[lovich]

>Now major chip companies are resorting to unusual tactics. They are placing orders far in advance and prepaying so that substrate companies have ample cash to build more factories. Some are committing to buying the entire supply of new production lines to give their suppliers confidence to invest.

If the chip makers are investing this much money into the substrate businesses, why aren't they just purchasing some of these companies wholesale?

[hfktktnfkgk]

The chip companies also know that they will definitely buy electricity in the future, yet they are not buying power plants.

In a sense, buying the future substrate production is just a future substrate contract.

Do you know how building a power plant is financed today? Basically someone buys 10 years of gas futures, sells 10 years of equivalent electricity futures, and uses the profit (since electricity is more expensive than gas) to build the power plant that will turn the bought gas into the sold electricity.

[HPsquared]

I suppose they could also just sell the electricity futures first, then build the plant, then buy gas as needed. That'd put them at risk of gas prices fluctuating though, might turn off investors.

I assume a random person can't just sell 10 years of electricity futures without some kind of collateral.

[CRConrad]

> I assume a random person can't just sell 10 years of electricity futures without some kind of collateral.

Maybe they run the negotiations in parallel, so they can kind of use one set of futures as collateral for the other?

[EmilioMartinez]

Any source, extra info or further insight on that last bit?

Comes to mind: "Be the change you want to see in the world, and get paid for it"

[azemetre]

It costs over $5 billion dollars to build a single fab. I can only assume the more complicated chips costs 3-4 times more.

It also takes over three years to build the plant itself. We aren't even talking about the people you'd even need to employ, design, and run the place (highly specialized labor).

This is such a huge burdensome cost. It's not even a guarantee that that such an investment would even make sense once 5-years pass.

I do think there should be more fabs and having them located in one area of the world is already playing out to be a geopolitical nightmare. It's obviously going to become a national security issue for nearly every industrialized country on earth over the next decade if not sooner.

[lovich]

In the article it says they are committing to buying the entire production lines output. Isn't that just committing to spending as much money as building the factories + the profit of the vendor, but with no control over the actual production ?

[andrewxdiamond]

It’s very unlikely that the deal says “we will buy everything you sell us no matter what.” It’s probably a typical purchasing agreement with tolerances and QA, just paid in advance.

Thus there is still a lot of risk being taken on by the fabs. The production could go sour, geopolitics could change, resources might be unavailable, etc.

The purchasing companies are externalizing all of this risk by purchasing the products instead of the company.

[hermitdev]

The chip companies probably have forms of insurance in place to cover losses if their independent supply chain partners fail to deliver that they wouldn't be able to get if the partners were integrated.

[lovich]

ah, that was the part I did not understand

[petra]

Buying a company is much more expensive than paying for a manufacturing line.

[adamdusty]

The margin isn't high enough on substrates to make buying and maintaining the process worth it.