Broadcom to move its headquarters back?

[calexanian]

The news is all aflutter about Broadcom moving its "Headquarters" back to the U.S. but if memory serves me they were always headquartered in Irvine CA. The president is saying it will bring jobs back, but if I am not mistaken Broadcom is fabless. I think we are experiencing Fake News in action.

[photon]

Broadcom, which was headquartered in Irvine CA, was bought by Avago who chose the Broadcom name but kept the Avago management. Avago was a private equity created monster to make some few people wealthy. The origins of Avago go back to the old HP semiconductor fab.

[raptor1956]

Generally speaking, the headquarters is but a small fraction of the total workforce so even if there was an actual return to the USA I doubt we'd be talking a significant number of people.  Apple makes most of there products outside the USA, mostly in China, so of the roughly one million people that work in some capacity for Apple the domestic workforce is a tiny fraction of the world total.  And, as many companies that make products in the semiconductor industry are actually fabless the actual workforce employed in the USA is mostly headquarters, design and programming.

Just as with accounting it's not hard to journal things around to make whatever picture you want.  The motion picture industry has played the game of paying big stars a percentage of the net for a film, but because accountants can move things around even hugely successful films can be made to appear as though they just broke even.  It's all a game!


Brian

[calexanian]

This leads to a big question. Do you think it will ever be viable to build a new consumer product level fab in the US? Most american semicon company's are either fables or high end on smaller wafer size lines.

[iainwhite]

I suspect the main reason for this is the expectation/hope that the President will pass his plan to reduce US corporation tax from 35% to 20ish%, and the fact that Broadcom's deal on Tax in Singapore will expire in 3 years or so.

I very much doubt this will create any US manufacturing jobs, although there will probably be a net job gain in the US for Broadcom.

[coppice]

This leads to a big question. Do you think it will ever be viable to build a new consumer product level fab in the US? Most american semicon company's are either fables or high end on smaller wafer size lines.

Fabs need few people, but huge amounts of cash. Cheap labour helps them a little, but sweet financing deals help them a lot. They tend to go where the financing takes them, ITAR rules permitting.

[tooki]

This leads to a big question. Do you think it will ever be viable to build a new consumer product level fab in the US? Most american semicon company's are either fables or high end on smaller wafer size lines.

Where’d you get that idea? AFAIK, lots of American semiconductor makers make the silicon in USA, and then ship the finished wafers abroad for testing, dicing, and packaging. (If they aren’t doing that domestically, that is.) For example, Intel makes 75% of its products in USA, and their most recent fab opened in USA in 2013.

I remember back in 2003, when Apple released the Power Mac G5, whose PPC970 chip was made by IBM, how IBM made a huge fuss about their new $3B fab in upstate NY. (And to this day, that’s also a contract fab for fabless semiconductor companies.)

[photon]

This leads to a big question. Do you think it will ever be viable to build a new consumer product level fab in the US? Most american semicon company's are either fables or high end on smaller wafer size lines.

Where’d you get that idea? AFAIK, lots of American semiconductor makers make the silicon in USA, and then ship the finished wafers abroad for testing, dicing, and packaging. (If they aren’t doing that domestically, that is.) For example, Intel makes 75% of its products in USA, and their most recent fab opened in USA in 2013.

I remember back in 2003, when Apple released the Power Mac G5, whose PPC970 chip was made by IBM, how IBM made a huge fuss about their new $3B fab in upstate NY. (And to this day, that’s also a contract fab for fabless semiconductor companies.)

Intel used to be the largest (by revenue) semi fab. They are now surpassed by Samsung.
IBM is no longer in the semi fab business. They essentially gave their fab business away to Global Foundries.
TSMC is the largest fab for fabless companies.
Intel is the only remaining competitive semi fab in the US but is quickly losing its technical leadership in 7 and 5 nm geometries of which TSMC and Samsung are the current leaders.
The remaining semi fabs in the US are trailing edge.
So I would answer the original question with "No, it does not seem likely that the US will have a leading edge semi fab, either new or Intel."

[photon]

BTW, there have been recent rumors that Broadcom wants to buy Qualcomm/NXP for $100B. https://www.eetimes.com/document.asp?doc_id=1332556&piddl_msgid=374640#msg_374640

[coppice]

Intel is the only remaining competitive semi fab in the US but is quickly losing its technical leadership in 7 and 5 nm geometries of which TSMC and Samsung are the current leaders.

Competitive is the wrong word there. Leading edge is a better fit. There are still quite a few fabs in America for parts which don't need a leading edge process which operate very competitively. Most devices really don't need a leading edge process, and can't benefit from one.

Isn't Samsung supposed to be building a new leading edge fab for memory in the US?

[photon]

Intel is the only remaining competitive semi fab in the US but is quickly losing its technical leadership in 7 and 5 nm geometries of which TSMC and Samsung are the current leaders.

Competitive is the wrong word there. Leading edge is a better fit. There are still quite a few fabs in America for parts which don't need a leading edge process which operate very competitively. Most devices really don't need a leading edge process, and can't benefit from one.

Isn't Samsung supposed to be building a new leading edge fab for memory in the US?

It's not wise to be trailing edge nor to be an outsourcing destination.

[tooki]

This leads to a big question. Do you think it will ever be viable to build a new consumer product level fab in the US? Most american semicon company's are either fables or high end on smaller wafer size lines.

Where’d you get that idea? AFAIK, lots of American semiconductor makers make the silicon in USA, and then ship the finished wafers abroad for testing, dicing, and packaging. (If they aren’t doing that domestically, that is.) For example, Intel makes 75% of its products in USA, and their most recent fab opened in USA in 2013.

I remember back in 2003, when Apple released the Power Mac G5, whose PPC970 chip was made by IBM, how IBM made a huge fuss about their new $3B fab in upstate NY. (And to this day, that’s also a contract fab for fabless semiconductor companies.)

Intel used to be the largest (by revenue) semi fab. They are now surpassed by Samsung.
IBM is no longer in the semi fab business. They essentially gave their fab business away to Global Foundries.
TSMC is the largest fab for fabless companies.
Intel is the only remaining competitive semi fab in the US but is quickly losing its technical leadership in 7 and 5 nm geometries of which TSMC and Samsung are the current leaders.
The remaining semi fabs in the US are trailing edge.
So I would answer the original question with "No, it does not seem likely that the US will have a leading edge semi fab, either new or Intel."

The question was not size or market share, nor whether they are leading edge, but whether any consumer product fabs are likely to be built. I provided evidence that that this is not only likely, but is in fact still happening. I question the premise that it isn’t happening.