Services: The New Software – Why the Next $1 Trillion Company Will Look Like a Services Firm
In March 2026, Sequoia Capital published a provocative essay titled "Services: The New Software". The core idea is as bold as it is timely: the next trillion-dollar company won't sell software tools. It will sell the actual work itself — powered by AI, delivered as a seamless service, and disguised as a traditional services business.
This thesis resonates deeply with many teams building in the AI space right now. We're not just seeing incremental productivity gains; we're witnessing a fundamental shift in how value is created and captured in the economy.
Intelligence vs. Judgement: Where AI Draws the Line
Sequoia draws a clear distinction between two types of cognitive work:
- Intelligence — Rule-based, complex but ultimately procedural tasks. Think writing code from a spec, running tests, debugging, translating medical notes into standardized billing codes, or screening thousands of resumes for basic qualifications.
- Judgement — The human domain of experience, taste, intuition, and high-stakes decision-making. Deciding which feature to build next, whether to take on technical debt for a faster release, assessing cultural fit in hiring, or advising on strategic risk.
AI has already crossed a critical threshold. It can now handle intelligence work autonomously at scale. Software engineering was the first profession to feel this shift — which is why more than 50% of all AI tool usage today comes from developers. But the wave is spreading rapidly to every knowledge profession.
As models improve, what counts as "judgement" today will increasingly become tomorrow's "intelligence," especially as AI companies accumulate proprietary domain data.
Copilots vs. Autopilots: Tools vs. Outcomes
- A Copilot is a productivity tool. It helps professionals do their job faster (e.g., AI assistants for lawyers reviewing contracts or developers writing code). The customer still owns the work and the responsibility.
- An Autopilot sells the finished work itself — the outcome. The customer buys a completed result, not a tool.
Why sell lawyers an AI tool to review NDAs faster when you can sell businesses a fully negotiated, compliant NDA as a service? Why sell IT management software to managed service providers when you can sell companies the simple promise that "your IT just works"?
The economics favor autopilots dramatically. Copilots compete for the relatively small "tool budget." Autopilots go after the much larger "work budget."
The Golden Ratio: $1 on Software, $6 on Services
Here's the staggering insight: for every $1 businesses spend on software, they typically spend $6 on services — consultants, implementers, agencies, outsourced teams, and external labor.
This isn't a small side market. It's a massive pool of spending that AI agents are now positioned to capture. Because AI can finally *do* the work rather than just assist with it, autopilots can eat into both outsourced services and eventually insourced labor.
The smartest playbook? Start with outsourced, intelligence-heavy tasks as a wedge. These have clear budgets, existing acceptance of external providers, and verifiable outcomes. Once you prove value and gather data, you expand upward into more judgement-oriented or insourced work.
The Innovator's Dilemma for AI Builders
Incumbent copilots — the breakout AI tool companies of 2025 — now face a painful choice. Many are trying to evolve into autopilots using their customer relationships and data. But there's a catch: if they start delivering the full work, they risk cannibalizing their own users (the agencies, law firms, accounting practices, and MSPs who currently buy their tools).
This creates a classic innovator's dilemma and opens a wide window for new startups that are built as autopilots from day one — companies that have no legacy tool business to protect.
Where the Biggest Opportunities Lie
- Recruiting and staffing ($200B+): Resume screening, initial matching, and high-volume sourcing are largely algorithmic. Judgement only kicks in later for culture fit and final offers.
- IT Managed Services ($100B+): Monitoring systems, granting access, triaging alerts — repetitive intelligence work that MSPs handle today.
- Medical billing / healthcare revenue cycle ($50–80B outsourced in the US): Converting clinical notes into tens of thousands of standardized codes for insurers.
- Accounting and audit ($50–80B outsourced): Journal entries, month-end close, compliance filings.
- Transactional legal work ($20–25B): Standard contracts, NDAs, routine filings.
Other large markets mentioned include insurance brokerage, claims adjusting, tax advisory, supply chain/procurement, and even parts of management consulting.
In each case, the pattern is similar: high volume of rule-based labor, structural shortages of human talent, and existing willingness to pay for outsourced outcomes.
Alo read:
- Marble 1.1 — World Labs Just Made Their World Model Significantly Better
- Unmasking Runway Characters: The Unexpected Rise of the Real-Time Avatar
- The Dawn of the Wisdom Era: Why Your Intelligence is No Longer Enough
The Bottom Line: Stop Selling Shovels — Start Digging the Holes
The next legendary company will look like a services firm on the outside — reliable, outcome-focused, client-facing — while operating like a highly leveraged software platform on the inside, powered by AI agents that handle the heavy intelligence lifting.
For those of us already building in this direction, Sequoia's piece feels less like a prediction and more like validation. AI isn't just changing how we work; it's changing what we sell. The winners won't be the best tool providers. They'll be the ones who deliver the best results — faster, cheaper, and at scales previously unimaginable.
The full original essay from Sequoia, with more examples and a helpful opportunity map, is well worth reading: Services: The New Software.
The shift is already underway. The question for founders and builders is no longer whether to move from copilots to autopilots — but how quickly and decisively to make that leap.