Intel Corporation (INTC) Fundamentals – Large Legacy CPU & Foundry Ambitions; PT: $25 v. $24.00

Posted on by Jerry Ng

Question #1: Is This A Good Business?

Subjective vs. Objective

  • Subjectively: Intel remains the historical leader in x86 CPUs for PCs and Servers, operating as an IDM (Integrated Device Manufacturer) with in-house fabrication. It has strategic importance to the semiconductor industry and a strong brand (especially in data center and PC CPU markets).
  • Objectively (via cash flow):
    • Free Cash Flow (FCF): Currently under pressure due to heavy CapEx for next-gen process nodes and foundry ambitions.
    • Revenue Growth: Had been declining in 2023–24 due to share loss in servers (to AMD, ARM) and a PC market slump. Potential re-acceleration possible if the product roadmap regains competitiveness and foundry strategy monetizes.
    • Profit Margin: Historical mid-50%+ gross margin is down to ~40% range. Operating margin was ~30% historically but has dipped closer to 15–20% recently due to underutilized fabs and intense R&D.
    • Capital Investment: Enormous outlays ($25–$30B/year range) to achieve advanced process leadership, plus the newly minted foundry business.
    • Terminal Value: If Intel reestablishes CPU leadership or effectively pivots to a profitable foundry model, the market could maintain a strong multiple. But if competition outpaces them, Intel’s large manufacturing assets could remain underutilized.

Business Value Creation Distilled to Five Factors

  1. Organic Growth: Currently negative yoy, but potential rebound if new CPU roadmaps deliver and foundry customers ramp.
  2. Margin Trajectory: Margins depressed near term by ramp costs; can rebound to ~25%+ operating margin if node execution improves and volumes fill fabs.
  3. Capital Intensity (ROIC)
    • Formula: ROIC=NOPAT/Invested Capital
    • 2023–24 ROIC: ~2–4% (very low given heavy CapEx).
    • Potential 2026+ ROIC: If volume recovers and foundry scales, could approach high single digits or ~10%.
  4. Capital Deployment: Currently, dividend was historically a priority, but it has been cut recently. Ongoing M&A (Tower Semi) and large CapEx overshadow share buybacks.
  5. Terminal Value Perception: If Intel regains CPU or foundry leadership, the multiple could expand. If not, it’s an aging x86 franchise with underutilized fabs.

Question #2: How Does This Business Make Money?

Revenue Model

  1. Client Computing Group (CCG)
    • What & How: Sells x86 CPUs for notebooks ($120–$140 ASP historically) and desktops ($150–$200).
    • Major Customers: PC OEMs (e.g., Dell, HP, Lenovo, Apple historically—but Apple is phasing out x86).
    • Contract Duration: Typically 1–3 years per platform. High switching costs for OEMs, but competition from AMD and Apple’s ARM-based silicon has grown.
    • Recurring Revenue? Mostly one-time CPU sales. Some incremental over-the-air updates are possible but minimal.
  2. Datacenter & AI Group (DCAI)
    • What & How: Sells Xeon CPUs for servers (ASP historically $600–$700) to enterprises, cloud providers, and HPC.
    • Major Customers: Hyperscalers (AWS, Azure, GCP), enterprise OEMs (Dell EMC, HPE, Lenovo), government HPC.
    • Revenue Concentration: Cloud ~50%, Enterprise ~30%, Comms ~20%.
    • Recurring Revenue? Minimal. CPU is a one-time sale. Some enterprise multi-year contracts exist but are not subscription-based.
  3. Network & Edge (NEX)
    • What & How: Supplies silicon for network infrastructure, vRAN, and edge computing.
    • Major Customers: Telecom OEMs, enterprise networking.
    • Recurring Revenue? Largely upfront, but longer design cycles.
  4. Intel Foundry Services (IFS)
    • What & How: A separate business to manufacture wafers for external customers. Revenue is tied to wafer volumes and utilization.
    • Pricing Model: Typically cost+ or market-based foundry rates.
    • Contract Duration: 3–5+ years for large engagements. Still a small portion of total revenue (<5%).
  5. Mobileye, Programmable Solutions Group (PSG), and Others
    • Small segments that provide automotive ADAS chips (Mobileye, partially IPO’d) or FPGAs (PSG) to enterprise/communication.

