Intel to Become one of TSM's best customers???

[Arthur Hanson]

If Intel's business unit is to become truly independent and competitive, would they not use TSM which is highly efficient and has better technology?

 

[Daniel Nenni]

I had expected Intel to be TSMC's #2 customer by 2025 but with the aggressive fab builds Pat Gelsinger is initiating I can no longer imagine it. Intel will be a big TSMC N3 chiplet customer but after that it all depends on the timing of 20A and 18A. TSMC seems to be a fail safe for Intel internal fabs.

 

[hist78]

I had expected Intel to be TSMC's #2 customer by 2025 but with the aggressive fab builds Pat Gelsinger is initiating I can no longer imagine it. Intel will be a big TSMC N3 chiplet customer but after that it all depends on the timing of 20A and 18A. TSMC seems to be a fail safe for Intel internal fabs.

From 2024 to 2025 and possible 2026, Intel will have to significantly increase products manufactured by outside foundries (such as TSMC) in order to compete, maintain its revenue, and improve its cash flow. Intel actually revealed it last week.

During the "Internal Foundry Model Investor Webinar" last week, Intel forecasted that IFS will have $20 billion 2024 revenue coming from Intel Internal needs and external fabless clients.

Let's make some assumptions:

1. Assume Intel will maintain a 50% gross profit margin for its CPUs and processors. If it gets lower than 50% gross profit margin than Intel overall revenue will be even smaller by the calculation method explained below.

2. Assume Intel will have limited external fabless clients for IFS. Let's say $5 billion for 2024.

3. From all those Intel new fab construction projects, such as Arizona, Ohio, Germany, and Israel, only Arizona can go into HVM in 2024/2025. All other fabs are still in early stage of planning and construction or will only be operational in 2025/2026 or beyond.


My Intel total revenue calculation is as follows:

Part 1: IFS revenue from external non-Intel clients: $5 billion.

Part 2: Revenue from selling processors made by IFS: $30 billion = ($20 billion (IFS revenue) - $5 billion (Part 1) ) * 2. (So Intel processors can maintain a 50% gross profit margin)

Part 3: Revenue from non-processor and non-IFS products and services, such as software licensing, consulting, selling equity in the investments or properties Intel owns: $5 billion

Part 4. Revenue from selling processors made by non-IFS foundries (like TSMC, UMC ...): $10 billion.

Intel 2024 total revenue will be the sum of Part 1 through Part 4: $50 billion.

Intel 2022 revenue was $63.05 billion and 2021 revenue was $79.02 billon. If Intel wants to bounce back to its 2022 revenue level, there is a $13 billion ($63 - $50 billion) gap needs to be filled.

Considering new fabs availability, high capital intensity ratio, and struggling free cash flow issues, Intel must sell more non-Intel manufactured products in order to improve its revenue and financial standing.

Of course this calculation is based on the $20 billion IFS 2024 revenue forecast announced last week. You can do your own calculation by applying your own numbers. With the forecasted IFS 2024 $20 billion revenue, more non-Intel IFS revenue will negatively impact Intel overall revenue due to the 50% gross profit margin effect.

 

[lilo777]

From 2024 to 2025 and possible 2026, Intel will have to significantly increase products manufactured by outside foundries (such as TSMC) in order to compete, maintain its revenue, and improve its cash flow. Intel actually revealed it last week.

During the "Internal Foundry Model Investor Webinar" last week, Intel forecasted that IFS will have $20 billion 2024 revenue coming from Intel Internal needs and external fabless clients.

Let's make some assumptions:

1. Assume Intel will maintain a 50% gross profit margin for its CPUs and processors. If it gets lower than 50% gross profit margin than Intel overall revenue will be even smaller by the calculation method explained below.

2. Assume Intel will have limited external fabless clients for IFS. Let's say $5 billion for 2024.

3. From all those Intel new fab construction projects, such as Arizona, Ohio, Germany, and Israel, only Arizona can go into HVM in 2024/2025. All other fabs are still in early stage of planning and construction or will only be operational in 2025/2026 or beyond.


My Intel total revenue calculation is as follows:

Part 1: IFS revenue from external non-Intel clients: $5 billion.

Part 2: Revenue from selling processors made by IFS: $30 billion = ($20 billion (IFS revenue) - $5 billion (Part 1) ) * 2. (So Intel processors can maintain a 50% gross profit margin)

Part 3: Revenue from non-processor and non-IFS products and services, such as software licensing, consulting, selling equity in the investments or properties Intel owns: $5 billion

Part 4. Revenue from selling processors made by non-IFS foundries (like TSMC, UMC ...): $10 billion.

Intel 2024 total revenue will be the sum of Part 1 through Part 4: $50 billion.

Intel 2022 revenue was $63.05 billion and 2021 revenue was $79.02 billon. If Intel wants to bounce back to its 2022 revenue level, there is a $13 bullion ($63 - $50 billion) gap needs to be filled.

Considering new fabs availability, high capital intensity ratio, and struggling free cash flow issues, Intel must sell more non-Intel manufactured products in order to improve its revenue and financial standing.

Of course this calculation is based on the $20 billion IFS 2024 revenue forecast announced last week. You can do your own calculation by applying your own numbers.

It looks like you are counting foundry revenues as if ("foundry business was a reseller" cross that) Intel is reselling the goods produced by the foundry. For example, Walmart may buy goods, say for $100B and sell them for $130B: there is a direct link between $100B and $130B. There is basically fixed margin that Walmart adds to the price it pays. Should not foundry business be detached from the price of the chips it facilitates manufacturing? Foundry business prices (and revenue) should depend on the price of the wafer and wafer volume but it does not determine how much the customer will be able to charge for resulting chips (well, there is some indirect link, but it's a weak one).

 

[hist78]

It looks like you are counting foundry revenues as if foundry business was a reseller. For example, Walmart may buy goods, say for $100B and sell them for $130B: there is a direct link between $100B and $130B. Should not foundry business be detached from the price of the chips it facilitates manufacturing? Foundry business prices (and revenue) should depend on the price of the wafer and wafer volume but it does not determine how much the customer will be able to charge for resulting chips (well, there is some indirect link, but it's a weak one).

I am talking about the overall Intel "revenue". Please remember IFS revenue generated from those Intel internal clients can't be counted into Intel overall revenue.

Intel stated goal is to achieve 50% to 60% gross profit margin in the future. IMO, even 50% gross profit margin is too optimistic due to various factors.

 

[MKWVentures]

Dan's comment (Intel should be TSMs 2nd largest customer) and the question "why would someone choose Intel Fab over TSMC Fab if they had free choice" are both great and accurate comments. we can show how both will be proven correct when we see what actually happens in 2025.

 

[Barnsley]

I am talking about the overall Intel "revenue". Please remember IFS revenue generated from those Intel internal clients can't be counted into Intel overall revenue.

Intel stated goal is to achieve 50% to 60% gross profit margin in the future. IMO, even 50% gross profit margin is too optimistic due to various factors.

Margins are high as competition not so easily able to utilise it like say Amazon and their "your margin is my opportunity"

Food n Beverage margins are less than 5% so require high volume.

Semicon seems relatively high volume so how come margins are not on a downward pressure?