For years, the software industry operated on a simple equation: more customers, more seats, more revenue.
Then AI arrived and threatened to blow the whole thing up.
The new generation of AI-native companies promised something different. Their economics weren’t built around selling another licence to another employee.
They were built around intelligence being consumed – agents acting, workflows running, decisions being made. In that world, the old SaaS kings suddenly looked less like category leaders and more like incumbents waiting for disruption.
Software survival
And yet, this summer, two of the greatest survivors of the SaaS golden age delivered two very different messages to Wall Street.
Salesforce had its best day on the stock market since 2020. Workday found itself at the centre of speculation over a potential $50 billion take-private deal.
One is fighting to reinvent itself in public. The other is being asked whether it would be better off reinventing itself in private.
The bigger question is whether either strategy can keep the SaaS giants relevant as the AI arms race accelerates.
Because perhaps the most valuable asset they own isn’t the software at all. It’s the data well underneath it.
Salesforce: from SaaS king to agentic enterprise
For those pedalling the “SaaS is dead” story, Salesforce’s comeback story is becoming increasingly difficult to ignore.
Its fiscal Q2 results beat expectations. Its shares jumped 22.6% in a single session to $252 — the company’s strongest market-day performance in years. Agentforce, its push into AI agents, was reportedly growing at more than 240%. An expanded partnership with Anthropic added another signal that Salesforce isn’t content to sit on the sidelines or be roadkill while the AI platform wars unfold.
Just days before the results, JPMorgan had upgraded Salesforce to Overweight with a $250 target. Then Salesforce beat it in a day.
Suddenly, the “SaaSpocalypse” narrative looked rather premature.
But the interesting part isn’t the earnings performance. It’s the attempted escape from the economics that made Salesforce so successful in the first place.
Traditional SaaS is fundamentally tied to seats. Hire more people, buy more licences. Add fewer people, sell fewer licences. Growth is therefore tethered to the size of the customer’s workforce.
AI agents change that equation.
An agent doesn’t need a desk, an email address or an annual licence in quite the same way a human employee does. It can work continuously, across thousands of transactions and workflows.
Salesforce is therefore pushing towards consumption pricing, with Agentic Work Units billed against Flex Credits.
That sounds like pricing innovation. It is something much more consequential: an attempt to sever the historic relationship between software revenue and customer headcount.
So, can Salesforce can pull it off? And can consumption revenue grow quickly enough to compensate for a slowing seat-based business?
Workday’s $50 billion question
Then there is Workday, founded in March 2005 by software visionaries Dave Duffield and Aneel Bhusri after a breakfast meeting in Truckee, California, following Oracle’s hostile takeover of their former company, PeopleSoft
The summer began with Reuters reporting that Silver Lake was exploring a take-private transaction potentially valuing the company at around $50 billion – a deal that, if completed at that scale, would rank among the largest software buyouts in history.
Workday’s shares immediately jumped, its strongest performance in a decade.
And then came the silence. On its Q2 call, Workday did not provide the kind of confirmation – or denial – that might have settled the speculation and, at the time of writing the situation appears unresolved.
Practical philosophy
But the philosophical question doesn’t require a deal to be signed. Why would one of the great SaaS companies of its generation even contemplate going private?
Perhaps because the public markets are increasingly impatient with the economics of yesterday’s software, even as concerns about the impact of AI grow.
AI-native platforms can command extraordinary revenue multiples – 15x, 20x, sometimes 25x or more.
Meanwhile, mature seat-based SaaS businesses – Salesforce and Workday’s original model – can find themselves trading at dramatically lower multiples, in the rough range of 3x to 6x revenue.
The market is effectively saying something brutal: We don’t care how much software you sell if growth remains mechanically connected to the number of people your customers employ.
That’s a problem when AI’s promise is to make companies more productive with fewer people.
The data underneath the software
But there is another way to look at this: Salesforce and Workday aren’t merely software companies.
They are data companies that happen to sell software.
Salesforce sits on some of the richest collections of commercial information in the enterprise: customer relationships, sales pipelines, interactions, pricing, service histories and transaction context.
Workday sits on another kind of gold mine: employee records, organisational structures, compensation, workforce planning and financial information.
These aren’t datasets you can simply download from the internet and feed into the latest frontier model. They are proprietary, deeply embedded and continuously refreshed.
Which brings us back to old ground that suddenly feels newly relevant.
Data may be the new oil, but context is king
Except there’s an important qualification – data is only valuable if you own the well.
