AI, SaaS, and the Myth of Inevitable Consolidation

There’s a familiar story people tell whenever a new technology wave hits.

First comes an explosion of startups.
Then comes consolidation.
Finally, a few winners emerge and capture most of the value.

This story is so deeply embedded in SaaS thinking that it’s often treated as a law of nature. And right now, it’s being applied wholesale to AI - especially in the context of AI startups built and sold to traditional white-collar companies.

The big problem with this thesis? White-collar industries themselves tell a very different story on consolidation … and AI is apparently going to eat white collar work.

The SaaS Mental Model Is Leaking

Classic SaaS economics assume a few things:

• Software is a tool, not the work itself

• Value is captured upstream of the platform

• High fixed costs justify consolidation

• Winner-takes-most dynamics emerge through distribution and lock-in

This works when software represents a small but critical slice of a broader workflow.

CRMs, ERPs, databases, cloud infrastructure - these are picks and shovels. They enable work, but they don’t do the work.

White-collar industries - law, accounting, consulting, finance - operate on a different axis entirely. Their value isn’t in interfaces or automation alone. It lives in judgment, accountability, trust, and context.

For decades, software has been maybe 10% of the value chain in these industries, with services making up the remaining 90%. And in that world, consolidation simply doesn’t behave the way SaaS people expect it to.

By way of example, the global legal services market is roughly a trillion dollars yearly. The largest law firm in the world captures around $8bn of that. That's right, the "market leader" is less than 1% the total market. Accounting, consulting, advisory - same pattern. Huge markets but with highly fragmented outcomes.

That’s not an anomaly. It’s a structural feature of expertise-driven work.

Why “Explosion → Consolidation” Often Fails in Services

The explosion-then-consolidation pattern tends to hold when innovation produces tools.

Think railroads, operating systems, cloud compute, or databases. These are infrastructure layers. They’re modular, substitutable, and price/performance driven. Scale wins.

But services behave differently.

When value is delivered through human judgment, responsibility, and situational nuance, markets resist consolidation. Expertise decentralizes. Trust fragments. Accountability remains local.

AI complicates this - but not in the way many expect.

The Picks and Shovels Analogy Breaks When the Pick Swings Itself

For years, AI was framed as another pick and shovel. A productivity layer. A helper.

But something fundamental has shifted.

AI is no longer just assisting the work. It is increasingly how the work gets done.

When software decides:

• what to do,

• how to do it,

• and produces the output,

it stops being a tool and starts behaving like labor.

This is the critical transition most SaaS narratives fail to grapple with.

If AI is 90% of the value chain, whoever controls it doesn’t just have pricing power - they have labor leverage. And labor markets don’t consolidate cleanly. They fragment.

The Fantasy of God-Like AI Companies

Some people extrapolate from this and conclude we’re heading toward monopolies or duopolies: single AI vendors that own the models, the workflows, the client relationships, and the outputs.

But follow that logic through.

If your supplier becomes your entire operation, then you’re no longer a firm - you’re a thin marketing reseller for someone else’s intelligence.

Concentrating that much leverage creates unavoidable pressure:

• regulatory scrutiny,

• liability concentration,

• client resistance,

• talent flight.

The result isn’t a stable monopoly. It’s inevitable decentralization.

The SaaS × Services Hybrid Trap

There’s a seductive pitch floating around right now:

“We’ll combine SaaS margins with services lock-in.”

History is not kind to this idea.

Services drag margins down.
SaaS strips differentiation out.

You end up stuck in the middle - too bespoke to scale like software, too standardized to command services premiums.

Services firms that over-productize disintermediate themselves.
SaaS firms that move downstream quietly become agencies with worse economics.

AI doesn’t resolve this tension. It accelerates it.

That’s why so many AI companies feel structurally broken - not because the technology doesn’t work, but because the business model doesn’t compute.

Expertise Always Decentralizes

Tools can centralize. Capital can centralize. Infrastructure can centralize.

Expertise almost never does.

Hotels are places to sleep. Taxis are places to drive. Platforms dominate those layers. But effort, risk, and quality remain decentralized.

White-collar work behaves the same way.

AI doesn’t eliminate that reality. It amplifies it.

What Actually Scales in an AI White-Collar World

If monopolies aren’t the endgame, what is?

What scales is not ownership of intelligence, but coordination of it.

• Marketplaces of cognitive labor

• Interfaces that standardize access to bespoke work

• Systems that let experts remain independent but augmented

• Platforms that orchestrate outcomes without absorbing all the IP

What doesn’t scale:

• closed, end-to-end intelligence stacks,

• single-vendor lock-in,

• SaaS pricing applied to labor replacement.

The mistake is assuming intelligence scales like code.

History suggests it scales like people.

The Real Bet - No Consolidation, Continued Proliferation

AI doesn’t lead to consolidation because it doesn’t behave like software.

It behaves like labor.

And labor markets don’t converge neatly into monopolies - they fragment, specialize, and rebalance power over time.

The companies that survive won’t be god-like owners of intelligence. They’ll be the ones that understand where software ends, where work begins.