Pumping the oil of the 21st century: TSMC versus INTEL

[user nl]

It is so easy these days to have AI make some historic financial overview of business performances. Below is a table generated by Google's Gemini
summarizing some key numbers for TSMC and INTEL, and it is really amazing when you compare these numbers.

It will be interesting to see what will happen when during the next 3 years 2026-2028 TSMC will "flood" the semi-system with some >150 B$ in capex and how their Fabs/nodes (2, 3, 5, 7 etc) will become (and are already) enormous cash printing machines.

TSMC may spend more in Capex in 2026 than the total (expected) revenue of INTEL this year.

Of course, INTEL and SAMSUNG will pump some overflow "made-in-USA-oil" orders from customers because of the lack of capacity at TSMC. But once that wafer and advanced packing capacity of TSMC is "on the market" around 2029, it is hard to see how INTEL Foundry will be an economical competitor to TSMC.

If we assume a (conservative) CAGR of 25% during 2026-2029 the expected revenue of TSMC will be 300 B$ and if they keep their net profit margin of 45-50% they will generate a net profit of some 150 B$ in a single year.

TSMC will be a one-member "OPEC" for the best (performance/watt) oil of the 21st century. Amazing story.........

 

[hist78]

It is so easy these days to have AI make some historic financial overview of business performances. Below is a table generated by Google's Gemini
summarizing some key numbers for TSMC and INTEL, and it is really amazing when you compare these numbers.

It will be interesting to see what will happen when during the next 3 years 2026-2028 TSMC will "flood" the semi-system with some >150 B$ in capex and how their Fabs/nodes (2, 3, 5, 7 etc) will become (and are already) enormous cash printing machines.

TSMC may spend more in Capex in 2026 than the total (expected) revenue of INTEL this year.

Of course, INTEL and SAMSUNG will pump some overflow "made-in-USA-oil" orders from customers because of the lack of capacity at TSMC. But once that wafer and advanced packing capacity of TSMC is "on the market" around 2029, it is hard to see how INTEL Foundry will be an economical competitor to TSMC.

If we assume a (conservative) CAGR of 25% during 2026-2029 the expected revenue of TSMC will be 300 B$ and if they keep their net profit margin of 45-50% they will generate a net profit of some 150 B$ in a single year.

TSMC will be a one-member "OPEC" for the best (performance/watt) oil of the 21st century. Amazing story.........


View attachment 4108

Great analysis! Thanks. Do you have any thoughts on how Intel can get out of this downward spiral?

Dominant does not necessarily mean monopoly. TSMC’s business practices are far more aligned with a “global collaboration model” and far less with anything resembling a monopoly. TSMC’s pure‑play foundry model is both more efficient and more profitable than Intel’s IDM model. While TSMC has enjoyed strong revenue growth and high profitability, its major customers, such as Apple, Nvidia, AMD, Qualcomm, MediaTek, and Broadcom, have also thrived. In contrast, although Intel’s profits were once strong, its major OEM customers (Dell, HP, Lenovo, Asus, Acer) have struggled for years with razor‑thin net margins of just 1% to 5%.

IMO, Without changing its IDM business model, Intel will continue to struggle with net losses and shrinking revenue. IFS can't save Intel.

 

[Daniel Nenni]

Great analysis! Thanks. Do you have any thoughts on how Intel can get out of this downward spiral?

Dominant does not necessarily mean monopoly. TSMC’s business practices are far more aligned with a “global collaboration model” and far less with anything resembling a monopoly. TSMC’s pure‑play foundry model is both more efficient and more profitable than Intel’s IDM model. While TSMC has enjoyed strong revenue growth and high profitability, its major customers, such as Apple, Nvidia, AMD, Qualcomm, MediaTek, and Broadcom, have also thrived. In contrast, although Intel’s profits were once strong, its major OEM customers (Dell, HP, Lenovo, Asus, Acer) have struggled for years with razor‑thin net margins of just 1% to 5%.

IMO, Without changing its IDM business model, Intel will continue to struggle with net losses and shrinking revenue. IFS can't save Intel.

