You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!
New Intel CEO Pat Gelsinger is not wasting any time in changing the course of the largest semiconductor company the world has ever seen. Today he announced the IDM 2.0 strategy which will better leverage Intel’s manufacturing abilities. There is a lot to talk about here but let’s focus on the...
The Intel announcement felt like an unexpected seismic event in the semiconductor industry to digest. Would love to get hear everyone's thoughts. In the meantime, here are my thoughts reading various article(s) throughout the Internet:
Intel Good News
Intel 7nm. Its progressing well & Intel is expected to start making its first 7nm CPU in 2Q21!
New research collaboration with IBM
Intel Foundry Services seems to have a lot of support (Qualcomm, Microsoft, Amazon, Google, etc..)
$20 billion investment in two new fabs in Phoenix
Planned yearly cadence of manufacturing updates to be more competitive
Big Questions
Will Intel be able to snag enough ASML EUV machines?
Can Intel be able to catch up or beat TSMC/Samsung on bleeding edge node technology?
How many of TSMC's/Samsung's customers can Intel snag?
TSMC Good News
Expanded business with Intel - Will produce Intel CPU cores using an unspecified node process in 2023
Intel 7nm will still be a generation behind TSMC 3nm
Bad News
A new competitor with a big wallet explicitly targeting its BIGGEST and MOST IMPORTANT customer (Apple)
Samsung Foundry Good News
Expanded business with Intel - Will produce Intel products in 2023
Bad News
Foundry relationship with IBM may be in jeopardy?
Intel 7nm may be a better process than Samsung 3nm
A new competitor with a big wallet
INTC announcement is great news for my
CPU intensive Verilog simulator. I thought
high performance was being abandoned
for low power. Did I hear that INTC is
now buying its R&D from IBM?
INTC announcement is great news for my
CPU intensive Verilog simulator. I thought
high performance was being abandoned
for low power. Did I hear that INTC is
now buying its R&D from IBM?
The high performance computing is and will be there always. The difference is that the way to achieve that has become diversified. Intel is just one of the solution provider among many.
I don't see Apple giving any significant business to Intel. Or Nvida, or AMD for that matter.
I can see Microsoft, Amazon, and Google working with Intel and having them manufacture custom chips for datacenters based on x86. But they pretty much already do this.
I can still this helping Intel keep business with cloud providers that they would otherwise lose, that's about it.
Not only Intel's Capex is lagging behind but their old equipment for old process nodes can't continuously produce more revenue and profit. It's because nobody will buy a new PC with an Intel CPU made last month based on a seven-year old process technology.
On the foundry business side, TSMC keeps producing chips using both old and new technology. In 2020, TSMC got 41% of their revenue from process nodes older than or equal to 28nm.
Intel has to phase out old process technology and old fab equipment while TSMC continuously collect profit generated from old equipment (some already fully depreciated). At the same time TSMC invests more on leading edge fabs than Intel does.
In the long run Intel and TSMC will become two drastically different companies in terms of capacity, capability, and revenue.