In February 2026, the SaaS sector lost roughly US$285 billion in market value in a single selloff. The financial press dubbed it the SaaSpocalypse. Jefferies downgraded Workday and DocuSign citing AI disruption. Thomson Reuters fell nearly 16% in one day. The long-held belief that SaaS was recession proof took a very public hit.
But behind the selloff is a quiet revolution in how software is priced. The flat, per seat subscription model that powered SaaS for 15 years is breaking down. Customers are not adding seats the way they used to, AI has made the cost of serving each customer wildly uneven, and usage-based pricing gives vendors a less confrontational way to keep growing revenue. Software is shifting from charging for access to charging for consumption.
The bigger question is what this means for your portfolio. Not every SaaS stock is going to zero. Some will adapt and emerge stronger. A few will be acquired. Your job, and mine, is to figure out which is which.
First, some perspective. I believe the “SaaS is dead” narrative is overdone. Recent data on actual corporate software spending shows that seat-based contracts still account for somewhere between 65% to 75% of what companies spend. Pure consumption pricing is still well under 10% of total spend. The shift is real, but it is unfolding over years, not weeks.
There is also a group of SaaS companies that sit on what analysts call a data moat. Intuit has decades of proprietary tax and accounting data. Microsoft has the entire Office and Azure ecosystem. Salesforce sits on enterprise customer data that no general purpose AI can replicate. These businesses are not disappearing. They are simply going to evolve how they charge you.
With those caveats in mind, here is what is actually likely to happen, and what to look for in your holdings.
Revenue gets lumpier, and that changes everything
The cleanest impact of consumption pricing is that quarterly revenue becomes less predictable. With seats, you knew almost exactly what next quarter would look like based on the contracts you already had. With consumption, you find out at the end of the quarter, after customers actually used your product.
Snowflake already lived through this in 2023, when cost-conscious enterprises throttled their compute spending and the stock got punished hard. Earnings became less of a victory lap and more of a hold your breath event.
For investors, this means the predictability premium that drove SaaS multiples to all-time highs is over. SaaS stocks will trade more like Snowflake, Datadog, and Twilio already do, with bigger swings on earnings as usage either surprises positively or negatively. That is not necessarily bad. It just requires a different stomach.
Why the incumbents are trapped
Here is what should worry investors most about the big, familiar names. Every major SaaS incumbent can see the consumption shift coming. The problem is that they cannot move quickly, because their entire business is built on the very model that is now under threat. This is the classic innovator’s dilemma.
Think about it from Salesforce’s perspective. If it aggressively pushed every customer onto consumption pricing tomorrow, it would blow up its own predictable seat revenue, the exact thing Wall Street has rewarded for two decades. The stock would crater on the uncertainty. So incumbents are stuck moving slowly, bolting AI consumption onto the side of their seat business rather than replacing it outright.
Meanwhile, AI native startups have no legacy revenue to protect. Cursor, the AI coding tool, reportedly went from zero to around US$1 billion in annual revenue in roughly two years, priced on consumption from day one. When GitHub, owned by Microsoft, finally moved its Copilot product to token based billing in June 2026, its 4.7 million existing subscribers woke up to a completely different product, and many were furious. That is the dilemma in a single example. The startup gets to build the new model cleanly. The incumbent has to rip up a model that millions of customers already depend on.
The takeaway is uncomfortable but clear. For these incumbents, the greatest asset they own, a massive base of locked in seat contracts, is quietly becoming their greatest liability.
Getting the price right is a razor’s edge
Here is the part almost nobody talks about. Moving to consumption pricing is not a magic fix. The pricing structure itself is brutally hard to calibrate, and getting it wrong hurts in both directions.
Set the price per token or per credit too high, and customers feel nickel and dimed. They constantly run out of their allowance, they get surprise overage bills, and they start resenting the product. Cursor learned this the hard way in mid 2025, when it switched to usage-based credits; customers got unexpected charges, and the company had to issue a public apology and refunds. GitHub Copilot triggered the same revolt in 2026 when power users watched their bills jump many times over almost overnight.
Set the price too low, and you get the opposite problem. AI features carry real, variable costs. If you price consumption too cheaply, your heaviest users will burn enormous amounts of compute and tokens, and the cost to serve them eats straight into your gross margin. You end up subsidizing your most demanding customers, which is exactly the trap consumption pricing was supposed to escape.
The winners will be the companies that find the narrow band in between. High enough to protect margin, low enough that customers do not feel constrained, and transparent enough that nobody opens an invoice and panics. This is why you are seeing constant pricing revisions across the industry. Salesforce moved Agentforce from 2 dollars per conversation to a credit-based action model within a single year. Expect a lot more of this trial and error before the dust settles, and treat a company’s pricing discipline as a real part of your investment thesis, not an afterthought.
What should you do as an investor?
1. Stop using ARR as your primary lens. For 15 years, annual recurring revenue was the holy metric of SaaS investing, because it sounded predictable, as if the revenue would simply repeat next year like clockwork. In a consumption world, that is closer to fiction. If a customer cuts usage 30% next quarter, that recurring revenue was never truly recurring.
Two better metrics deserve your attention:
- Net revenue retention measures how much existing customers spend year over year, including both expansion and contraction, so a company above 120% is growing from its existing base alone, while one below 100% is quietly leaking.
- Gross revenue retention strips out expansion and shows how much you keep before any upsell, which reveals the true stickiness, or the floor, of the business.
Watch those two far more closely than headline ARR.
2. Reassess your seat heavy holdings. If you own SaaS companies whose entire model charges per user, per seat, per host, or per document, ask how easily an AI agent can replace the human doing that work. The answer largely determines the next five years of your returns from those names.
Judge how well each company prices. A company that moves to consumption but botches the calibration will either bleed margin or bleed customers. Watch for clean, transparent, predictable pricing and usage dashboards. Repeated chaotic pricing changes and customer revolts are a warning sign, not a rounding error.
3. Watch the transition winners. Big incumbents with deep data moats, like Microsoft, Salesforce, ServiceNow, and Intuit, may go through ugly transitions, but they have the assets to emerge with stronger pricing power than before. The recent selloffs in some of these names could be opportunities if you have the patience and the stomach.
The fifth perspective
The flat per seat SaaS model is being dismantled in real time. Some of the SaaS stocks you own today will look very different five years from now. Some will be much stronger. A few will be gone. The market has already started pricing this in, which is exactly why February 2026 happened.
The good news is that this is not a sudden collapse. It is a slow rotation. Investors who watch the metrics that actually matter, who understand which categories are structurally tailwind versus headwind, and who reward companies that price with discipline, can position themselves well ahead of the crowd. SaaS as a category is not dying. It is being rebuilt with a new pricing engine underneath.
Stay patient. Stay selective. And stop staring at ARR.