Customer Concentration

  • Heavily reliant on a few major PC OEMs and cloud providers. No single end-customer >20% of total revenue, but the top 5 might represent ~40–50%.

Recurring Revenue

  • Very limited on the CPU side. If Intel Foundry Services grows with multi-year wafer supply agreements, that could bring more stable recurring revenue.

Question #3: Nature of the Cost Structure

A. Granular Breakdown of R&D

  • For 2024E, total R&D ≈ $15B or ~25–30% of revenue (excl. Mobileye). Example sub-allocation:
    • Process Node R&D (~50%): e.g., Intel 3, 20A, 18A transistor development, advanced packaging.
    • CPU/GPU SoC & Architecture (30%): new CPU microarchitectures, GPU expansions, AI accelerators.
    • IFS & EDA Tools (10%): foundry enablement, internal EDA flows.
    • Safety & Validation (5–10%): advanced reliability, security.

B. Fixed vs. Variable Costs

  • Fixed: R&D, a portion of G&A, overhead for fab depreciation.
  • Variable: Materials (silicon wafers, chemicals), direct labor, packaging, testing.
  • Operating Leverage: Mixed. Fabs have large fixed depreciation, so higher wafer volumes can significantly boost GM. In 2023–24, underutilization weighs on margins.

C. Incremental Margins

  • Historically, each +10% revenue can yield +15–20% EBIT growth if fabs are well utilized. Right now, idle capacity reduces incremental margin.
  • If demand returns to fill capacity, incremental GM can jump to 60%+ on extra units.

Fixed Costs

  • Depreciation: A major line item (tens of billions).
  • R&D: ~$15B.
  • G&A: ~$6–$7B.

Variable Costs

  • BOM for each CPU or chip, plus packaging costs.
  • Foundry external wafers (for some designs, e.g. GPU, certain chiplets).

Operating Leverage Proxy

  • Under normal conditions, a +10% revenue growth might drive +15–25% EBIT improvement. Currently, negative leverage while volumes remain below the fab’s cost-recovery threshold.

Comparison with Comps

  • TSMC: Pure-play foundry, runs at ~50%+ operating margin, more consistent utilization.
  • AMD: Fabless, with lower overhead, invests heavily in design.
  • NVIDIA: Primarily GPU-based, outsources manufacturing, ~70% GM, leaner overhead.

Question #4: Key Drivers of the Business

Driver #1: Product Competitiveness in Server CPUs

  • Narrative & Debate: Intel’s new server chips (Granite Rapids, Sierra Forest on Intel 3) must regain performance/watt leadership vs. AMD Genoa/Bergamo. If Intel’s roadmap slips again, share loss will continue.
  • Financial Impact: Each ~5% server share shift ~$2B in revenue swing. Potential margin pivot if volumes saturate or if it remains underutilized.

Driver #2: PC TAM and Market Share

  • Narrative & Debate: Post-COVID slowdown decimated PC shipments from 350M to ~250M. Can Intel maintain ~70% share vs. AMD, Apple, ARM entrants?
  • Impact: The PC segment is ~40% of Intel’s revenue. Even partial share changes can move EPS significantly.

Driver #3: Foundry (IFS) Ramp

  • Narrative & Debate: Can Intel attract external customers for advanced nodes by 2026–27? Potential synergy or conflict if Intel Product competes with foundry customers.
  • Impact: If IFS hits >$10B by 2027 with decent margin, it’s a new growth pillar. If adoption is slow, the large idle capacity drags overall margin.

Driver #4: AI Acceleration

  • Narrative & Debate: AI training/inference is shifting to GPUs/accelerators. Intel invests in Gaudi (Habana) and GPU, but needs traction. CPU-based AI PC is also a new theme.
  • Impact: Gains in AI adjacency can add billions in the data center or client sales, or it could remain overshadowed by Nvidia’s GPU dominance.

