Does Intel still sell equipment to TSMC?

Although returning chip production to factories is strategically important to Intel, the company acknowledges that TSMC is its strategic partner, producing some of Intel's most important products, such as the Core Ultra 200V (codenamed Lunar Lake) processors. Intel CEO Pat Gelsinger emphasized in an interview with Yahoo Finance that Intel even provides equipment to TSMC.

“TSMC is a great company, they serve their customers well, and they serve us well,” Gelsinger said. “The AI ​​PC [platform] Lunar Lake that we talked about earlier, we couldn’t have done it without TSMC. […] Having said that, we are a customer, we are a competitor, and we also work with TSMC on industry standards, advanced packaging using TSMC and Intel chips, which is critical to our collaboration.”

Gelsinger continued: “In the end, I provided them with some advanced equipment. It was a complex relationship that was important to Intel, to TSMC, and to the industry. I appreciate the many things we did with them and for them. I look forward to this relationship continuing for many years.”

帕特·基辛格提到的先进设备是IMS Nanofabrication多电子束光掩模写入工具，这在EUV时代至关重要。使用多束电子束写入光掩模可显著加快光掩模的生产速度，这是一项关键的进步，因为使用EUV光刻技术，光掩模的退化速度会更快。

In addition, these multi-beam tools enable manufacturers such as Intel and TSMC to quickly fine-tune masks, which helps increase yields, reduce performance variation, and improve overall performance. In the future, these tools are expected to become essential tools for producing masks for new technology nodes using high-NA EUV lithography.

One of the reasons Brian Sozzi of Yahoo Finance asked Pat Gelsinger about the relationship with TSMC was a Reuters report that claimed Intel had lost discounts at TSMC due to comments made by its CEO, severely damaging the relationship between the two companies.

Intel reportedly struck a favorable deal with TSMC, which offered a deep discount, charging $23,000 per wafer using 3nm-class manufacturing technology, about 40% less than Intel. TSMC was offended by Kissinger’s comments about Taiwan’s vulnerability and his suggestion that TSMC was not a stable manufacturing center.

Over the past few years, there has been a high level of anxiety among U.S. policymakers about the need for domestic advanced fab capacity in the U.S. The media has been filled with all the stories about how the U.S. is falling behind, and how we are no longer able to produce state-of-the-art chips. This anxiety stems both from geopolitical concerns about competing with China and from fears that the U.S. has somehow lost the ability to produce chips, just as we seem to have lost the ability to make many things.

We believe this issue is more of an economic than a technical one.

In short, Intel can produce cutting-edge semiconductors today, they just can’t do it economically. They have the technology and the equipment. Their shortcoming is simply that they can’t yet produce 18A, 3nm chips profitably.

Imagine everyone’s favorite fear-mongering scenario — what would happen if the U.S. lost access to TSMC’s factories in Taiwan? War, blockade, alien invasion — whatever the reason. Beyond that, imagine the U.S. facing a military conflict where the lack of cutting-edge chips becomes a serious national security issue.

This would obviously cause huge disruption to the economy, but how long would it take Intel to get its process up and running? It's reasonable to assume that the government could throw enough money at Intel to get its process up and running quickly.

At first, Intel's yields would be terrible, and the government would actually make up for those losses - buying wafers with only 5 or 10% good chips at a fixed price. In the semiconductor industry, volume speaks volumes, and in times of need, they could afford all the bad wafers to gain enough experience to increase production.

Obviously, this is not a scenario anyone wants to see, but we mention it here because it reflects an important but widely overlooked reality today. The conventional narrative about TSMC’s rise to success is that Intel missed the boat in mobile, and TSMC became the foundry of choice for mobile phones, which drove Intel’s production.

With that scale, they were able to learn faster than others and win process leadership over time. All of this is true, but it misses a key factor. During that time, TSMC enjoyed a lot of subsidies. The most famous of these were direct subsidies from the Taiwanese government, which enabled them to import wafer manufacturing equipment in the early days. But an even bigger subsidy was an indirect subsidy - the undervalued New Taiwan Dollar.

The New Taiwan Dollar is ostensibly a free-floating currency, but as the chart above shows, it appears to have been pegged to the US dollar for over 20 years. Economist Brad Setser has written extensively about the mechanisms Taiwan uses to achieve this (TL;DR – commercial banks, then life insurance companies), essentially managing the currency to keep Taiwan competitive.

The US dollar has appreciated significantly during this period, and the New Taiwan dollar has actually depreciated. The Economist's Big Mac Index is the best proof of this.

This is a handy way to show how undervalued or overvalued a currency is relative to the U.S. dollar. As this chart shows, the Taiwan dollar has steadily fallen against the U.S. dollar over the past two decades as TSMC has risen. We understand that this decline actually began with the Asian financial crisis in 1998 and has only widened over time.

This provides TSMC with a huge indirect subsidy. They can pay their employees wages that are competitive in Taiwan, but are actually much lower than their US competitors.

TSMC's revenues are denominated in U.S. dollars, but its employees, who are the company's real assets, are paid in New Taiwan dollars. This discount has been accumulating for decades. We tend to think that currency depreciation provides a way for exporters to undercut foreign competitors by offering lower prices. Instead, TSMC uses the effects of this currency suppression to invest in its own talent pool.

We have read studies that suggest the TWD is actually undervalued by about 30% (using a more comprehensive methodology than the Big Mac Index, which suggests levels are over 50%). Therefore, we do not think it is a coincidence that TSMC has indicated that costs at its US factories will be 20%-30% higher than wafer costs at its Taiwan factories.

To be clear, we are not disparaging TSMC's technical talent. They have tremendous capabilities and unparalleled human capital. But we think it is important to understand how they achieve this. The real strong move by TSMC management would be to invest their monetary advantage in their own talent pool rather than squandering it on broad acquisitions that are not tightly tied to the core business.

This all begs the question: What can anyone do to compete with this?

The U.S. government is fully aware of the current state of Taiwan’s currency and has consistently refused to take any action on it. Competing in the foundry space requires an external pool of capital. For Intel, this further fuels calls for direct government support.

We would obviously prefer a more commercial solution of obtaining investment from potential customers, but there are growing calls for the US government to "save" Intel by subsidizing some equity holders.

The less obvious question is what Samsung will do with it. The won is effectively on a par with the Taiwan dollar for similar reasons, but Samsung chose to spend its windfall elsewhere.

Samsung Foundry may need support from the Samsung Group in the short term. It is unclear how interested the group is in supporting the foundry. The memory business is important to the entire group. Will the foundry jeopardize this business? Will the cash burn of the foundry be so great that the group will eventually decide to take a back seat and focus on the memory business?

It's clear that both Intel and Samsung will need a lot of external support to continue to be competitive.

END

Focus on the trends and developments of the global semiconductor industry