Intel may sell part of Intel Foundry in the future - Intel at Citi 2025 Global TMT Conference

[hist78]

You may listen the Intel official webcast audio replay: https://www.intc.com/news-events/ir-calendar/detail/20250904-citis-2025-global-tmt-conference

Audio overview is generated by Google NotebookLM AI in the "Critique" mode: https://notebooklm.google.com/noteb...tifactId=666ba588-f683-40cc-adf6-6961fde46fd4

Summary below is generate by Google NotebookLM AI:

Intel Citi TMT Conference 2025 Briefing

Date: September 4, 2025

Presenter: Dave Zinsner (Intel CFO)

I. Executive Summary

Intel CFO Dave Zinsner provided a comprehensive update on the company's strategic initiatives, financial health, and future outlook. Key takeaways include: successful recent capital raises from the U.S. government and SoftBank, a strengthened balance sheet enabling debt deleveraging, the operational and financial separation of the Foundry and Product businesses, a disciplined approach to 14A node capacity build-out, and a nuanced view of competitive positioning across different product segments. Zinsner also touched on talent management post-restructuring and hinted at an upcoming detailed AI strategy.

II. Key Themes and Important Ideas/Facts

A. Strategic Capital Raises and Financial Strengthening

Intel has significantly bolstered its financial position through two major capital injections:

1. U.S. Government Investment:

• "A great deal for Intel" and a "win-win for everybody." The government took an equity stake in exchange for $5.7 billion in outstanding grants (plus $2.2 billion already received) and over $3 billion in secure enclave funding.
• Elimination of Uncertainty: This transaction removed uncertainty and "clawback rights" associated with previous grants, providing a guaranteed cash inflow.
• Improved Balance Sheet & Flexibility: The upfront money supports Intel's balance sheet, enabling deleveraging and maintaining liquidity for capital expenditure (capex) needs. It also removed "milestones and handcuffs" from the grants, granting Intel more operational flexibility.
• Silent Partner: The government is a "silent partner" with "no governance associated with it, no operational input." They have agreed to vote with board recommendations for the most part and are aligned with other shareholders in seeking a return on their investment.
• Warrants: The government received 5-year warrants associated with Intel stock, triggered if Intel sells more than 50% of the foundry business. Intel's motivation is "probably not to sell below 51%."

1. SoftBank Investment:

• Relationship-Driven: CEO Lip-Bu Tan's long-standing relationship with Masa (SoftBank CEO) was a key factor.
• AI Focus: SoftBank saw an opportunity to invest in a company that can leverage AI, with the foundry business being a key area. Masa gained "a lot of conviction around Lipu in terms of his ability to transform."
• Attractive Valuation: SoftBank saw the investment as an opportunity at an "attractive valuation."
• Capital Inflow: The SoftBank investment is expected to bring in $2 billion by the end of the quarter, pending regulatory filings.

1. Overall Capital Inflow & Deleveraging:

• This quarter has been "a significant quarter in terms of incremental capital raise."
• Intel also sold "almost a billion dollars worth of Mobileye stock" and expects to close the Altera deal soon, adding another "$3.5 billion."
• Deleveraging Goal: With this cash inflow, Intel intends to pay off $3.8 billion in maturing debt this year without refinancing any of it, essentially "wipe it out because of all this cash." This aligns with Lip-Bu Tan's goal of a stronger balance sheet and deleveraging.

B. Foundry Business Strategy and Operational Separation

1. Foundry as a Subsidiary:

• The foundry business was set up as a subsidiary to create "operational separation" for customers and to enable taking "money into the foundry business separately."
• Future investments from "customers on the foundry side" are not inconceivable, but unlikely "anytime soon" as it's "not quite investable yet." Any such sales would likely be "less than 49%" to avoid triggering government warrants.

1. Disciplined Approach to 14A Capacity (No Blank Checks):

• Lip-Bu Tan's "no more blank checks" statement means Intel "will build the capacity when we know we've got the demand for that capacity and we will not do it before that." This signifies a shift from past practices where Intel "spent money ahead of demand."
• 14A Requirements: 14A is "more expensive than 18A" due to the expected use of "high NA EUV tools" and more complex manufacturing steps, requiring higher wafer production to justify the ROI.
• Internal vs. External Demand: While Intel products will be a "big customer on 14A," external customers are crucial. 18A will be "milked for our internal products as much as possible," meaning 14A "doesn't flip on as significantly as early in the life cycle" with internal demand. External customers are needed to "fill in the gap."
• Confidence in 14A: Lip-Bu Tan is "getting increasingly more confident around 14A" due to early data and customer engagements. A "relatively low likelihood" of not getting the required demand.
• Timeline for 14A Decision: Intel expects to "have a good feel for how things are going" with 14A "sometime in '26."

