On September 4, Intel's CFO, David Zinsner, spoke at Citigroup's Global Technology Conference. Below are some key insights from the conference.
Intel Will Not Make Their 20A Arrow Lake Chip Internally
At first, Intel was planning to make Arrow Lake themselves, but now they're going to skip 20A, focus their efforts on 18A, and use "external partners" to manufacture Arrow Lake. 20A was supposed to help Intel improve their manufacturing process so they could successfully make 18A, but since the defect density per square centimeter for 18A is 0.4, they felt comfortable going straight to 18A. Defect density per square centimeter measures the amount of defects per square centimeter on a semiconductor wafer. A lower number is preferred.
Intel did not mention the company they will use to manufacture Arrow Lake, but everyone believes it's TSMC since TSMC is already making their Lunar Lake processor.
Intel's rough patch is teaching them how to leverage the semiconductor ecosystem to their advantage. They're designing chips with software from EDA companies, manufacturing chips with EUV machines from ASML, outsourcing chip production to TSMC, and building fabs with money from the government and private equity firms. They're transforming their business into a flexible, nimble operation.
Lip Bu-Tan Was Right About the Cost Cuts
One of the reasons Lip Bu-Tan quit Intel's Board was because he believed the company was moving slow to cut costs. At the conference, David Zinsner mentioned that Intel's original plan was to spend 2024 analyzing the company's cost structure, and in 2025 they would cut costs, but when they saw that the back half of 2024 was going to be soft, they decided to cut costs in 2024 instead of 2025. Just like Lip Bu-Tan said.
To reduce costs, Intel will reduce the number of people who are not directly involved in the design, manufacturing, and sales of their products. They are also reducing the number of labs they have. According to Zinsner, Intel has facilities all over the world, and almost every facility has a lab, which is unnecessary. Intel is also scaling back on the amount of shell space that they need.
By next year, these cuts should reduce their cost of sales by $1 billion, operating expenses by $3.5 – $4.5 billion, and capital expenditures by $10 billion.
Accelerating these cuts is good because when Intel's business picks up, more of their revenue will fall to the bottom line. It's interesting that despite Intel's troubles, they still control over 70% of the PC and server CPU market. Investors minimize the fact that Intel is still the big dawg in the CPU space. In Q2 of 2024, Intel's client revenue was ~5x AMD's.
Once Intel stabilizes their business, they'll be able to grow their market share and revenue with a reduced cost structure.
Cuts Won’t Affect Chips Act
Market noise makes it seem like Intel is wasting the government's money because they're firing people and delaying their fab projects. But the CFO revealed that Intel has not received any money from the government yet. They still need to finalize their agreement with the government and hit certain milestones before money is released to them. The milestones include finishing certain levels of construction, using certain equipment in their operations, producing a specified number of wafers, and signing up customers.
It's important that investors do their own research when investing because market noise can fill our heads with misinformation. The market's criticism of Intel for wasting money that they haven't received yet seems like a small thing, but it's part of the overall narrative surrounding the company where everything they do is wrong. This type of hate towards Intel will depress the company's stock price and create a windfall for investors when the company turns around.
Potential Customers for Intel Foundry Services (IFS) Are Concerned About the Separation Between the Products Group and IFS
David Zinsner mentioned that customers want to see a separation between IFS and Products in order for them to feel comfortable using Intel's foundry services. I think most of Intel's foundry customers will be companies that don’t directly compete with them. I will be surprised if AMD or NVIDIA use Intel to make their chips. I can see them use Intel to package their chips, but not to manufacture them. The risks are too high for AMD and NVIDIA to expose their IP to one of their main competitors. They will most likely use TSMC and Samsung instead of taking a risk with Intel, unless they are forced to use Intel because some issue arises with TSMC and Samsung.
Intel’s Advanced Packaging Business Has Good Margins.
The margin for Intel's Advanced Packaging business is ~40%. This is a good business for Intel to lean into. Especially until they start manufacturing chips for customers in 2026–2027.
David Zinsner Spoke About Intel Products and Foundry as Two Businesses Under the Same Roof
There's a rumor in the market that Intel might get rid of their foundry business and focus on being a fabless company. David did not mention anything like this at the conference. Instead, he talked about Intel's long-term plan of reaching a 60/40 margin profile and how their advanced packaging business was going to help them achieve those margins. He also talked about cutting down on hot lots and testing times, and other things related to operating the foundry business long-term. It doesn't seem like they're going to split from the foundry business anytime soon.
The company is meeting this month to decide which direction to take the company in. There are rumors swirling around that they might sell Mobileye, their foundry business, and Altera. We should hear the outcome of their meeting by the Q3 earnings call.
Stay strong, stay blessed, and God willing, I will see you next week.
Do not judge by appearances, but judge with right judgment.
John 7:24 ESV
コメント