Chapter 6 / Essay
Chapter 6 № 06 · 2026

With the effort of outsourcing,
you can build it yourself.

The friction of outsourcing, plus the customer's invisible labor — for the same effort, you can build it yourself

The effort customers pay to commission an SIer — requirements, vendor selection, contracts, project management, acceptance — consumes as much labor as building it AI-natively, sometimes more. For the same effort, you can build it yourself.

Chapter 5 showed that customers can become the builder and that nine-tenths of the work can close inside customer plus AI. This chapter takes up the other side — why "commissioning an SIer makes life easier" is now an illusion — by decomposing the commission process step by step.

The cost of outsourcing carries, on top of vendor payments, a stack of invisible costs on the customer side. That stack is the focus of this chapter.

The SIer commission model has a longer process than it looks

Moving a single SIer engagement requires a process like this:

"Hand it to the vendor and we are done" does not describe this. Customer-side work continues throughout the engagement. The same is true for small projects and for very large ones — at every step of the process, somebody inside the customer has to stay attached, or the project does not move.

flowchart TB subgraph Sier["SIer commission model — process steps"] direction TB S1["Requirements / RFP
(customer: weeks to months)"] S2["Vendor selection
(customer: weeks to months)"] S3["Contract negotiation
(customer + SIer: weeks)"] S4["Project management
(customer + SIer: full duration)"] S5["Acceptance / UAT
(customer: weeks)"] S6["Ops & maintenance handover
(customer + SIer: ongoing)"] S1 --> S2 --> S3 --> S4 --> S5 --> S6 end subgraph AI["AI-native process steps"] direction TB A1["Customer + AI on requirements
and design
(days to weeks)"] A2["AI implements
(days to weeks)"] A3["Evaluate and integrate
(continuous)"] A1 --> A2 --> A3 end classDef good fill:#e8f5e9,stroke:#7a9a6d,color:#3a4d34 classDef bad fill:#fef3e7,stroke:#c89559,color:#5a3f1a class AI good class Sier bad

The customer's "invisible labor" is the real body of the cost

The thing easiest to miss in commission-cost discussions is the customer's internal labor.

The amount the customer pays the SIer is written in the contract. The time spent inside the customer's organization, to make the engagement move, is not:

None of this hits the invoice line as "labor cost." But it is actually being consumed. IT staff, affected-department managers, sign-off owners — every engagement pulls non-trivial time out of each.

Empirically, in mid-size SIer engagements, the customer-side invisible labor amounts to a significant fraction of the SIer payment itself (the exact ratio varies, but it is never negligible). Even so, until now there was no alternative — building in-house meant hiring and retaining coders, which cost more.

The real cost of commissioning = SIer payment + customer-side invisible labor. The second item is not in the contract, but in practice it carries half the weight.

With the same effort, you can build it yourself

This is the central claim of the chapter.

"Even if the SIer is expensive, we cannot build it ourselves, so we have no choice" — that was the old argument. It does not hold in the AI-native world.

Why not? Because the customer-side labor consumed by an SIer engagement (requirements, vendor selection, management, acceptance) overlaps with the customer-side labor consumed by AI-native in-house development (requirements, design, delegation to AI, evaluation, integration).

The historical difference sat in writing the code. That was where enormous money and person-months landed. Once AI took execution, that difference disappeared.

In other words, the effort to commission an SIer and the effort to build AI-natively in-house are now the same order of magnitude. If you are going to spend the effort anyway, zeroing out the SIer payment is clearly the more efficient route.

Once "the effort to outsource" and "the effort to build yourself" become equal, the rational reason to outsource disappears.

Why SIers cannot absorb this diseconomy

"If SIers themselves use AI, their internal efficiency goes up" — true. Many SIers are integrating Claude or GPT into their workflows. Even so, the SIer commission model cannot, structurally, reach parity with AI-native in-house development.

Four reasons:

The result is that when SIers internally use AI, AI sits on top of person-months kept in place to defend revenue. Costs drop; prices do not. From the customer's perspective, the total cost of an SIer engagement stays anchored at a level above what an in-house AI-native build would cost.

SIers will shrink and reconstitute

This is not "SIers all disappear at once." It is structural shrinkage: nine-tenths moves to the customer side, the SIer share concentrates in the remaining tenth.

Transition speed, Japan-specific dynamics (multi-tier subcontracting), and labor mobility are taken up in Chapters 10 and 11. What this chapter fixes is the structural claim: the SIer commission model cannot reach parity with AI-native in-house development.

SIers do not vanish, but they cannot avoid the 9 : 1 shrinkage and the reshaping of their contract forms.

Where the next chapter goes

This chapter has shown that "for the same effort, you can build it yourself." The next question is: not by effort, but by money, how far apart are they? Putting an SIer quote next to the cost of an AI-native in-house build.

The next chapter takes up that price gap.


Related articles