Nearly thirty years of knowing why a 2011 decision left a column nobody uses.

That is the thing Porsche just sold, and no line item on the deal describes it.

The arithmetic of the week

On August 24, Porsche agreed to sell MHP to Tata Consultancy Services. MHP is the management and IT consultancy Porsche started buying into in 1998 and fully owned in early 2024: more than 4,500 people, about 743 million euros of 2025 revenue, roughly 300 clients across automotive, manufacturing, aerospace, defense and the public sector.

The price is 320 million euros of enterprise value, in cash. TCS disclosed it in its stock-exchange filing; Porsche’s own announcement does not carry the figure.

Then Porsche signed a five-year partnership with TCS worth roughly 1.25 billion euros, to run an AI and mobility center of excellence.

One payment in. Four times that, out, over five years.

The accounting is fine. That is not my objection.

Porsche had a brutal 2025. Operating margin fell from the mid-teens to barely above break-even. A sale that lands in one year against an expense spread across five is exactly the shape of transaction a CFO uses to steady a balance sheet, and I would not pretend otherwise.

So the financial case closes. Fine.

My problem is a different ledger, the one nobody keeps.

Execution transfers. Compounding does not.

A consultancy that has lived inside your systems for two decades holds knowledge that exists nowhere else. Which plant reports differently. Which integration breaks at month end. Who to call when the official process fails.

None of that is in a document. It is in people, and people now work somewhere else.

Under contract you keep the ability to ask for work and receive it, probably done well. What you lose is accumulation. Every project MHP delivers from here improves an asset that belongs to TCS, and TCS can sell that asset to every other carmaker on earth.

We made the same argument from the opposite direction in the piece on AI companies buying SaaS instead of building it. The question underneath both is which judgment you want living inside your walls.

Your version of this is smaller and identical

Almost no mid-sized company owns a 4,500-person consultancy. Almost every one of them is making this exact decision right now without naming it.

You hire someone to build the agents. Someone else to write the evaluations. A third party to wire the model into the ERP. Each contract is defensible on its own and probably correctly priced.

Two years later nobody inside can explain why the system decides what it decides, and renegotiating with the vendor is not a negotiation anymore.

I am not telling you to stop outsourcing. We are an outsourced partner, so that would be a strange position to hold. I am telling you to split the decision in two: what you want done, and what you want your company to learn how to do. Those are different questions and they almost never get asked separately.

At IQ Source the process map and the Go/No-Go gate come before any tooling choice, and part of that work is marking which steps are permanently yours and which can leave without strategic cost. We went deeper on that split in building is cheap, deciding not to is expensive.

Three questions before the next contract

If this vendor disappeared in eighteen months, could anyone inside explain how the thing they built works?

Does the knowledge from this project end up in a format your people actually open, or in the head of someone who bills hourly?

Are you buying an outcome or a capability? If it is the second, the contract should say how it transfers.

Porsche made a legitimate financial move in a bad year and it will help. The rest of the invoice does not come due in 2026. It comes due the day they need to rebuild something they already knew how to do.

Let’s separate what you outsource from what your team learns