Intel — Funding the Foundry
Download case PDF (6 pages)Request teacher’s pack
Decision date and information cutoff: 16 September 2024, 4:04 p.m. ET
Decision scope: Governance and staging of Intel Foundry investment
Case convention: This is an author-created management decision exercise. The assignment and options are hypothetical; company disclosures are historical. No private board deliberations are represented.
The decision
At 4:04 p.m. on September 16, Intel CEO Pat Gelsinger sent employees a public update on the company’s next phase. Intel would move toward making its manufacturing operation—Intel Foundry—an independent subsidiary within Intel. The stated aim was to give external customers and suppliers clearer separation while preserving collaboration with Intel’s product businesses. The subsidiary had not yet been established. P7
The update came after Intel’s difficult second quarter and August cost-reduction plan. It also followed two announcements that appeared to support the Foundry strategy: a multi-year, multi-billion-dollar framework with Amazon Web Services (AWS), including a custom AI fabric chip planned for Intel 18A and a custom Xeon 6 on Intel 3; and an award of up to $3 billion in direct funding for the U.S. Secure Enclave program. P6, P9
Those announcements offered evidence of customer and government interest. They did not disclose a minimum AWS wafer volume, detailed unit economics or funding receipts. An award ceiling is not cash received. Intel’s near-term external Foundry revenue remained small beside the manufacturing operation’s total reported segment revenue. Much of that segment’s revenue came from Intel’s own products business. P3, P9
Assignment: Imagine that you are part of a hypothetical strategy team preparing recommendations for Intel’s directors at the next governance checkpoint. Recommend how Intel should govern and stage Foundry investment over the next two years. Explain how your approach would preserve—or deliberately limit—the links between product design and manufacturing. Address capital discipline, external-customer confidence, capacity timing and what evidence should trigger the next investment step. Use only information available by 4:04 p.m. ET on September 16, 2024.
Two businesses under one roof
Intel designed processors and related products, including Core client processors and Xeon server processors. It also owned and operated the factories that developed process technology and manufactured many of those products. In its new internal-foundry reporting model, Intel Products included Client Computing (CCG), Data Center and AI (DCAI), and Network and Edge (NEX). Intel Foundry combined process development, manufacturing and supply chain with the Foundry Services business. P2, P3
The model made manufacturing costs more visible. Foundry recorded charges to Intel Products for wafers and related services; Intel said intersegment prices were intended to approximate market prices. But segment revenue did not represent outside customers alone. Intersegment revenue and profit were eliminated when Intel consolidated its statements. The company also said some Foundry costs benefited other segments and that it did not regularly provide its CEO with discrete asset information by segment. P2, P3
Intel presented integration as an advantage: product designers and process engineers could optimize chips, manufacturing and packaging together. Its “systems foundry” proposition extended beyond wafer production to advanced packaging, design tools and ecosystem support. The strategy also asked external customers to entrust manufacturing to a company that sold products competing in some of their markets. Separate reporting could improve visibility and accountability, but did not by itself establish customer trust, yield or competitive process economics. P5, P7
A capital-intensive recovery plan
In February 2024 Intel introduced a goal of delivering five process nodes in four years. Its leading-edge roadmap included Intel 18A, with volume production planned for 2025. By August, Intel reported that the 18A Process Design Kit (PDK) 1.0 had been released to Foundry customers and that the first client and server products on 18A had powered on. These were company-reported development milestones; Intel did not disclose 18A yield, cost per wafer or high-volume production performance. P3, P4
Intel’s financial results put pressure on the timeline and funding plan. For the second quarter ended June 29, 2024, Intel reported $12.833 billion in consolidated revenue and a GAAP net loss attributable to Intel of $1.61 billion. Non-GAAP EPS was positive $0.02 while GAAP EPS was negative $0.38. In the recast segment presentation, Foundry reported $4.320 billion in revenue and a $2.830 billion operating loss. The company attributed that loss to lower intersegment product and services profit, lower external revenue, factory startup and capacity-curtailment charges, impairments and higher process-technology investment, offset in part by lower excess-capacity charges. P3, P4
The composition of Foundry revenue mattered. Of the $4.320 billion reported for Q2, $4.243 billion was intersegment and $77 million was external. In 2023, external Foundry and assembly/test revenue was $953 million, within a segment that reported $18.910 billion in total revenue. Those are different measures: segment totals include Intel internal activity, while consolidated accounts eliminate intersegment sales. They should not be presented as evidence of $18.9 billion in outside demand. P2, P3
Exhibit 1. Selected Intel Foundry segment results
USD billions unless noted; segment presentation recast under Intel’s 2024 internal-foundry model.