1. Operational Separation of Products and Foundry:

• Separate P&Ls: The biggest step has been creating "separate P&Ls" for Products and Foundry, which drives managers to "optimize around those various variables."
• Different Metrics: Products are measured by margins (excluding wafer loading), while Foundry focuses on "wafer pricing versus wafer cost" and asset productivity.
• Ongoing Integration: There is still "a lot of intertwining between the business."
• ERP Systems: Implementing new ERP systems (one for Foundry, one for Products) is underway and expected to be "done by the end of '27," which will further separate processes. "Every quarter you'll start to see some changes in behavior and how we manage things."

1. Customer Engagements:

• 18A Challenges: Intel missed an early window for the "first wave of customers" on 18A due to not reaching optimal performance and yield in time. Lip-Bu Tan wanted "real confidence around the performance and the yield" before committing.
• Current 18A Focus: The priority is to see "Panther Lake and Clearwater Forest do well in terms of ramps." Panther Lake's first SKU is expected "out by the end of the year."
• Future 18A Opportunities: There will be "multiple waves of opportunities for 18A" and Intel is confident in securing customers, including "Department of Defense related" business.
• 14A Engagements: Customer engagements for 14A are in "early, early, early stages of PDK maturity" but have been "good." The learning cycle from these engagements helps refine the process.

C. Manufacturing Strategy and Competitive Landscape

Use of TSMC:

• Intel will "continue to use them now" as they are "a great partner for us" with "great" support and technology.Current Outsourcing: Intel is "probably at a relatively high level of outside wafers" currently, around 30% external versus 70% internal.
  
  Future Outsourcing: This 30% might "come down" but will "still be pretty high" compared to 10 years ago. Products like Lunar Lake are "almost all or actually it's all TSNC," and Arrow Lake is "mostly outsourced."
  
  Smart Capital Model: Outsourcing provides flexibility to manage demand volatility, avoiding building unused capacity, which is part of Intel's "smart capital" model. It also gives the products business "autonomy around choosing the right process for the product."
• 
  Product Roadmap and Node Progression:
• 
  
• 18A as a Workhorse: "18A is going to be a workhorse node for us." Peak volume on 18A is not expected until the "2030 time frame."
• Panther Lake: First SKU out by end of year, ramp through next year.
• Clearwater Forest & Diamond Rapids: Also on 18A.
• Noval Lake: Will also use 18A and is expected to "completely address that situation" on the client side.
• 14A Introduction: Intel products on 14A are expected in a "28 '29 time frame," aligning with external foundry customer introductions.
• 14A Performance Advantage: 14A, "particularly with backside power," is expected to be "a great offering" and a "bigger step up or leap up" than 18A. Intel Foundry will sell to "anybody," including competitors.

1. Competitive Positioning (CPU Market Share):

• Notebook: "Feel great about" and Panther Lake will be "another opportunity."
• Desktop: "Not as good" but "feel pretty good about how things will evolve there" with the roadmap.
• Server: Market share losses have "largely completed" and Intel is "more stable than we've been in the past." However, share is "not absolutely stable," and it will take "a couple years of product introduction" to fully stabilize and potentially grow share.
• Data Center Design: Intel has "work to do just on the pure design side" to be competitive across all segments. While doing "great" in certain areas like "AI workloads for single threaded performance," it doesn't "check all the boxes."
• Lip-Bu Tan's Involvement: Lip-Bu Tan's review of every design aims to ensure the "portfolio into the right place" by incorporating customer feedback.

D. Cost Structure and Talent Management

1. Workforce Reductions:

• "Regrettable loss of headcount has been relatively consistent." However, current reductions involve "quality people" and are "necessary."
• Targeted at Bureaucracy: Lip-Bu Tan's reductions are "largely targeted at eliminating bureaucracy." Intel had "roughly like 11 management layers" and he "cut it basically by half" to "speed up the process." This will be felt more by employees "through the end of this year and into next year."
• Impact on Products and Foundry: Reductions have impacted both sides of the business, with a focus on streamlining and reducing overhead.

1. Talent Attraction:

• Lip-Bu Tan is an "attractor of talent," even amidst reductions. Despite the expected "grind" over the next couple of years, he has been "surprised at the people that he's been able to attract who a lot of cases don't need a job."

E. AI Strategy

1. Upcoming Unveiling: Lip-Bu Tan will "unveil that and talk about it" in the "next few months," possibly at the next earnings call.
2. Jaguar Shores: Mentioned as "the product that, you know, is kind of where we want to end up."
3. Leveraging Internal Technology: Intel has "a lot of technology within the company that we can leverage into the AI space."
4. Talent Attraction: AI is an area where Intel is "attracting some interesting talent."

