This happened in the late 1980s. Gender perspective needs to be developed, or and/or and other agencies, but his broad point is very relevant in 2024.
Fabric is a semiconductor manufacturing company. These gigantic structures require billions of dollars and several years to build. They are incredibly difficult to get right.
Intel has been the best at this for decades. Then, around 2018, its lead began to collapse, due to a series of mistakes. TSMC, from Taiwan, is steadily developing and is undoubtedly the best chipmaker in the world.
Bloomberg’s chip reporter Ian King wrote about this in late 2018. Since then, Intel has fallen and TSMC has risen. Intel now has a capital of less than $ 100 billion. It doesn’t even break the top 150 companies based on this rating. TSMC has a net worth of about $1 trillion these days, placing it in the top 10.
Why countries must be strategic now
This is a dramatic fall from grace for Intel, and a major strategic and political problem for the US.
If you want the best chips, you have to go to Taiwan to make them. Or perhaps South Korea, where Samsung has built an impressive so-called core business that makes semiconductors for other companies.
Many of the famous chipmakers we think of in the US today actually don’t make chips. Nvidia, Qualcomm, AMD, and the rest of the chip designs, and often have TSMC making them. Apple and many other big tech giants also have TSMC making the chips they design in-house.
Also, producing these complex products in large quantities, without minor defects, is very difficult.
If China invades Taiwan, or somehow controls this island nation, this will be a disaster for the US, and Europe for that matter. No faster iPhones for a while. Nvidia will have to look elsewhere to make its own GPUs. AI development can stop.
That’s why, today, “real worlds with fabrics” is a better phrase. Chips power the modern economy. If you have to source these parts overseas, you are at risk.
Qualcomm can’t make up for Intel’s massive decline
That’s why Intel’s decline is so worrying. It is the only company in America that knows how to make chips that are powerful in size. (Globalfoundries also makes semiconductors, but not advanced ones).
The WSJ reported on Friday that Qualcomm approached Intel recently about a deal. That followed a report from Reuters two weeks ago that Qualcomm was considering buying parts of Intel.
If the deal happens, it won’t solve America’s chipmaking problem. One analyst called the possibility of a tie-up “unusual.”
Qualcomm probably isn’t interested in Intel’s manufacturing process. As Reuters reported, it is interested in some of the chip-design process.
Intel has two major businesses. For one, it creates semiconductors for PCs, data center servers and other devices. Another part is making these plans.
For decades, Intel’s design and manufacturing processes were tightly integrated. This has worked well for many years. The company can set up its factories to the exact specifications of its in-house designers.
Then, the world began to go in a different direction, which was opened by TSMC. Instead of designing and making your own shifts, why not run fabrics and make shifts for other companies?
In the late 1980s, when TSMC started, this idea was laughed at. (Hence the sexist comments from AMD’s Sanders about real men with cloths).
But TSMC’s approach continued slowly, aided by Intel’s mistakes and the march of modern technology.
The big one happened when Intel missed out on making chips for the iPhone when that revolutionary device first came out. Apple ended up going with TSMC. Qualcomm is a major smartphone developer as well, and has TSMC making most of these components. Other chip makers, including AMD, also began switching to TSMC.
This gave TSMC a large, diverse range of products that it needed to learn how to make chips better than anyone else. Ian King’s 2018 article explains this concept well:
With billions of transistors on chips, a problem with a small number of tiny switches can render the entire circuit useless. A production run can take six months and involves hundreds of steps that require maniacal attention to detail. Whenever something goes wrong, the factory operator has the opportunity to make tweaks and try a new method. If the change works, that information will be saved to try the next challenge. The longer the production, the better. And TSMC has the most these days.
While TSMC was learning from this range of big customers, Intel’s manufacturing process was held back by one customer: itself.
When smartphone chips became the biggest game in town, Intel didn’t have the volume to keep up with TSMC in the manufacturing race. AI has made this situation worse. Nvidia leads here by a mile, and TSMC makes its own GPUs.
A stubborn miasma of productivity
Raising the manufacturing miasma around Intel will be an expensive, risky and complex endeavor.
Intel has already started paying TSMC to make some of its chips. With that as a starting point, there is a long way to go.
The American company took a big step recently when it separated its foundry business from its chip-design business. This enables external customers to trust Intel to make their own chips without having to compete with them.
The next challenge is the biggest: In fact, it is good to make chips as well.
Intel’s Foundry business won’t be able to challenge TSMC until it gets more big customers. Also, to be able to make chips you need large, different volumes in order to see errors, correct processes, and feed that knowledge into your fabrics.
It’s a chicken and egg situation. Without large volumes, external customers will be afraid to have Intel make their own key chip designs. But without customers, Intel will not improve.
Does everyone love Raimondo?
One way to end this disruption: Get the US government to force other companies to use Intel fabrics. This is exactly what is happening now, according to CNBC.
Secretary of Commerce Gina Raimondo has been trying to encourage shareholders in companies such as Nvidia and Apple to recognize the economic benefits of having a US facility that can produce AI chips, CNBC reported recently.
About 4 days later, Intel announced a partnership where it will make an AI chip for Amazon Web Services. Intel stock was in the news because AWS is the largest cloud provider and designs many of the chips that will be used in its big data centers. This is the kind of quality Intel needs.
18 A hope
On the technology side, Intel has a new processor node called 18A. This is a set of rules for chip design and the accompanying manufacturing process. If all goes well in the coming years, this could make Intel compete with TSMC’s flagship nodes again.
The AWS partnership is based on this 18A technology, and Microsoft said earlier this year that it will have an in-house chip made in this area of operation, too.
Intel’s Foundry business needs just these customers. And the 18A technology has to work well for that to happen.
“18A would be a great bet as the company bets on its success,” Bernstein chip analyst Stacy Rasgon wrote in a recent note to investors.
Can separation work?
In the meantime, Qualcomm is unlikely to seek to acquire this part of Intel.
This raises the specter that Intel is broken, something that has been speculated in recent months. Reuters says that Qualcomm is interested in some of the chip-design process, while the WSJ reported on Friday that Qualcomm may sell some of Intel’s parts to other customers.
How would Intel’s Foundry business function as a separate company, separate from the design divisions?
The problem here, too, is about volume. Currently, Intel’s Foundry business needs the amount of production it gets from the company’s chip designs. Besides, it may have very little input to learn from.
“We don’t see how it can be done now,” Rasgon wrote in an early September letter. “The production arm cannot sustain itself now because of the heavy losses and the lack of scale.”
“The division really makes sense as soon as it attracts a large third-party business that looks like it will be years (if it ever happens),” he added.