| Period | Segment revenue | External revenue | Intersegment revenue | Segment operating income (loss) |
|---|---|---|---|---|
| FY2023 | $18.910 | $0.953 | $17.957 | $(6.955) |
| Q2 2024 | $4.320 | $0.077 | $4.243 | $(2.830) |
| H1 2024 | $8.689 | $0.104 | $8.585 | $(5.304) |
Sources: P2, FY2023 recast segment table and Foundry loss discussion; P3, Q2 2024 Form 10-Q, Note 2 and Foundry MD&A. Amounts converted from millions. Segment revenue is not consolidated revenue; external revenue is not a measure of contractual backlog. Segment loss includes technology development and manufacturing costs and is affected by Intel’s allocation and transfer-pricing model.
Cash, costs and capacity
For the first six months of 2024, Intel reported $1.069 billion in cash from operations and $11.652 billion in additions to property, plant and equipment (PP&E). It also reported capital-related government-incentive proceeds and partner contributions. Intel’s adjusted free cash flow was a non-GAAP measure that included specified partner/incentive cash flows. In August, Intel defined gross capex as GAAP PP&E additions and net capital spending as PP&E additions net of capital-related government incentives and partner contributions. The measures should not be mixed. P3, P4
Intel’s August plan aimed to reduce operating and capital spending by more than $10 billion in 2025 versus previous estimates. It forecast gross capex of $25–27 billion and net capital spending of $11–13 billion in 2024, followed by gross capex of $20–23 billion and net capital spending of $12–14 billion in 2025. It also targeted non-GAAP research and development and marketing, general and administrative expense of about $20 billion in 2024 and $17.5 billion in 2025, with a workforce reduction greater than 15%, most completed by year-end. These were management targets, not realized savings. P4
The September 16 update described a more selective geographic buildout. Intel planned to pause its Poland and Germany projects for about two years. It would complete construction of a new advanced-packaging factory in Malaysia but align startup with market conditions and use of current capacity. Intel said it would continue projects in Arizona, Oregon, New Mexico and Ohio, and that its Ireland fab would remain its European hub for the foreseeable future. The company described these changes as adjustments to the timing and scope of expansion, not a withdrawal from its U.S. manufacturing plans. P7
Exhibit 2. Recent financial capacity and planned spending
USD billions; six months ended June 29, 2024, except management targets.
| Measure | Amount | Basis |
|---|---|---|
| Cash provided by operating activities, H1 2024 | $1.069 | GAAP cash flow |
| Additions to PP&E, H1 2024 | $11.652 | GAAP investing cash flow |
| Gross capex target, FY2024 | $25–27 | Intel guidance; PP&E additions basis |
| Net capital spending target, FY2024 | $11–13 | Intel non-GAAP definition; net of qualifying capital incentives and partner contributions |
| Gross capex target, FY2025 | $20–23 | Intel guidance; PP&E additions basis |
| Net capital spending target, FY2025 | $12–14 | Intel non-GAAP definition; net of qualifying capital incentives and partner contributions |
Sources: P3, Consolidated Condensed Statements of Cash Flows; P4, “Reducing Capital Expenditures” and non-GAAP definitions. Gross and net capex are not interchangeable. Net capital spending is a forecast, not an amount of cash already funded or a site-level investment authorization.
