Available for orders from 15 Feb 2026.
Available for orders from 15 Feb 2026.
A two-part essay on why the firms that win the AI era will not merely sell labor — they will sell orchestrated business consequences
Every big technology shift comes with the same prediction: this time, the middlemen disappear. The tool gets smarter, the interface gets simpler, and suddenly we’re imagining a world where you just want something and it happens — no specialists, no operators, no layers of people standing between intention and result.
It’s a seductive story. It flatters the technology and, honestly, it flatters us too. We like the idea that once a machine gets smart enough, all the messy institutional stuff just… dissolves.
But that’s not really how it goes.
What actually happens is more interesting — and more ironic. Every breakthrough clears one kind of complexity and immediately releases another. Something becomes easier to access on the outside, just as it becomes harder to govern on the inside. A component gets cheaper; the orchestration around it gets more valuable. A task gets automated; accountability becomes more strategic. So the spoils rarely go only to whoever built the tool. They go to whoever figures out how to turn the possibility into something that actually runs.
That’s the insight behind Sequoia’s recent essay, Services: The New Software. Its core argument isn’t just that AI makes software more powerful — it’s that the next great company might look less like a traditional software vendor and more like “a software company masquerading as a services firm.” The essay also draws a sharp line between intelligence and judgment, distinguishes between a copilot (sells the tool) and an autopilot (sells the work itself), and makes the rather blunt but important observation that for every dollar spent on software, six are spent on services.
That thesis is right. But I think it’s still underselling the moment.
Because the real question isn’t whether software is becoming more service-like. It’s whether service companies can become the new execution layer of the AI-native enterprise. That’s a bigger claim — and a more interesting one.
History, if you actually look at it, is pretty clear on this. Service firms didn’t grow large by fighting technology transitions. They grew large by stepping into the dislocation those transitions created. When a new stack emerged, when systems got more interdependent, when enterprises needed to migrate, integrate, govern, and operate at scale — that’s when service firms thrived, because someone had to absorb the complexity the technology left behind.
IBM is the clearest example. By 1999, its Global Services arm was generating $32.2 billion in revenue, up from $25.2 billion just two years earlier. That wasn’t accidental. As enterprise computing became more networked and more strategic, someone had to stitch it all together, run it, and stand behind it. IBM didn’t survive the transition to services — it built its next act around it.
The Indian technology majors followed a similar arc, though the machinery looked different. Their rise was never purely a labor-cost story. It was about building an industrial model for enterprise change: global delivery, process discipline, quality control, the ability to absorb enormous volumes of business and technology work with consistency. A recent NITI Aayog report describes India’s tech services sector as accounting for roughly 7% of GDP — and Reuters, citing Nasscom’s 2026 review, projects the sector will reach $315 billion in FY26, with AI-driven services revenue already estimated at $10–12 billion. That’s not a picture of an industry dying. That’s a picture of one adapting under pressure.
Which brings us to the first principle most commentators still miss: technology doesn’t kill services when it makes the enterprise more powerful. It often enlarges services because it makes the enterprise more complex.
The Sequoia essay gets at this by separating intelligence from judgment. Writing code, triaging alerts, reconciling invoices, matching records — that’s intelligence work. It’s complex, but it follows patterns and rules. Judgment is different. Judgment is knowing when a process should bend and when it should stop. It’s reading institutional context, accepting responsibility for a consequence, deciding whether an exception should be allowed. AI is moving fast into the intelligence layer. The judgment layer still needs humans — or at least, humans remain accountable for it.
That distinction matters a lot for service firms. Because it means the future isn’t just a headcount story. It’s an architecture story.
For years, service firms scaled through labor pyramids — hire more people, distribute work across delivery centers, build process rigor around repeatable tasks, and monetize volume plus reliability. In the ERP era, that meant integration and migration work. In the cloud era, it evolved into platform engineering and managed operations. In the AI era, it evolves again. The new factory runs on agents, orchestration, enterprise context, exception management, and human checkpoints. The factory doesn’t go away. The machinery inside it changes.
That’s why the “AI will commoditize services” argument feels too shallow to me. Yes — it will commoditize certain tasks. It will pressure firms that simply rent out human effort for repeatable work. It will compress labor intensity in parts of the value chain. But enterprises don’t buy isolated tasks. They buy resolution. They buy a reduction in uncertainty. They buy cycle-time compression with accountability attached. They buy the confidence that the work will get done, the exceptions will be handled, the controls will hold, and the output will stand up under scrutiny.
This is also why Sequoia’s copilot-vs-autopilot framing matters so much. A copilot sells a productivity tool to the professional. An autopilot sells the work itself to the buyer. Start with outsourced, intelligence-heavy tasks, Sequoia argues, because those tasks already have external budgets, already have owners, and already represent purchased outcomes rather than internal reorganizations. That’s especially relevant for large service firms — outsourced work is exactly where they already know how to sell, govern, and expand. The market isn’t asking them to invent new categories of work. It’s asking them to reinvent the machinery through which the work gets done.
So no — this isn’t the twilight of services. It’s the beginning of the end of a particular formula for services.
The old formula sold effort. The next one has to sell something more elevated: orchestrated business consequences.
I use that phrase deliberately, because “outcomes” — while technically correct — has become too sanitized. Outcomes sound like the last slide of a consulting deck. Consequence carries more weight. It implies change made real, decisions that actually show up in operations, intent that gets translated into measurable movement. And orchestrated matters just as much, because in the AI era, nothing of real consequence happens through raw model capability alone. It has to be composed — sequenced, governed, verified, observed, and continuously tuned.
That’s the genuine opening for service companies right now. The firms that win won’t just help clients use AI tools more effectively. They’ll become the operating layer through which enterprises convert machine intelligence into institutional movement. They won’t be remembered as sellers of time. They’ll be remembered as governors of intelligent execution.
That’s why the obituary for services is being written too early.
The real story isn’t disappearance. It’s metamorphosis.
Which leads to the harder question: if services aren’t dying, what exactly must they become?
We will continue this discussion in the next part!
(Part II appears here)
”Sadagopan Singam is a global business and technology leader and the author of Agentic Advantage. He advises boards and executive teams on GenAI-driven transformation and autonomous enterprise models.”
