Late last year Intel announced massive plans to increase its global chip production capacities by upgrading, expanding, and equipping four of its fabs. As it turns out, the company is not going to stop there and is considering to build a brand new fab in Israel, which will cost $11 billion.

Moshe Kahlon, Israeli Finance Minister, on Tuesday confirmed that Intel had applied for a grant of about $1 billion from the government for its new investment plan to build a fab that would cost $11 billion. If everything goes as planned, the investment will not only be the biggest of its kind in Israel, but Intel will also build the largest semiconductor fab in the region. Furthermore, the expansion will add 1,000 new employees to Intel’s staff of approximately 13,000 in Israel.

“The moment the company comes to Israel and invests $10 billion, and it receives a grant of 9%, that means 91% of it stays here,” said Mr. Kahlon in an interview on Army Radio. “There are always such discounts, there are always incentives.”

Intel and Ministry of Finance reportedly started talks about the new fab several weeks ago, so the production facility is a few years away. In addition to securing incentives from the Israeli government, Intel will have to apply for permissions from various local authorities. That said, peculiarities of the fab as well as process technologies that it will use remain to be seen.

Since Intel is already investing in two fabs in the U.S. (Fab 42 in Arizona and D1X in Oregon), a fab in Ireland, and a fab in Israel, it is unlikely that Intel needs to start building another fab in Israel as soon as possible.

In addition to manufacturing operations, Intel also has rather vast R&D operations in Israel. The company develops CPU microarchitectures, Mobileye advanced driver-assistance systems (ADAS), RealSense 3D cameras, communication, and other technologies in the country. The addition of an advanced fab will clearly increase significance of Israel for Intel.

Related Reading

Sources: Globes, Reuters

Comments Locked

37 Comments

View All Comments

  • eva02langley - Tuesday, January 29, 2019 - link

    I wonder how Intel, in this moment, think having more fabs is a good thing.

    1. Their process is behind.

    2. The cost of new process is only going to explode.

    3. They are not TSMC, process node is not their BnB.

    Yes, they have money right now, but what in 10 years?
  • jtd871 - Tuesday, January 29, 2019 - link

    They are apparently selling everything they can produce. Alot of that stuff is reportedly high-margin, so a good way to maximize profits is to make more stuff.
  • iwod - Tuesday, January 29, 2019 - link

    1. Assuming their 7nm is on schedule , they are not far behind, and if needed, roughly the same time as TSMC for HP node. ( Although I doubt it because if I remember correctly Intel has yet to put down / increase orders for their EUV Scanners, and I bet Samsung and TSMC will be grabbing most of it )

    2. The cost of new process is going to increase whether they sell the same amount of CPU or even more with GPU. They need capacity. ( And they were again very late to increase capacity )

    3. Process node are their BnB. Intel has CPU to continue to fund the Node development, Samsung has NAND and DRAM to fund their Custom Fab, TSMC is just funding it via their own business.

  • Santoval - Wednesday, January 30, 2019 - link

    Regarding #1, if Intel's 7nm node is as dense as they have announced (~238 million transistors per mm^2) it will be quite denser than TSMC's 5nm node (I have read numbers around 180 MTr/mm^2), perhaps around as dense as TSMC's 3/4nm node or a bit less dense than that.
    I strongly doubt Intel will release 7nm based CPUs before 2022 though, by which time TSMC should have its 3/4nm node ready for orders, assuming high volume release of their first gen 5nm in 2020 and their second gen 5nm+ in 2021.

    By the way, at Intel's 7nm node and at TSMC's 3/4nm node (and at Samsung's 2/3nm), apart from going smaller it is also expected that GAA (Gate All Around) tech will be adopted, and possibly even III-V semiconductors. The latter is still not likely, but GAA-FET transistors are apparently a certain bet.
  • HStewart - Wednesday, January 30, 2019 - link

    This is why I dislike the nm rating - because it only realistic base on who makes the node process - I know it more than just density of transistors - it technically is judge also by gate design.

    CPU actually depends also on more than just node also - but how the architexture of cpu is designed - like execution units and such.
  • Jorgp2 - Tuesday, January 29, 2019 - link

    I mean Intel is cutting consumer orders to be able to fill high-margin orders.

    They aren't AMD
  • HStewart - Tuesday, January 29, 2019 - link

    High margin orders must be laptop creators like Dell, Lenovo, HP, Asus and many others.

    Just take a trip into you local BestBuy and go to laptop section.
  • fazalmajid - Wednesday, January 30, 2019 - link

    No, they can’t even fill all the server Xeon orders, despite that being much lower volume than laptop or even desktop CPUs
  • HStewart - Wednesday, January 30, 2019 - link

    What proof do you have on this - this could mean just the demand is higher than supply and will indicated why it needs to spend $11 billion on new Fab. That does not sound like a bad thing but a good thing.
  • Wilco1 - Sunday, February 3, 2019 - link

    They have several 10nm fabs producing zero chips for several years when they should have been producing hundreds of millions. If 10nm actually worked then they wouldn't be in this terrible situation. If you look at the news, Intel looks like a head-less chicken: fired their CEO, cancelled many product lines, still don't have 10nm working, and released nothing new apart from a few overpriced overclocked CPUs for crazy fanboys. And you believe all this is a good thing?!?

Log in

Don't have an account? Sign up now