I do not see Intel reclaiming the glory days of the x86 monopoly but I do see the possibility of them being a thriving business. x86, foundry, and ASIC business. It would be nice if Intel can get in on the AI gold rush but the window is closing for that.

Can Intel Foundry succeed? Yes, definitely. The NOT TSMC market is ripe for the picking and Samsung is still floundering. There is no trust there after many years of claiming to be first only to not yield.

By the end of 2026 you will see some big customers lined up to use IFS, absolutely.

 

[user nl]

Great analysis! Thanks. Do you have any thoughts on how Intel can get out of this downward spiral?

Dominant does not necessarily mean monopoly. TSMC’s business practices are far more aligned with a “global collaboration model” and far less with anything resembling a monopoly. TSMC’s pure‑play foundry model is both more efficient and more profitable than Intel’s IDM model. While TSMC has enjoyed strong revenue growth and high profitability, its major customers, such as Apple, Nvidia, AMD, Qualcomm, MediaTek, and Broadcom, have also thrived. In contrast, although Intel’s profits were once strong, its major OEM customers (Dell, HP, Lenovo, Asus, Acer) have struggled for years with razor‑thin net margins of just 1% to 5%.

IMO, Without changing its IDM business model, Intel will continue to struggle with net losses and shrinking revenue. IFS can't save Intel.

No idea.

It is kind of interesting that the USA, because of national security reasons, has decided to subsidize its national "21st century oil" producer for the coming decade(s ?). I think this is a bi-partisan decision, see the Chips-act and the 10% stake.

So, for as long the USA-China great power battle continues INTEL Foundry will be kept alive by INTEL Products, the US administration and some other big companies like NVIDIA, Microsoft, Apple that will provide IFS with some small orders the coming 5-10 years. Enough to keep IFS from starving to death, but probably not enough to have IFS reaching ever 50% gross margin on leading edge nodes.

How IFS at some point will finance new fabs and tools, CEO LBT has made it clear he will not ramp 14A when there is no outside commitments for capacity. So, perhaps the US administration can use all these tarrif and "made-in-USA" rules, that they have already developed, to force NVIDIA, Microsoft, Apple, Google, AMD to commit enough wafers to IFS to subsidize them enough to ramp 14A in 2028-2030.


At the same time TSMC can not complain: they got Chips-act money, cheap loan, 35% tax break and loan guarantees by the Taiwan government in the 250 B investment treaty with US. In total it means all the Arizona investments of 165 B$ by TSMC are also heavily subsidized, and they can import chips from Taiwan for 0 tariff for their US customers as long as the Arizona Giga-campus is under construction. So, all this money for TSMC will help them as well to grow their gross margin and net profit margin.

So, no wonder TSMC feels so confident that they'll manage for the foreseable future (2029 and further) to keep their gross margin (well) above 56%, in 2026 probably already around 63-65%.

As said an amazing story of a cash printing quasi-monopolist, also supported by the US government...........I'm sure there will be many academic papers and books written in 5 to 10 years about how the US got TSMC "striking oil" in Arizona to start their big "oil-production" facility.........;

--------------------------------------------------------------------------------------------------------------------------------------------
Here is what Gemini states about the TSMC support by the US government:

Total Package Value: When combining the $11.6 billion in direct cash/loans with the projected $57.7 billion in tax credits (35% of $165B), the U.S. government is effectively subsidizing roughly 42% of the first three phases of the Phoenix site.

The 35% Tax Break: Why It Changed​

Originally, the CHIPS Act (Section 48D) offered a 25% Investment Tax Credit. However, in mid-2025, Congress passed the BASIC Act (Building Advanced Semiconductors Investment Credit), which TSMC and other manufacturers lobbied for to offset the higher costs of U.S. labor and energy.

• The Incentive: The credit was raised to 35% for projects that commenced construction or expanded production capacity before December 31, 2026.
• The Financial Impact: For TSMC’s current $165 billion investment roadmap, this 10% increase represents an additional $16.5 billion in tax savings over the life of the project.