F. Margin Targets and Timelines

1. Near-Term Goal: The immediate goal is to drive margins "into the 40s" from the current "30s."

• Tailwinds:18A Cost Structure: "18A as just a cost structure, a margin mix benefit to Foundry will certainly be a tailwind."
• Panther Lake Architecture: Expected to drive "better margins" compared to Lunar Lake.
• Cost Focus: Overall improvement in "cost focus."

1. Ultimate Driver: "At the end of the day, the biggest thing we can do is get products out that are really competitive. Get products out that are really competitive, you end up with better pricing, you end up with better margins, and you end up with better share."

Bottom of Form

!! NotebookLM can be inaccurate; please double check its responses.

 

[hist78]

Use of TSMC:

Intel will "continue to use them now" as they are "a great partner for us" with "great" support and technology.

Current Outsourcing: Intel is "probably at a relatively high level of outside wafers" currently, around 30% external versus 70% internal.

Future Outsourcing: This 30% might "come down" but will "still be pretty high" compared to 10 years ago. Products like Lunar Lake are "almost all or actually it's all TSMC," and Arrow Lake is "mostly outsourced."

Smart Capital Model: Outsourcing provides flexibility to manage demand volatility, avoiding building unused capacity, which is part of Intel's "smart capital" model. It also gives the products business "autonomy around choosing the right process for the product."

 

[DanX]

Great job by LBT.
Only if he is in charge of the US government and balances the debt of the US.

 

[hist78]

About Intel Foundry Spinoff:

Christopher Danely
Citigroup Inc. Exchange Research
Could other semi companies joined the investment parade, either straight in until now or the foundry whenever you spin out the foundry? Is that on the table?


David Zinsner
Executive VP & CFO
Yes. Look, it's certainly a consideration. I think what you'd like to see. And one of the reasons we took foundry, and we set it up as a subsidiary of the parent company, was to enable -- partly was it to create operational separation a little bit because customers are going to want to see that. But in addition to that, it was to set it up in a way that, hey, we could take money into the foundry business separately. And there are customers on the foundry side that might like the notion of investing in the future of that foundry business. So it's not inconceivable that we do that.

I think the likelihood is that it won't happen anytime soon because it's not quite investable yet. But at some point, down the road, I could see that happening. Obviously, the one thing I didn't mention in the structure of the government, financing is that they also got warrants associated with Intel stock and it triggers off of us selling below -- or selling more than 50% of the business.

I think as long as we hold 51% essentially, it doesn't trigger, and it's a 5-year warrant. And so our motivation will probably be not to sell below 51% because that would dilute investors significantly unless it made economic sense for investors for us to do that. And so the likelihood is if we are selling stakes in foundry, it would be something less than 49% that would be sold off.

 

[samwilde]

If you spinoff Intel Foundry it won't help things but make them worse. They are totally dependent on Intel as a customer and their PDKs are no good.

 

[MKWVentures]

Good summary, I would read the transcript to get a little better feel for what was actually said.

On foundry a couple items:
1) Will foundry be break even as committed in 2027 (its doesnt look like it unless massive write off in 2025/6). Lets see those Arizona Financials. "Foundry is not investable"
2) On Foundry independence. Is Product group allowed to choose who the want to use or are they told to use 18A? There are well documented ways to divide today IF you wanted to (Bonuses, Stock, no shared managers, Documentation of intercompany discussions)
3) Are there any external customer product tapeouts planned in 2025 or 2026. It sounds like no.

 

[tooLongInEDA]

So, let's take this statement (implied that it was made by David Zinsner, if the AI filtering didn't get that wrong):

Current Outsourcing: Intel is "probably at a relatively high level of outside wafers" currently, around 30% external versus 70% internal.

Two obvious comments here:

1) "Probably" ? Doesn't he actually know ? Surely that's his job ...

2) Is this by wafer volume (implied) or $ value ? I'm assuming Intel buys wafers over a range of processes and there's some cost spread over these.

I find the comment unhelpfully - perhaps deliberately - too vague.

 

[lispwright]

1) "Probably" ? Doesn't he actually know ? Surely that's his job ...

He's been a finance guy in semiconductor manufacturing for a quarter century, but only at Intel for 3.7 years.

As the CFO of a public company, it's his job to tell the truth as he knows it, and was he speaking extemporaneously? Does he know 18A is going to go cleanly into HVM, especially with all the layoffs? At one extreme, might the single Panther Lake SKU by the end of the year become another Cannon Lake? What about the rumors of parametric yield problems with 18A, and the concerns about it being the first industry node with backside power delivery?

Intel has put only one major node since 10 nm into HVM, 18A failed to attract any whales and as far as we know only 2 small fry, he'd be irresponsible to not hedge both his statements and strategy. For all we know, Intel still has a large reservation of TSMC capacity in case of 18A problems. Perhaps one it can surrender to other companies like Nvidia if it doesn't turn out to need it?