A new Goldman Sachs research note published in mid-March 2026 argues that autonomous AI agents are poised to fundamentally reshape the economics of the entire software market — potentially redirecting the majority of future profit pools away from traditional SaaS subscriptions toward agentic workflows that execute tasks end-to-end with minimal human supervision.
The Headline Projection
According to the analysts, by 2030 more than 60% of industry operating profit in the broad software sector could accrue to systems built around AI agents rather than conventional SaaS products. That would represent a dramatic reallocation of value away from today’s dominant subscription-based revenue model.
The core thesis is straightforward:
- Current SaaS companies sell access to tools (CRM, project management, analytics dashboards, etc.).
- Agentic systems sell outcomes — they don’t just provide software; they complete the work.
Once agents can reliably own multi-step business processes (lead qualification → proposal generation → contract negotiation → invoicing → support hand-off), the willingness to pay shifts from monthly seat licenses to outcome-based pricing or percentage-of-value-created models.
Why the Shift Is Inevitable
Goldman identifies three converging forces:
1. Rapid capability gains
Agent reliability, long-term memory, error recovery, and multi-step planning have improved dramatically in the past 12–18 months. What was mostly demo-ware in 2024 is already moving into controlled production pilots in 2026.
2. Economic logic
Knowledge work is expensive. If agents can deliver the same (or better) output at 20–40% of the human cost, companies will reallocate budgets accordingly. The surplus value created by that productivity gain becomes the new profit pool — and the agent platform that captures it stands to earn outsized margins.
3. Changing buyer behavior
Forward-leaning enterprises increasingly ask vendors: “Can your product do the job, or do I still need people to use your product?” When the answer becomes “yes, it can do the job,” the purchasing conversation moves from “how many seats?” to “what percentage of value created will you take?”
The Missing Infrastructure
While the agent vision is compelling, Goldman cautions that the industry still lacks several critical layers required for mass adoption at enterprise scale:
- Stable platform layer — reliable orchestration, observability, rollback, audit trails;
- Identity & security primitives — strong agent authentication, granular permissioning, verifiable execution provenance;
- Data integrity & lineage controls — ensuring agents operate on correct, up-to-date, governed data sources.
The report estimates that meaningful standardization across these areas could emerge within the **next 12 months** (i.e., by mid-2027), after which agent adoption should accelerate sharply.
Earliest High-Confidence Use Cases
Goldman highlights four domains where autonomous agents are already showing strong pilot results and could scale fastest:
- Customer support — full ticket resolution (investigation + response + follow-up);
- Sales development — outbound prospecting, personalization, initial qualification;
- Marketing operations — campaign creation, A/B testing, creative iteration, performance reporting;
- Software development tooling — code generation, test writing, PR review, deployment automation.
In each of these areas the economic case is particularly clear: high labor intensity + reasonably well-defined workflows + acceptable risk when agents err.
Winners and Losers in the Transition
The report sketches a bifurcated future for software vendors:
- Winners — companies that successfully wrap existing business processes inside agentic wrappers and capture a share of the productivity surplus they create. These players become the new “interface” to knowledge work.
- Losers — pure SaaS vendors that remain tools rather than outcomes providers and cannot pivot to agentic delivery fast enough.
Goldman notes that many of today’s highest-multiple SaaS businesses are already racing to build agent layers on top of their core products — precisely because they understand the long-term threat to their current model.
Also read:
- The Humble Origins of the "Like": A Scaling Hack That Conquered the Internet
- Prada's Versace Acquisition Reshapes Luxury Fashion: The Rise of the Big Three
- The Great Divergence: America's K-Shaped Economy Accelerates into a Post-Labor Future
Bottom Line
If Goldman’s 60%+ profit-share forecast proves directionally correct, the software industry is facing one of the largest value-migration events in its history. The winners will be the platforms and workflows that move from selling shovels to actually digging the gold — and taking a cut of the treasure they help unearth.
For now the race is still early. But the direction of travel is unmistakable: the future of software looks less like monthly logins and more like autonomous digital employees that quietly get work done in the background.
The full report is available at goldmansachs.com/insights/articles/ai-agents-to-boost-productivity-and-size-of-software-market.