 

[Daniel Nenni]

Remember, Intel is also strong in packaging. Intel Foundry can get a chiplet and packaging order since TSMC does not package foreign die. Hopefully IFS will get full chip orders but chiplets at IFS seem to be a no brainer especially if you need packaging to go with it.

 

[MKWVentures]

Remember, Intel is also strong in packaging. Intel Foundry can get a chiplet and packaging order since TSMC does not package foreign die. Hopefully IFS will get full chip orders but chiplets at IFS seem to be a no brainer especially if you need packaging to go with it.

do you think Intel is stronger than ASE or Amkor in Packaging? If so, why is there no revenue from people using packaging

 

[Daniel Nenni]

do you think Intel is stronger than ASE or Amkor in Packaging? If so, why is there no revenue from people using packaging

Yes Intel is stronger. Intel is in the lead with BSPD for example. Others are following but Intel is the leader. Packaging revenue will come but it will not be big margins. There really needs to be wafers packaged in there. Again, this is the NOT TSMC packaging market.

 

[Artificer60]

I've heard Intel claim EMIB-T should be a real differentiator for packaging as well. That is due out in the second half of this year.

 

[hist78]

Can Intel Foundry succeed? Yes, definitely. The NOT TSMC market is ripe for the picking and Samsung is still floundering. There is no trust there after many years of claiming to be first only to not yield.

The “NOT TSMC” market is shrinking or even disappearing day by day, partly thanks to Trump.

When Trump forced TSMC to commit $250 billion to build fabs and advanced packaging facilities in the United States, that decision effectively reshaped the competitive landscape and severely limited the future of Intel Foundry Services for external customers.

Assuming this $250 billion will be spent over the next eight years, and considering that TSMC has already invested about $40 billion in its Phoenix, Arizona projects, TSMC will be spending an average of at least $26.25 billion per year in the US alone over that period.

At the same time, TSMC will continue to invest even more CapEx in Taiwan over the next eight years, building additional capacity beyond what it is constructing in the United States.

As a result, TSMC’s customers will inevitably support more production at TSMC Arizona along with more orders at TSMC Taiwan to reduce the overall cost and to support global growth. Their urgency or need to add Intel as a second foundry source is now much lower than before. One consistent lesson from TSMC’s history is that whenever it builds new capacity, customers must commit to orders before the expansion is publicly announced. Morris Chang often emphasized that TSMC builds fabs based on customer commitments, not speculation.

This is especially true for large customers such as Apple, Qualcomm, Nvidia, AMD, Broadcom, and even Intel. When TSMC agreed to invest $250 billion in the US, it is reasonable to assume that these companies had already committed to placing sufficient orders to justify the scale of the investment.

So where is the future of the “NOT TSMC” market? There is not much room left for IFS to grow. Thanks to Trump.

 

[hist78]

By the end of 2026 you will see some big customers lined up to use IFS, absolutely.

Do you think it will be big customers with small orders, or big customers with large orders?

 

[hist78]

How IFS at some point will finance new fabs and tools, CEO LBT has made it clear he will not ramp 14A when there is no outside commitments for capacity. So, perhaps the US administration can use all these tarrif and "made-in-USA" rules, that they have already developed, to force NVIDIA, Microsoft, Apple, Google, AMD to commit enough wafers to IFS to subsidize them enough to ramp 14A in 2028-2030.

I can assure you that if Intel or Intel Foundry Service needs this level of government intervention just to survive, similar to many state‑owned enterprises in communist countries, then Intel is effectively finished, dead.

By the way, when Trump forced TSMC to agree to invest $250 billion in building fabs in Arizona, the tariff threats and “Made in USA” requirements were effectively removed from TSMC’s and TSMC US customers' list of concerns.

 

[hist78]

No idea.

It is kind of interesting that the USA, because of national security reasons, has decided to subsidize its national "21st century oil" producer for the coming decade(s ?). I think this is a bi-partisan decision, see the Chips-act and the 10% stake.

