Apple's 2028 iPhones to Use 1.4nm A22 Pro Chips

[melgross]

I wouldn't be surprised if Apple simply put Intel chips in all U.S. market iPhones and Macs.

The volumes of both Pro and base chips are so high, it doesn't make sense to artificially limit Intel to specific products. This avoids the whole "customer return" issue as well.

Apple can't satisfy the whole "chips made in U.S." if only a random model or SKU may contain Intel chips. It has to be definitive for Apple to satisfy the White House and avoid potential tariffs.

I think that the effort would be enough. Trump won't be here after 2028 so His policies will no longer be in effect anyway. It takes years for these things to happen and can't possibly be completed by the end of his term.

 

[JPack]

I think that the effort would be enough. Trump won't be here after 2028 so His policies will no longer be in effect anyway. It takes years for these things to happen and can't possibly be completed by the end of his term.

That assumes his ideas just vanish and a successor reverses everything. The reality is the current policies are driving re-shoring efforts for chip manufacturing.

 

[deconstruct60]

I seriously doubt that. Dual sourcing is rather cost prohibitive (re-design, new tape out/mask set etc) and there will be performance differences. They might put the Intel chip in a different device though...

It doesn't have to be "dual sourced". The A18 and A18 Pro are two different dies. The M4 , M4 Pro , and M4 Max are also three different dies. (the M5 Pro/Max use chiplets to shrink that down to two, but if Apple did a specific PCC die that would bring count back up to three). The watch chip is also a different die. The R1 is a different die. The displayless glasses may get their own chip/die. At this point Apple has a decently large family dies that need to be made.


This could be done by pruning off a specific die size(s) for Intel. For example could do plain Mn and 'data center' on Intel and the others would be TSMC. The Ann Pro + plain Mn could be done and rest TSMC. There are lots of possible combos there.

If assign the die size to a fab then there is only one mask, one tape out for that die size. There is a narrower set of costs of taking the basic architecture and deploying to a process design kit (PDK) it 'twice' , but that is much lower than what are getting at. There are design tools 'helpers' that can help with that if using a set of design tools that span both fabs.

 

[Reason077]

Apparently Elon Musk is building the biggest and most advanced chip plant in the work in Texas for $20Bn

Firstly, Terafab is a partnership with Intel. Secondly, it will neither be the biggest nor the most advanced chip factory when it opens. There are grand plans to make it huge eventually, but Elon's grand factory plans rarely reach full fruition. Tesla's Gigafactory Nevada was planned to be enormous originally, but only a fraction (about 1/3) of the original plan has actually been built. At Gigafactory Texas, the factory building is enormous but must be pretty empty inside because it's only reached a fraction of its planned production capacity.

Yes, it sounds like Terafab is going to be experimenting with new fabrication technologies, which I guess is where the "most advanced" claim comes from. But that means it will be very limited in production volume initially. New technology in chip manufacturing rarely works out well the first time - something that Intel knows very well!

Apple is crazy not to do the same... they have the money, and they need the chips... what's keeping them?

It takes many, many years to build up the knowledge for advanced chip fabrication, and it's extremely capital-intensive. China has been trying for years and still hasn't caught up with TSMC, Samsung, or even Intel for that matter. Elon can take a punt on it because he has virtually no risk aversion and totally dominates the boards of his companies. But for a public company with good governance like Apple, it's much more difficult to commit serious capital to such a risky gamble.

Plus, the current AI boom and semiconductor shortage is obviously spurring a massive increase in capital spending on fab plants anyway. This should benefit Apple in the coming years when there will likely be an excess capacity of advanced chip production, so Apple can play off the fabs against each other to get the best deals on capacity at the most advanced fabs.

 

[Reason077]

I think that the effort would be enough. Trump won't be here after 2028 so His policies will no longer be in effect anyway.

But I'm not sure that JD Vance will reverse many of Trump's policies.

 

[name99]

As others have pointed out, the proper way is 14A. Apologies.

For Intel it's A14; for TSMC it's 14A!
Which one will Apple use? Well...

 

[Axemantitan]

That assumes his ideas just vanish and a successor reverses everything. The reality is the current policies are driving re-shoring efforts for chip manufacturing.

The Chips Act was a Biden policy.

 

[JPack]

The Chips Act was a Biden policy.

It was the current administration that bought a 10% stake in Intel. The tariffs (or threat of) on imported chips, was also by the current admin.