Driver #5: Cost Cuts & Capital Efficiency

  • Narrative & Debate: Management commits to $8–$10B in cost savings by 2025. Also exploring Smart Capital (shell-first, government grants, Brookfield JV).
  • Impact: If successful, Intel can preserve FCF and dividends. If underutilization persists, negative GM leverage continues to hamper EPS.

Model Sensitivities & Debate

  • Where is the Debate? Execution on 18A node, the timeline for regaining server leadership, foundry traction, and the magnitude of cost discipline vs. needed R&D.
  • Shiny Ball Syndrome? Robotaxi, AI GPUs, and foundry expansions all overshadow the core x86 CPU cyclical reality. The real driver remains CPU competitiveness.

Question #5: Business Momentum

Recent Accelerations / Decelerations

  • Revenue: 2023 ~$52–$54B (-13–15% yoy). Down from $63B in 2022.
  • 2–3 Year Stack: Down ~25% vs. 2021 peak, reflecting share loss + cyclical slump.
  • Potential reacceleration in 2025 if server roadmap improves and PC normalizes.

Trajectory of Key Drivers

  • Bull Case: Intel successfully launches Granite Rapids on schedule, stabilizes server share at 70% by 2025, and foundry wins a handful of external customers. EBIT margin climbs to high-20s, revenue recovers to $65B+ by 2026.
  • Base Case: Slight server share declines continue, foundry adoption is modest, cost cuts partially offset idle capacity. Revenue stagnates near $55–$60B range, with ~20% operating margin.
  • Bear Case: Repeated roadmap delays, foundry flops, major HPC, and enterprise shift to AMD/ARM. Revenue <$50B by 2026, margins stuck ~15%.

Expectations Gap

  • The market is discounting a prolonged share loss offset by partial cost improvements. The stock trades near 1.2x book value, reflecting subdued growth prospects.
  • Some see an upside if Intel’s new leadership revives CPU competitiveness or effectively monetizes foundry JV strategies. Others remain skeptical given repeated misexecution.

Putting It All Together
Intel’s fate hinges on server CPU competitiveness, foundry scaling, and disciplined cost management. Underutilized fabs weigh on near-term margins, but if future node and product roadmaps succeed, a margin expansion to ~25–30% is feasible. Investors are cautious, reflecting historical delays, but any positive surprise on execution or foundry traction can lead to re-rating.

Thesis: “Intel – Potential Turnaround if CPU Roadmaps Deliver, But Big If”

Price-Embedded Key-Drivers

  • The stock (~$24) seems to reflect partial CPU share erosion and high capital burdens.
  • If Intel demonstrates on-time CPU product launches with improved performance vs. AMD, we see potential multiple expansion.
  • Foundry success remains uncertain, but even moderate uptake helps fab utilization.

Base Case View on Key Drivers

  • PC TAM normalizes around 250–270M by 2025; Intel holds ~65–70% share.
  • Server share stabilizes near 60–65% by 2025 if new leadership swiftly addresses product execution.
  • Foundry revenue remains single digits ($3–$5B) by 2026.

Why the Mispricing or Misperception Exists

  • Past node delays degrade confidence. Some investors assume perpetual server share decline.
  • Street not fully crediting the cost cuts or potential synergy from government subsidies.

When the Mispricing Might Close

  • Could close by late 2025 if Intel hits major CPU milestones (e.g., on-time next-gen server CPU) or announces meaningful external foundry customers.
  • Alternatively, a spin-off or partial separation of IFS could crystallize value.

Valuation

Under Our Bull Scenario

  • We assume revenue recovers to $65B by 2026, operating margin rebounds to ~28%.
  • EPS surpasses $3.00, at ~13x multiple = mid-$40 stock.
    Base Case
  • Stays near $55–$60B rev, mid-teens to 20% margins, EPS $2.00 by 2025–26, at ~12–13x = $24–$26 stock.
    Bear Case
  • Fabs remain underutilized, share loss to AMD/ARM, EPS near $1.00 or less, stock ~$15–$18.

Final Thought
If Intel can revitalize its CPU roadmap and keep capital discipline, the stock could re-rate from its depressed multiples. But skepticism runs high given repeated misses. We remain neutral (Equal-Weight) with a price target of $25.

Unknown's avatar

Published by Jerry Ng

Leave a comment