For more than a decade, the dominant logic in venture capital has been straightforward: fund software companies that sell subscriptions. The SaaS model — software as a service — rewarded scalable code, recurring revenue, and high gross margins. Now one of Silicon Valley's most influential firms is signaling that the era's defining playbook may be losing its edge. Julien Bek, an investor at Sequoia Capital's London office, has published a thesis titled "Services: The New Software," arguing that the next trillion-dollar company will not sell tools at all. It will sell outcomes.
Bek's central claim is that AI-native firms can deliver finished results — a resolved customer inquiry, a completed financial audit, a fully designed marketing campaign — rather than the software platforms traditionally used to manage those tasks. In this framing, the product is the outcome itself, and the underlying technology is invisible to the buyer. By combining AI systems with human expertise where necessary, these companies would function less like conventional SaaS providers and more like business process outsourcers, albeit with fundamentally different cost structures and scalability profiles.
From tools to results
The distinction Bek draws is not merely semantic. Traditional software companies sell capability: a CRM platform gives a sales team the tools to track leads, but the team still does the work. An outcome-oriented company, by contrast, would guarantee the leads themselves — or the closed deals. The customer pays for what was accomplished, not for access to a dashboard.
This reframes the competitive landscape in important ways. Software margins have historically been attractive because code scales at near-zero marginal cost. Services businesses, by contrast, have been penalized by investors for their reliance on human labor, which scales linearly and compresses margins. The AI-native model Bek describes attempts to resolve that tension: if machine intelligence handles the bulk of execution, the cost structure can approach software-like margins while the revenue model looks like services. The hybrid is, in theory, the best of both worlds.
The idea has precedent, even if the technology enabling it is new. The outsourcing industry — from Accenture to Infosys — has long sold outcomes in various forms, bundling labor with process expertise. What changes with AI is the ratio of human to machine effort, and therefore the economics. A customer service operation that once required hundreds of agents might be handled by a small team overseeing AI systems. The gross margin profile shifts dramatically.
The tensions ahead
Sequoia's framing carries weight because of the firm's track record in identifying structural shifts early — from the rise of mobile to the cloud transition. But the thesis also introduces questions that remain unresolved.
Pricing is one. SaaS companies benefit from predictable, subscription-based revenue that public markets reward with high multiples. Outcome-based pricing is inherently more variable: if the AI fails to deliver, the company may not get paid. That volatility could make these businesses harder to value and harder to finance at the growth stage.
Accountability is another. When a company sells a tool, the buyer owns the result. When a company sells the result itself, liability shifts. Errors in AI-delivered outcomes — a flawed legal summary, a mishandled customer complaint — raise questions about who bears responsibility and how contracts are structured.
There is also the question of defensibility. Software companies build moats through switching costs, network effects, and data accumulation. An outcome-selling company must demonstrate that its results are consistently superior, which may depend more on proprietary models, training data, and operational discipline than on traditional software moats.
Bek's thesis arrives at a moment when the AI industry is moving past its initial infrastructure phase — chips, foundation models, cloud capacity — and into the application layer, where the question of how AI creates tangible business value becomes urgent. If Sequoia is right, the companies that capture the most value will not be the ones building the best models, but the ones using models to replace entire workflows.
Whether the market rewards that model as generously as it rewarded SaaS remains an open question — one that depends on execution, pricing innovation, and the willingness of enterprise buyers to hand over not just tasks, but trust.
With reporting from Fortune.
Source · Fortune
