OpenAI's Sora and DALL-E and Midjourney as free or cheap substitutes chipping at Adobe's moat. It did not help that the CFO left for Marvell in mid-June.

The numbers, though, are not breaking. Adobe's Q2 results on June 11 (its own newsroom has the release) were a record: revenue of $6.62 billion, up 13 percent, total recurring revenue of $27.1 billion, and AI-first revenue that tripled year over year to more than $500 million. The company even raised its guidance. The stock still fell about 6 percent, because Adobe said it would lean into a free, freemium funnel and hold off on planned price increases, which spooked investors about near-term growth. On the product side, Adobe is not sitting still: it launched a Firefly Creative Agent in April and signed an NVIDIA partnership in March to build its next-gen AI models. The question is whether "commercially safe" Firefly, baked into tools pros already use, is enough to defend the franchise against an army of free generators.

The rest of the field: winners, losers, and the in-between

The selloff has not hit everyone the same way, and the split is telling:

• Microsoft (MSFT): the biggest software company of all, and the clearest AI winner. It is down about 18 percent this year, far less than the rest, trading near $369, roughly a third below its high, with a market cap around $2.7 trillion. Microsoft is not the one being disrupted, it is selling the disruption: its Microsoft 365 Copilot is a $30-per-user add-on now on more than 20 million paid seats, its AI business is running at a $37 billion annual rate (up 123 percent), and Azure grew about 40 percent last quarter. The worry here is not seat loss, it is the roughly $190 billion it plans to spend on AI build-out this year and whether the payoff justifies it.
• Salesforce (CRM): the CRM giant, down about 30 percent this year. It is leaning on Agentforce, its AI-agent platform, which has grown to roughly $1.2 billion in annual recurring revenue, and CEO Marc Benioff has spent months publicly defending the SaaS model. It also announced a $25 billion buyback. Still very much in the crosshairs because its revenue is seat-based.
• Workday (WDAY): HR and finance software, down about 33 percent, and the name the market treats as most exposed, because autonomous HR and finance agents threaten user licenses directly. Jefferies downgraded it. Workday is fighting back with its own AI push (it bought the AI startup Sana and says emerging AI products now top $400 million in ARR), but the market has mostly shrugged.
• Snowflake (SNOW): the data-cloud company, and the clearest winner of the bunch. It is actually up on the year. The reason is its pricing: Snowflake charges by consumption, not by the seat, so more AI activity means more usage, not fewer licenses. AI agents need a data layer, and Snowflake sells the data layer.
• Intuit (INTU): the maker of TurboTax and QuickBooks. It partnered with Anthropic to build AI agents into its products, says tens of millions of transactions a week already run through its AI, and is moving toward usage-based pricing. A consumer-finance angle on the same shift.

Notice the pattern. The company that owns the AI distribution (Microsoft) and the consumption-priced infrastructure names (Snowflake) held up best. The classic per-seat application vendors (Workday, Salesforce, and ServiceNow) got hit hardest. The market is not selling "software," it is selling the seat-license business model.

So are they needed in the AI game?

This is where the smart money openly disagrees. On the bull side, Nvidia's Jensen Huang flatly said "the markets got it wrong," arguing agents will use software, not replace it. Bank of America's Vivek Arya called the selloff "indiscriminate" and "overblown." Morgan Stanley's Katy Huberty framed it as a "sentiment-driven" dislocation, not a fundamental one. Wedbush's Dan Ives called it a "generational opportunity." The consulting firm Bain made the most concrete bull argument: incumbents own the systems of record, the customer data, the trust, and the distribution, which is a real "right to win" even in an agent world.

The honest read is that AI is not killing software so much as changing who gets paid and how. The shift is from charging for seats to charging for outcomes, and that transition is going to mint winners and gut losers. The companies that own the data and the workflows, and that turn AI into something customers pay more for, come out fine or better. The ones that mostly rent out seats an AI no longer needs are in real trouble. Microsoft, with Copilot riding on top of 450 million Office users, has largely proven it belongs in the first camp. ServiceNow, Adobe, Salesforce, and the rest are still trying to prove the same. Whether today's beaten-down prices are a bottom or a value trap comes down to a question no chart can answer yet: when the dust settles, whose pricing power is still standing? We dug into the broader version of this fight in did the AI bubble just pop, and at one famous bear's software bets in Michael Burry's Microsoft and Palantir trades.

Nothing here is investment advice. Prices and figures are as of June 26, 2026 and move constantly. These are volatile stocks in the middle of a genuine debate, with smart investors on both sides. Do your own research.

Cover photo: source code by Martin Vorel / Wikimedia Commons, CC BY-SA 4.0, cropped.