Intel is "saved" -- US will continue to manufacture computer chips

Tom's Hardware ^

[powerset]

Intel late on Tuesday [Mar 2026] announced that its board of directors had elected Dr. Craig H. Barratt as independent chairman, who will assume his role in mid-May after the company's Annual Stockholders’ Meeting. Craig H. Barratt, who has an engineering background, will replace Frank D. Yeary, who will retire from Intel's BoD after spending around 17 years there.

Barrat replaces Frank D. Yeary, who has a financial background and who once tried to split Intel into products and manufacturing companies and then get rid of the company's manufacturing assets. Instead of splitting Intel, Craig H. Barratt seems to envision the company as an integrated devices manufacturer with rigorous execution. .....

---

TOPICS: Business/Economy; Computers/Internet
KEYWORDS: apple; governmentchips; intel; manufacturing; tesla; tsmc

[Dr. Sivana]

I remember when Japan thought it wanted to get in the supercomputer business, when the country was in its ascendancy in the 1980s. A big government program threw a lot of resources at a big program, and the technology leapfrogged before they came out with the first product.

I just got laid off from one of those non-cutting edge companies I had mentioned earlier (and make no mistake: EVs, drones, and missile systems use our non-cutting-edge chips. Programming is as important as small size and low-energy use.) I was not up on design or manufacturing, so I do not pretend to be up on the entire industry.

Other than clearing the field of regulations, and maybe tax incentives at the state and local levels to encourage new plants (e.g. tsmc), and tariffs to encourage locating HERE, I do not know what we CAN do. I do not believe the political will is there to do it all despite the president’s efforts, especially if we are going to try to be cutting-edge in general AI as well, which tends to suck the oxygen out of the room of EVERYTHING.

[Dr. Sivana]

How many CCP moles are embedded in that company? You really can’t rely on Taiwan.

Right. And Intel doesn't have the tech chops anymore. I don't think pouring money into Intel will solve the problem, though it might keep a bad problem from getting worse.

[Pete from Shawnee Mission]

ASML Holding N.V. (originally standing for Advanced Semiconductor Materials Lithography) is the company in the Netherlands which is the major supplier of semiconductor lithography processing equipment. Wikipedia has a good article about the company.

TSMC is developing a fabrication facility in Phoenix. (Pretty certain that in the event of invasion the Taiwanese facilities would destroyed to prevent their takeover.) I heard that they were having problems with Staffing. Differences in management style. Helium is required for some semiconductor clean room processes, and they are closer to U.S. sources of helium.

A bit of a slide. There is a U.S. semiconductor materials company in Denver that does not require rare earths or other exotic materials. Its electro-optical polymer product can be incorporated without retooling Semiconductor fabrication facilities (except for a final encapsulation step) and has the ability to reduce Data Center power usage and cooling costs. Current focus is on Data Centers, but they anticipate other uses in general electronics, aviation, and Areospace.

While they are a Tech R&D, they have agreements with Semiconductor Manufacturing companies and is moving from phase 3 development to Stage 4. (The tech works, need to move from final development into actual production.)

https://www.lightwavelogic.com

Lightwave Logic, Inc. (Nasdaq: LWLG) is a technology platform company, leveraging its proprietary technology to develop next-generation Electro-Optic – “EO” – polymers which increase the efficiency of internet infrastructure by allowing more data to be transmitted at significantly higher speeds and with less power than existing solutions."

https://tradertimes.com/lists-and-rankings/pivotal-points-en/pivotal-point-tracking-lightwave-logic-lwlg-efficiency-turbo-for-nvidi-68079.html

>>> “ Lightwave Logic is advancing efficiency in data centers with its patented "Perkinamines" platform. On March 12, a positive progress report on commercial licensing agreements and the confirmation of scalability for 800G and 1.6T data transmission triggered a surge in stock prices. The company develops organic polymers that convert electrical signals into light pulses significantly faster and with much lower energy consumption than conventional solutions. As demand for bandwidth in the AI sector explodes, the technology is now coming into focus for mass production.

The company is massively benefiting from the extensive expansion of AI infrastructure by hyperscalers like Microsoft and Nvidia. As traditional semiconductor materials reach physical limits at extremely high transmission rates, Lightwave Logic offers a cost-effective retrofitting solution for existing silicon photonic platforms. The polymers enable a drastic reduction in operating costs in server farms through lower power requirements and reduced heat generation. With the transition from research to commercialization, Lightwave is addressing a multi-billion-dollar market for optical transceivers. “

[powerset]

I was talkng about state-of-the-art computer chips (CPU, GPU, NPU etc.).

From DDG search:

[Texas Instruments] focus is on developing analog chips and embedded processors, which account for more than 80% of its revenue. TI also produces digital light processing (DLP) technology ...

NXP Semiconductors N.V. is a Dutch semiconductor manufacturing and design company with headquarters in Eindhoven, Netherlands.

Onsemi is a leading supplier of discrete and power modules, power management, sensors, wireless connectivity, signal conditioning, motor control, custom and ASSP, interfaces, timing, logic and memory products. Explore their products by technology, market, solution and design.

[powerset]

Nvidia does not *manufacture* any chips (never has, AFAIK). It designs them and they are manufactured by TSMC (Taiwan) and occasionally Samsung (South Korea).

[powerset]

Micron makes memory chips, not CPU (etc.) chips. Currently most of their manufacturing is outside the US.

[Dr. Sivana]

NXP Semiconductors N.V. is a Dutch semiconductor manufacturing and design company with headquarters in Eindhoven, Netherlands.

They manufacture stateside down the street from me in Chandler, AZ.

[Zathras]

I was working at Intel when they were fighting 10nm on 193nm lithography.
Yields were close to zero on way too many wafers

[dennisw]

Intel 10nm chips can do more than enough for 90% of computer users. I mean 11th and 12th gen. Yes I do understand the pursuit of lower and lower nm chips by TSMC Intel and others because they run cooler with less current draw. And for AI.

[Spktyr]

Yup, heard about that. From the outside, it appeared the main reason Intel was having so many problems was due to Not Invented Here syndrome. They did grudgingly buy that 2nm process that IBM developed but that didn’t go much of anywhere fast, and now people are doing sub 2nm pilot runs while Intel hasn’t got a working 2nm run of any kind.

[Spktyr]

It’s not the desktop or laptop consumer’s CPU that’s the issue. It’s the GPU that is, even in consumer products. CPU more or less got ‘fast enough’ for Intel around 9-10th gen. (Though AMD and Apple both have blown *way* past Intel since, they use TSMC as their fab.) GPU... that’s where a lot of market demand is at and consumers need/want the power.

In military and AI applications... we need all the power we can get.

[dennisw]

Thanks. You are right lower and lower nm is more crucial for discrete gpu for gaming and AI done with desktop GPU video cards. Also for chips that go into AI data centers. Now google has its own AI chips Tensor/TPU that are competitive with Nvidia AI chips for use in data centers. TSMC made.