The AI model may be interchangeable. The model you use today may be obsolete tomorrow. Frontier intelligence is becoming increasingly commoditised as models improve, prices fall and capabilities converge.
But the customer data flowing through an enterprise system is another matter.
Your CRM knows who your customers are. Your HR system knows who your people are. Your financial system knows where the money goes.
The company that controls those systems controls the context in which AI operates. And context may ultimately be more valuable than intelligence.
The sovereignty subplot
That is why the emerging battle over data sovereignty matters almost as much as the AI model race.
Workday’s launch of its EU Sovereign Cloud – built on AWS, with EU-resident operations and AI processing – arrives at precisely the moment Europe is becoming increasingly concerned about where its critical data lives, who can access it and which technology stack ultimately controls it.
Brussels’ growing enthusiasm for initiatives such as EuroStack reflects the same anxiety.
The argument is no longer simply about whether Europe can build an AI model competitive with Silicon Valley. It is increasingly about whether European companies can retain control of the infrastructure, applications and data on which their economies depend.
And that brings us back to Salesforce and Workday. They may not own the most powerful AI models. They don’t need to. They may own something harder to reproduce.
The data well.
Private equity: rescue mission or slow death?
But there is a darker interpretation of all this.
Perhaps private equity is becoming the natural final chapter for yesterday’s category kings. Look at the recent activity.
Thoma Bravo’s $12.3 billion acquisition of Dayforce.
Hg’s take-private of OneStream.
These aren’t obscure companies. They are category leaders.
But once a software company goes private, something changes. Management gets time. Quarterly earnings pressure disappears. Investment horizons lengthen.
A company can cut costs, restructure its product portfolio, change pricing and make a multi-year bet without explaining every move to Wall Street every 90 days.
Category denial
That can be a powerful form of freedom. But it can also look suspiciously like a retirement home.
When a category leader goes private because public investors no longer believe in its growth story, is private equity rescuing the business?
Or is it simply providing a dignified exit for a category that has run its course?
Workday now sits directly in that debate.
If the Silver Lake approach has indeed been rebuffed, independence could represent extraordinary confidence: the belief that Workday can reinvent itself without needing to escape public markets.
Or it could represent something less flattering. Category denial.
The rarest move in business
Salesforce, meanwhile, is attempting something even harder. Even heroic.
It is trying to win a category twice.
The first time, it built the modern CRM category and became its defining company. Now it wants to redefine the category itself.
CRM becomes the agentic enterprise. The software isn’t simply recording what employees do. AI agents are supposed to act on the company’s behalf – selling, servicing, analysing, automating and executing.
That is a remarkable strategic move. It amounts to saying: We won the old game. Now we’re changing the rules before somebody else can beat us at the new one. Pretty ballsy.
Workday faces a different kind of test.
It must persuade investors that being the system of record for human capital and finance remains strategically valuable when AI can increasingly perform the work those systems support.
That distinction matters. Because the future may not belong to the company with the best AI.
It may belong to the company with the best position from which AI can act.
The wild frontier
So, does proprietary data beat the frontier model? That is the real category question.
The AI arms race has produced extraordinary models, but models are becoming abundant. Enterprise-grade proprietary data is not.
There will be dozens of powerful models. But there will be far fewer systems that contain decades of trusted customer, employee and financial context.
And there is a potentially enormous difference between an AI that knows how to do something and an AI that knows your business.
That is Salesforce’s opportunity. It is Workday’s opportunity too.
It may also be the reason these companies are not relics of the SaaS era, but infrastructure for the AI era.
But owning the well isn’t enough. You have to monetise it.
Salesforce’s consumption model is an attempt to do exactly that. Workday’s sovereignty strategy is another way of defending the value of the well itself.
Both are ultimately answering the same challenge: How do you turn proprietary data into an economic advantage when intelligence itself is becoming cheaper?
Oil or troubled waters?
The SaaS kings aren’t dead. But they are no longer entitled to rule.
Salesforce is betting that it can reinvent its economics and recrown itself as the platform for the agentic enterprise.
Workday is betting – at least for now – that independence gives it the room to reinvent without surrendering control.
And private equity is waiting in the wings with a much simpler proposition: if the public markets won’t value the well, we will and we’ll take control.
That leaves us with a provocative possibility. The winners of the AI era may not be the companies building the smartest models.
They may be the companies that already own the deepest wells of proprietary data – and figure out how to let intelligent machines access them.
Because data may be the new oil. But the real moat is owning the well, knowing the value of the resource and controlling who has access to it.