Intel's effort to build a foundry biz is costing far more – and taking longer – than expected Three years after CEO Pat Gelsinger announced Intel would create a foundry business that took on contract manufacturing gigs, Chipzilla has committed to more than $185 billion in spending across new and existing fab, packaging, and test sites. Now with $8.5 billion in US CHIPS Act subsidies in hand and up to $11 billion of …
Now that is an intelligent choice for a chip foundry.
Just below the Great Lakes, water in abundance, it is a much better choice than Arizona.
Good on Gelsinger for having done that. Now why did he have to go and put two fabs in the driest place of the US of A which is subsisting only because it is sucking up an entire river of water that is failing ?
A new advanced complex chip design costs upwards of $500 million with much of that being specific to the type of fab (ie much of the costs would have to be repeated if the production was moved from one fab company to another). When eventually Intel's new fabs come online Intel will have an uphill battle to attract customers as not many will want to pay over $500 million on an unproven process when they can use TSMC for the same price. Intel will probably have to operate at a loss with deep price cuts to customers for years until their processes are proven.
Will Intel's shareholders agree to keep funding the fab construction or will they lose their nerve?
I see your point. There does however seem to be a significant push in North America and Europe to re-shore some strategic industries, like pharmaceuticals and chip-making, to prevent the sort of supply chain issues that occured during COVID and still exist today (two years later). Plus, geopolitics evolved to the point where Western tech advances (including in the military) appear to be on the verge of being overtaken by China's, which could lead to serious turmoil given its apparent allies (Russia, North Korea, Iran ...). It'll be the patriotic thing to do to bake chips locally once the fabs are online IMHO, and so much the better for Intel if it manages to operate 10 of them.
> Will Intel's shareholders agree to keep funding the fab construction or will they lose their nerve?
Depends on whether they can convince them that with this investment Intel will improve its chances of maintaining market dominance and thus be able to operate as a monopoly. Ie. Borrow the business plan of Uber et al.
Foundry us contract labor low ASP.
Microsoft as a cloud account for Intel foundry =$29.00 per device, no IP and customer can move tomorrow to TSMC. Same customer was using them at $500.00 per device and Intel had the IP but...they screwed it up with icelake....Foundry is not good business for IDM, it means you failed.
Hmm ...
Being Intel's fab customer -- or anybody else's fab customer -- means your profitability will be limited vs that of your competitors who develop their own fab tech, because the latest and best processes are going to be used only by the companies which develop them.
Huh?! I don't see TSMC selling chips. I do see Apple being "rather" profitable by designing chips for TSMC to fab.
I think Samsung use Qualcomm chips, forming a "sandwich", if you will, where quite likely the fabless "filing" of Qualcomm makes more of profit than the Samsung "bread".
@Mark 124: I agree with what you wrote, but that wasn't what I was writing about.
I was writing about, e.g., the profit potential of Company A, which fabs its own chips, and sells them to everyone, and the profit potential of Company B, who outsources the fabbing of their chips, and who also sells their chips to everyone.
For closely-performing chips, the profit potential of Company A is higher vs Company B, because Company A can take advantage of higher chip densities available from the latest fab processes, and thus make their chips more-cheaply than can Company B.
Will the US State Department allow Intel to export products from these fabs without purchase by purchase authorization to protect National Security? That has been one of the biggest impediments to building fabs in the US. Since they are highly automated, labor isn't even a rounding error in the costs.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