So, for as long the USA-China great power battle continues INTEL Foundry will be kept alive by INTEL Products, the US administration and some other big companies like NVIDIA, Microsoft, Apple that will provide IFS with some small orders the coming 5-10 years. Enough to keep IFS from starving to death, but probably not enough to have IFS reaching ever 50% gross margin on leading edge nodes.

How IFS at some point will finance new fabs and tools, CEO LBT has made it clear he will not ramp 14A when there is no outside commitments for capacity. So, perhaps the US administration can use all these tarrif and "made-in-USA" rules, that they have already developed, to force NVIDIA, Microsoft, Apple, Google, AMD to commit enough wafers to IFS to subsidize them enough to ramp 14A in 2028-2030.


At the same time TSMC can not complain: they got Chips-act money, cheap loan, 35% tax break and loan guarantees by the Taiwan government in the 250 B investment treaty with US. In total it means all the Arizona investments of 165 B$ by TSMC are also heavily subsidized, and they can import chips from Taiwan for 0 tariff for their US customers as long as the Arizona Giga-campus is under construction. So, all this money for TSMC will help them as well to grow their gross margin and net profit margin.

So, no wonder TSMC feels so confident that they'll manage for the foreseable future (2029 and further) to keep their gross margin (well) above 56%, in 2026 probably already around 63-65%.

As said an amazing story of a cash printing quasi-monopolist, also supported by the US government...........I'm sure there will be many academic papers and books written in 5 to 10 years about how the US got TSMC "striking oil" in Arizona to start their big "oil-production" facility.........;

--------------------------------------------------------------------------------------------------------------------------------------------
Here is what Gemini states about the TSMC support by the US government:

Total Package Value: When combining the $11.6 billion in direct cash/loans with the projected $57.7 billion in tax credits (35% of $165B), the U.S. government is effectively subsidizing roughly 42% of the first three phases of the Phoenix site.

View attachment 4112​

The 35% Tax Break: Why It Changed​

Originally, the CHIPS Act (Section 48D) offered a 25% Investment Tax Credit. However, in mid-2025, Congress passed the BASIC Act (Building Advanced Semiconductors Investment Credit), which TSMC and other manufacturers lobbied for to offset the higher costs of U.S. labor and energy.

• The Incentive: The credit was raised to 35% for projects that commenced construction or expanded production capacity before December 31, 2026.
• The Financial Impact: For TSMC’s current $165 billion investment roadmap, this 10% increase represents an additional $16.5 billion in tax savings over the life of the project.

"At the same time TSMC can not complain: they got Chips-act money, cheap loan, 35% tax break and loan guarantees by the Taiwan government in the 250 B investment treaty with US. In total it means all the Arizona investments of 165 B$ by TSMC are also heavily subsidized, and they can import chips from Taiwan for 0 tariff for their US customers as long as the Arizona Giga-campus is under construction. So, all this money for TSMC will help them as well to grow their gross margin and net profit margin."

I believe there is an error in your table (or Google's AI generated table) and analysis.

The $250 billion loan credit guarantees are for an additional $250 billion of investment in the United States, this is not part of the $250 billion direct investment in the US (the majority is coming from TSMC). The $250 billion loan guarantees are backed by Taiwan’s government, not by the US government, and not for TSMC, or at least not totally for TSMC.

 

[hist78]

The 35% Tax Break: Why It Changed​

Originally, the CHIPS Act (Section 48D) offered a 25% Investment Tax Credit. However, in mid-2025, Congress passed the BASIC Act (Building Advanced Semiconductors Investment Credit), which TSMC and other manufacturers lobbied for to offset the higher costs of U.S. labor and energy.

This is another mistake made by Google Gemini. It fabricated a claim that, in mid‑2025, Congress passed the BASIC Act (Building Advanced Semiconductors Investment Credit). ChatGPT did not make this mistake.

The truth is that on May 5th, 2025, the bill was only introduced and referred to the House Committee on Ways and Means. As of today, it still has a long legislative process to go before it can pass both chambers of Congress and become law.

https://www.congress.gov/bill/119th-congress/house-bill/3204/text