Signals from customers and competitors
Intel had reported more than $15 billion in expected lifetime deal value from external Foundry customers in April 2024 and a large test-chip pipeline. Those figures were company expectations, not recognized sales or independently verified bookings. P5, P8
On September 16, Intel and AWS announced an expanded collaboration. The companies described a multi-year, multi-billion-dollar framework covering products and wafers, co-investment in custom chip designs, an AI fabric chip planned for 18A, and a custom Xeon 6 planned for Intel 3. They also said the collaboration would accelerate development of manufacturing in Ohio. The public announcement did not state a minimum wafer purchase, project economics, schedule or how much value would be recognized as Foundry revenue. It was a meaningful customer signal, not guaranteed wafer revenue. P6
Intel competed with established foundries and with fabless chip designers that relied on them. In the second quarter of 2024, Taiwan Semiconductor Manufacturing Company (TSMC) reported $20.82 billion of revenue, 53.2% gross margin and 42.5% operating margin. TSMC forecast Q3 revenue of $22.4–23.2 billion and gross margin of 53.5–55.5%. It said it planned capacity in close coordination with customers and sought to fill fabs with demand across multiple customers and product generations. TSMC’s product mix, geographic footprint and accounting differ from Intel’s, so these results are context rather than a direct comparison of economics. C1, C2
Intel also faced a customer-confidence challenge. A customer could value U.S.-based capacity, advanced packaging and close collaboration with Intel engineers, while worrying that Intel might compete with it, prioritize internal products or expose proprietary design information. The company said a subsidiary could offer clearer separation and future financing flexibility. It had not yet disclosed the subsidiary’s governance, customer protections, capital structure or operating agreements with Intel Products. P7
Three routes forward
The following alternatives are designed for classroom analysis. They are not known proposals from Intel directors or executives.
Keep Foundry closely integrated and fund the roadmap through Intel. Preserve tight product–process coordination and prioritize the 18A ramp, packaging and the existing geographic network. The model could support internal product differentiation and use Intel Products demand to build process experience. It would keep Intel responsible for most of the capital burden and could leave external customers uncertain about neutrality and priority.
Give Foundry stronger independence and external financing options. Establish a subsidiary with a separate operating board, explicit customer/IP safeguards and transparent cost and capacity reporting. Consider independent financing only against defined business conditions. This could strengthen customer confidence and accountability, but legal separation would not create yield or volume. It could also raise financing costs, divide shared engineering teams or create friction with Intel Products.
Use a demand-gated hybrid. Preserve integration where process and product development are tightly linked, while setting independent Foundry governance and customer protections. Stage capacity and construction starts against process milestones, customer commitments and utilization evidence; use third-party capacity or partner funding where the economics justify it. The hybrid could retain flexibility but requires clear rules about transfer prices, engineering priorities, shared assets and investment gates.
For any route, directors would need to decide how much authority Foundry has over pricing, customer selection, intellectual property and capacity allocation; how Intel Products commits to using internal manufacturing; what evidence is sufficient to authorize each major construction or tool purchase; and what should happen if demand, yield or cost targets fall short.
Recommendation
Prepare a two-page recommendation for the hypothetical strategy review. Select a route or specify a hybrid. State the conditions that would justify the next investment step, and how directors should monitor customer trust, 18A readiness, internal and external demand, capital efficiency and product competitiveness. Explain which decisions should be made now, which should remain conditional and what evidence would cause you to change course.
Discussion questions
- What are the economic and strategic advantages of Intel’s integrated product and manufacturing model? Which activities require close coordination, and which could be separated?
- How much do the Q2 Foundry segment results tell you about outside-customer economics? What do they fail to show?
- Does a Foundry subsidiary make Intel more credible to external customers? Which governance or operating safeguards matter beyond the legal structure?
- How should Intel stage its capacity and capital plan across 18A, the U.S. sites, Europe and Malaysia? What observable evidence should trigger the next step?
- What is the strongest objection to your recommendation? What would you measure over the next year to detect whether the strategy is working?
Sources and conventions
P2. Intel, 8-K Exhibit 99.1: retrospective recast of segment results, furnished 2 April 2024.
P3. Intel, Q2 2024 Form 10-Q, period ended 29 June 2024; filed 2 August 2024.
P4. Intel, Q2 2024 earnings release and cost-reduction plan, 1 August 2024, Exhibit 99.1.
P5. Intel, Foundry financial framework, 2 April 2024.
P6. Intel and AWS, expanded strategic collaboration, 16 September 2024, 4:01 p.m. ET.
P7. Pat Gelsinger, September 16 strategy update, 16 September 2024, 4:04 p.m. ET.
P8. Intel, Q1 2024 Form 10-Q, period ended 30 March 2024; filed 26 April 2024.
P9. Intel, Secure Enclave award announcement, 16 September 2024, 2:02 p.m. ET.
C1. TSMC, 2024 Q2 results, 18 July 2024.
C2. TSMC, Q2 2024 earnings call transcript, 18 July 2024.
All financial figures are U.S. dollars. Segment results reflect Intel’s 2024 recast reporting model; see cited filing notes. GAAP and non-GAAP measures remain distinct. No later outcomes, subsequent filings, or actual private board decisions are included. No reuse license has been assigned.