Schedule a talk with one of our advisors to learn more about Summitry and how we can help you get a foothold on your financial life. For career opportunities please visit careers at Summitry.
For two decades, enterprise software was built around a simple premise: hire a person, buy them a seat. Revenue scaled with headcount. That model appears to be evolving in a way that is more structural than anything the industry has seen since the move to cloud computing.
The implications for how software is priced, sold, and valued are significant. And, for Bay Area professionals whose compensation, portfolios, and careers are tied to the technology sector, understanding what is actually changing versus what is noise matters. In this article, we walk through how the shift is unfolding, what it may mean for a couple of the companies most exposed, and how we’re thinking about it as investors.
From Toolbox to General Contractor
For twenty years, the economics of enterprise software were straightforward. Companies paid for software licenses based on how many people used them. There was one seat for each employee. If a business grew from 500 to 1,000 employees, their software costs roughly doubled. Revenue for SaaS companies scaled predictably with their customers’ headcount, and the model worked well for everyone.
AI is beginning to challenge that relationship. Agents can now perform discrete, repeatable work like processing invoices, triaging support tickets, and writing and reviewing code without a software license seat. A company can increasingly get more done without adding headcount, which means the seat-based pricing model that underpinned two decades of SaaS growth is under pressure.
The stakes for software vendors depend entirely on which side of this shift they land on. Think of it this way: traditional SaaS was like a toolbox. The vendor provided the tools, but the customer still had to supply the labor. AI-native software is more like hiring a general contractor. You pay for the finished outcome, and the vendor brings both the tools and the work. That represents a different value exchange, which may support different pricing models.
For companies that can prove their AI actually does the work and delivers measurable results, this may merit an expansion of what they can charge. For those that cannot make that case, customers will simply need fewer seats as they need fewer people, and revenue will follow headcount lower.
We think markets appear to be repricing this uncertainty, and software has been among the weakest performing sectors in recent months. But the picture is more nuanced than a broad sector decline suggests. Job postings for software developers, tracked by Indeed, are increasing, not decreasing. If software were being hollowed out, the companies deploying it should know first. Many of them are hiring. The transition is real, though the narrative around it may be getting ahead of the evidence.
How Summitry Evaluates Our Holdings in The AI Era
Our investment framework has not changed in response to AI. We continue to focus on economic moats, durable earnings, capable management, and valuations that offer a margin of safety. What has changed is the set of questions we apply within that framework: does AI broaden or erode a company’s competitive position? Does it expand their addressable market or cannibalize their existing model? And, is management executing with clarity in an environment where the facts are shifting quickly?
We think there are two prominent tech companies illustrate how differently the same macro shift can play out at the company level. Take Microsoft and Salesforce, for example. Please note that these are illustrative examples, not recommendations.
Our research shows Microsoft appears well positioned to benefit regardless of which specific AI application wins. Its strategy is to become the essential infrastructure layer upon which AI is built, deployed, and scaled. Whether a startup is building a bespoke model, an enterprise is deploying internal Copilots, or a major AI lab is running its production workloads, the Azure infrastructure underneath typically carries a Microsoft toll. Azure revenue grew 39% in the most recent quarter, and backlog grew over 110%.
There are risks worth watching. Microsoft plans to spend close to $100 billion this year on data centers. If the return on investment for customers (many of which are still unprofitable in their AI deployments) does not materialize, that capital expenditure becomes a drag on profitability. We are monitoring both the infrastructure build and the performance of Microsoft’s Office suite, which faces the same seat-versus-agent pressure as the broader software market.
Salesforce is navigating a more direct version of this challenge. CEO Mark Benioff has pivoted the company aggressively around Agentforce, reorienting Salesforce from a system that helps companies track customers to a platform that performs customer-facing work on their behalf. The company now describes what it sells as digital labor and has introduced a new metric called the agentic work unit to capture the value AI provides independently of traditional seat counts. Annual recurring revenue from agentic products was up over 160% in the most recent quarter.
The underlying logic is durable: Salesforce controls the customer data, and AI agents are only as effective as the data they can access. A competitor’s agent, however capable, may be at a structural disadvantage without the depth and history that Salesforce’s system of record provides. The legacy business is slowing, and the transition is early. We expect continued share price volatility until the new model clearly offsets that deceleration. But the moat is real.
Summitry’s Philosophy
Zooming out from individual names, our broader positioning follows one of our consistent principles: favor companies that own the infrastructure or the data, apply discipline on valuation regardless of the excitement the theme generates, and hold the analytical question steady: does AI broaden this company’s moat or erode it?
A meaningful portion of our portfolio also consists of businesses whose core value proposition is largely independent of how the AI transition resolves in the near term. People will still buy coffee, maintain their homes, seek healthcare, and return to the brands they grew up with. These are not primarily AI-driven businesses. They are businesses whose competitive advantages are taste, trust, habit, nostalgia, and which have historically demonstrated the ability to adapt to technological change without losing what makes them valuable.
Louis Vuitton was founded in 1854, before electricity, the telephone, or the automobile. Nintendo manufactured playing cards for most of its early history before pivoting to video games in the 1980s. Universal Music has navigated the transition from vinyl to cassette to CD to digital download to streaming. Each format shift looked, at the time, like a potential existential threat. Each time, the underlying asset proved more durable than the delivery mechanism.
Our Investment Principles Haven't Changed
In our view, AI does not require a new investment philosophy. It requires more rigorous application of the one we already have.
Moats still matter, but you have to assess whether AI is widening or narrowing them. Earnings durability still matters, but the source and timeline of that durability may be shifting. Management quality matters more than ever, because execution in a fast-moving environment is genuinely difficult. And margin of safety always applies, especially when valuations embed significant optimism about outcomes that are still uncertain.
We are not avoiding software. We are being more selective about it and favoring companies that control infrastructure and data over those whose competitive positions are more exposed to the agentic shift. The transition is real, the uncertainty is real, and the opportunities it creates are real. Navigating all three with discipline is exactly what this moment calls for.
The securities identified and described do not represent all of the securities purchased, sold or recommended for client accounts. The reader should not assume that an investment in the securities identified was or will be profitable. The opinion contained herein are subject to change without notice and contain forward looking views which may not be accurate.
General commentary for informational purposes only—not investment, legal, or tax advice, and not a recommendation of any security or strategy. Forward-looking statements are subject to change. There is no guarantee that any strategy will be successful. Past performance is not indicative of future results. The S&P 500 Index is unmanaged and not available for direct investment.
GET THE NEXT SUMMITRY POST IN YOUR INBOX:
