How Intel Makes its Money: Revenue Breakdown

How Does Intel Make its Money?

Intel (NASDAQ: INTC) is one of the world’s largest semiconductor companies, designing and manufacturing processors for personal computers, data centres, networking equipment, and embedded systems. Intel generated $54.2 billion in total revenue for fiscal year 2024 — flat year-over-year — while reporting a net loss of $1.6 billion, marking one of the most difficult financial years in its history.

Intel’s defining characteristic — and its defining strategic challenge — is that it is an Integrated Device Manufacturer (IDM): a company that both designs its own chips and operates its own fabrication facilities (fabs). Every other major chip designer of scale (Qualcomm, AMD, Apple, Nvidia, ARM-based designers) is fabless — they design chips but outsource manufacturing to foundries like TSMC. Intel’s integrated model was its greatest competitive advantage for four decades. Since approximately 2016, it has become its greatest liability, as Intel’s manufacturing processes fell behind TSMC’s, its chips lost performance-per-watt leadership, and customers began defecting to AMD and Apple-designed alternatives.

The company is now in the most complex corporate turnaround in semiconductor history: attempting to simultaneously regain manufacturing process leadership, rebuild its product roadmap, launch an external foundry business to justify its fab investments, and maintain financial stability — all while reporting operating losses and burning capital at extraordinary rates.

CEO change: Pat Gelsinger, who launched the IDM 2.0 turnaround strategy in 2021, was removed by the board in December 2024. Lip-Bu Tan, former CEO of Cadence Design Systems and a highly respected semiconductor industry veteran, was appointed as Intel’s new CEO in March 2025. The leadership change signals that the board was not satisfied with the pace of turnaround progress under Gelsinger and is resetting strategic priorities.

Key Takeaways

• Intel generated $54.2B in FY2024 revenue — flat year-over-year — while reporting a $1.6B net loss and a -4.8% operating margin; a stark reversal for a company that historically earned 55-60% gross margins
• Client Computing Group (54% of revenue) remains resilient in Windows PCs despite AMD competition; Intel still commands ~70%+ market share in laptop and desktop processors
• Data Center & AI (DCAI, 24% of revenue) is the crisis segment: Xeon server CPU revenue has declined as enterprises shift data centre spending from CPUs to Nvidia GPUs for AI workloads; Gaudi AI accelerators have not gained meaningful traction
• Intel Foundry (formerly IFS, 4% of revenue) is the long-term bet — manufacturing chips for external customers — but currently loses billions annually and needs marquee customer wins to justify its $25B+ annual capex
• 18A process node is the make-or-break manufacturing milestone: if Intel’s 18A technology achieves competitive performance vs. TSMC’s N2/N3, it validates the entire foundry investment thesis; delays or underperformance would be devastating
• Pat Gelsinger departed in December 2024; Lip-Bu Tan became CEO in March 2025 — a reset that signals the board’s dissatisfaction with turnaround pace and may lead to strategic pivots including potential separation of Products and Foundry
• CHIPS Act funding ($8.5B in grants + $11B in loans from the US government) is essential to Intel’s fab investment programme; these commitments reduce but do not eliminate Intel’s capital risk
• Gross margin of 33.2% vs. the historical 55-60% range is the clearest single indicator of how far Intel has fallen — every percentage point of margin recovery requires manufacturing cost reduction, product mix improvement, and competitive pricing restoration
• Intel’s turnaround is a 5–7 year programme at minimum — not a 2–3 year story; investors are valuing a probability-weighted outcome between full recovery and significant strategic retrenchment

Intel (INTC) Business Model

Intel’s business model has been in strategic transition since 2021 and is best understood by contrasting what it was, what it is now, and what it is trying to become.

The Historical Model: IDM Monopoly Power

For most of Intel’s history (roughly 1990–2015), its business model was one of the most powerful in technology:

• Intel designed x86 processors that ran Windows, the dominant desktop and server operating system
• Intel manufactured those processors in its own fabs, which were technologically superior to any competitor’s
• AMD was the only licensed x86 competitor but chronically lagged Intel in manufacturing technology
• The combination of x86 software lock-in (decades of software compiled for x86 architecture), Windows dominance, and manufacturing superiority gave Intel nearly unchallenged pricing power in PC and server CPUs
• Gross margins of 55-65% on CPU products reflected this monopoly-like position

How Intel charged customers: Intel sells processors and chipsets to PC manufacturers (Dell, HP, Lenovo), server manufacturers (Dell, HPP, Supermicro), and through the retail/DIY channel. Pricing is driven by a tiered product line — Core i9 (premium), Core i7, Core i5, Core i3 (consumer) and Xeon (enterprise server) — with significant price differentiation by performance tier.

The Transition: IDM 2.0 and the Foundry Ambition

In March 2021, newly reinstated CEO Pat Gelsinger launched “IDM 2.0” — a strategic framework with three elements:

1. Regain process leadership: Intel committed to closing the manufacturing gap with TSMC through a defined roadmap of process nodes: Intel 7 (delivered, roughly equivalent to TSMC N7), Intel 4 (delivered), Intel 3 (ramping), Intel 20A (intermediate step), Intel 18A (the critical target, targeting leadership-class density and performance by 2025).

2. Use external foundries selectively: Rather than manufacturing all its chips internally, Intel would use TSMC for some products where TSMC’s process advantage was too large to overcome in time. This pragmatic acknowledgement that Intel’s fabs were behind was a significant cultural shift.

3. Open fabs to external customers (Intel Foundry Services): Intel would transform its US and European fabs into a commercial foundry business — manufacturing chips for fabless companies that currently rely on TSMC or Samsung. This would spread the enormous fixed cost of running advanced-node fabs over a larger revenue base, make the US government’s CHIPS Act investment logical (domestic foundry capacity), and eventually generate a standalone revenue stream.

The strategic logic is sound; the execution has been painful. Regaining manufacturing leadership requires: sustained R&D investment in process technology, construction of new fab capacity ($25B+ per year in capex), recruitment and retention of thousands of expert process engineers, and patience through years of losses before the investment pays off. Intel has struggled with all of these simultaneously.

The New Reality Under Lip-Bu Tan

Lip-Bu Tan’s appointment as CEO in March 2025 introduced several strategic reconsiderations:

• Products vs. Foundry separation: Tan has signalled openness to a more formal separation between Intel’s chip design business (Products) and its manufacturing business (Foundry) — potentially creating two subsidiaries or even two separately valued entities. This would allow each business to be evaluated on its own merits and could unlock valuation
• Foundry customer focus: Tan’s Cadence background (EDA software for chip design) gives him deep relationships with fabless chip designers who are potential Intel Foundry customers — a Rolodex that Gelsinger lacked
• Cost discipline: The new regime is prioritising cost reduction, headcount rationalisation, and capital allocation discipline over the aggressive “build everything simultaneously” approach of IDM 2.0
• 18A delivery: The singular near-term milestone remains 18A process node validation with external customers — Tan inherited this and its outcome will largely define his early tenure

Intel Competitors

Intel faces fundamentally different competitive challenges in each of its business segments:

Client Computing (PC processors) — AMD and Apple

AMD is Intel’s primary competitor in x86 PC processors and has been Intel’s most significant competitive threat since AMD’s Zen architecture launched in 2017. AMD’s Ryzen processors for consumers and EPYC processors for servers have consistently matched or exceeded Intel’s performance-per-watt since AMD switched to TSMC’s manufacturing. AMD’s competitive position: better process technology (TSMC N3/N5 vs. Intel’s Intel 7/4), better power efficiency in many workloads, and aggressive pricing.

Apple Silicon (M-series chips based on ARM architecture) displaced Intel from all Macs starting in 2020. Apple’s M1, M2, M3, and M4 chips have demonstrated performance-per-watt ratios that Intel’s x86 chips cannot match in mobile form factors — a humiliating loss of the premium laptop segment. The ARM architecture threat is broader than Apple: Qualcomm’s Snapdragon X Elite chips and other ARM-based designs are entering the Windows PC market with competitive performance claims.

Data Center (server CPUs) — AMD EPYC and Nvidia (different competition)

In server CPUs, AMD’s EPYC processors have taken substantial market share from Intel’s Xeon, particularly in cloud deployments. AWS, Google Cloud, and Microsoft Azure all deploy EPYC-based instances alongside Xeon. AMD’s advantage in server CPUs mirrors its advantage in consumer — better manufacturing technology translating to more cores per die, lower power consumption, and competitive pricing.

Nvidia is not a direct CPU competitor but represents a more profound threat: the shift of data centre spending from CPUs to GPUs for AI workloads. Every dollar hyperscalers spend on Nvidia H100/B200 GPUs for AI training and inference is a dollar not spent on Intel Xeons for traditional compute. Intel’s response — Gaudi AI accelerators — has not gained meaningful market traction.

Foundry — TSMC and Samsung

TSMC is the foundry competitor Intel is most directly targeting with IFS. TSMC has ~55% global foundry market share, a customer list including every major fabless chip company (Apple, Nvidia, AMD, Qualcomm, Broadcom), and a manufacturing process lead Intel is trying to close. Samsung Foundry (~15% market share) is the other major competitor.

For Intel Foundry to succeed, it must convince fabless chip companies to trust Intel as their manufacturer — despite Intel being a direct product competitor in CPUs and data centre chips. This conflict-of-interest concern is real and requires structural and contractual firewalls to overcome.

Revenue Breakdown

Financial data sourced from Intel SEC Filings. Note: Intel Foundry revenue includes internal intersegment revenue (sales from Foundry to Intel’s own product groups); external foundry revenue is significantly smaller.

Client Computing Group — 54% of Revenue ($29.3B)

Intel’s largest segment and its most resilient — CCG sells PC processors (Core series for consumers, Core Ultra for AI PC-capable laptops) and laptop/desktop platform chipsets to PC manufacturers and through retail.

Why CCG has been stable: Intel still commands approximately 70%+ unit share in x86 laptop and desktop processors for Windows PCs. AMD has made significant inroads — capturing 20–25% of PC CPU unit share — but Intel’s entrenched position with OEMs (Dell, HP, Lenovo), its supply chain relationships, and the sheer installed base of Intel-compatible software and systems provide enormous structural continuity.

The AI PC initiative: Intel has positioned its Core Ultra (Meteor Lake, Arrow Lake, Lunar Lake) processors around Neural Processing Units (NPUs) — on-chip AI accelerators that run local AI workloads (Microsoft Copilot, real-time transcription, camera effects) without sending data to the cloud. The “AI PC” category is Intel’s attempt to drive a PC upgrade cycle: if AI PC features require an NPU that only newer Intel (and AMD) chips have, consumers and enterprises have a reason to replace 3–5 year old machines. The success of the AI PC thesis is the most important near-term CCG growth driver.

The ARM threat in Windows: Qualcomm’s Snapdragon X Elite and Snapdragon X Plus chips (ARM-based, manufactured on TSMC) have entered Windows PCs with credible performance claims and superior power efficiency in some workloads. Microsoft’s Copilot+ PC initiative is agnostic to x86 vs. ARM — this is the first time ARM has had a genuine software-compatible path into Windows laptops. If Snapdragon X takes meaningful unit share, it would erode Intel’s most stable revenue base. Current ARM Windows share is small (~5%); the trajectory matters more than the current level.

Data Center & AI — 24% of Revenue ($12.8B, -17%)

Intel’s most troubled segment and the one where the competitive failure is most acute. DCAI sells Xeon server processors and Gaudi AI accelerators to cloud providers, enterprises, and high-performance computing users.

Xeon server CPU decline: AMD’s EPYC processors have taken meaningful server CPU market share, particularly in cloud data centres where total cost of ownership (performance per watt per dollar) is optimised ruthlessly. AWS has deployed AMD EPYC instances extensively; Google Cloud and Microsoft Azure offer EPYC options. Intel still has the majority of server CPU installed base but is losing incremental share. New Xeon (Granite Rapids) launched in late 2024 with competitive improvements, but restoring the performance leadership that drove Intel’s historical server pricing power requires the 18A process node.

Gaudi AI accelerator failure: Intel’s Gaudi AI training chips (Gaudi 2, Gaudi 3) were positioned as lower-cost alternatives to Nvidia H100 and A100 GPUs. Despite competitive pricing and Gaudi 3 offering solid raw performance numbers, enterprise and hyperscaler customers have overwhelmingly chosen Nvidia. The reasons: Nvidia’s CUDA software ecosystem has a decade of developer tools, libraries, and frameworks that Gaudi cannot match; AI teams are trained on PyTorch/CUDA workflows and do not want to retrain; and the cost of software migration exceeds the hardware cost savings. Gaudi revenue has been negligible — a $12B DCAI segment earns nearly all its revenue from Xeon, not AI accelerators.

Network & Edge — 11% of Revenue ($5.8B)

Processors and communications chips for telecommunications infrastructure (5G base stations, network routers), industrial automation, retail, and edge computing applications. This segment is stable but not a growth driver — it provides consistent cash generation without strategic excitement. Key products include the Xeon D (data centre and network), Atom (embedded/edge), and communications chips for radio access networks.

Intel Foundry — 8% of Revenue ($4.3B)

Critical accounting note: Intel Foundry’s reported $4.3B in revenue is overwhelmingly internal — it represents charges from the Foundry segment to Intel’s own product groups (CCG, DCAI, NEX) for manufacturing their chips. External foundry revenue — what Intel actually earns from third-party customers — is a small fraction of this figure and is not separately disclosed.

This accounting structure makes the Foundry segment appear larger than its commercial traction warrants. When evaluating Intel Foundry as a standalone business, the relevant question is: how much revenue is Intel generating from external customers who are choosing Intel Foundry over TSMC? The answer is currently very small — a handful of US government and defence customers, and early trials with a few commercial customers.

The Foundry segment reports a massive operating loss — approximately $7B in FY2024 — because it bears the full cost of running Intel’s fabs (depreciation of $25B+ in annual capex, fab employee costs, process R&D) while generating minimal external revenue. This operating loss is the primary driver of Intel’s overall net loss.

Mobileye — 3% of Revenue ($1.8B, -14%)

Mobileye (NASDAQ: MBLY) is Intel’s publicly traded subsidiary developing autonomous driving and ADAS (Advanced Driver Assistance Systems) technology. Intel took Mobileye public in October 2022 at a $17B valuation but retained ~88% ownership. Mobileye’s EyeQ system-on-chip and SuperVision/Chauffeur software platforms are used in ADAS systems at major automakers.

Revenue declined 14% in FY2024 as automakers drew down elevated chip inventory accumulated during the post-COVID supply chain disruption. Mobileye’s longer-term opportunity — if fully autonomous vehicles (SAE Level 4) reach commercial scale — is substantial, but the near-term is constrained by automaker inventory normalisation and the slower-than-expected progression of autonomy timelines. Intel has been exploring options for Mobileye including a full sale or further stake reduction.

Revenue Trend (3-Year)

The three-year trend tells the Intel story in three numbers: revenue declining from $63B to $54B, gross margin collapsing from 47% to 33%, and net income swinging from $8.0B profit to $1.6B loss. Every metric reflects the combination of competitive share loss (in DCAI to AMD and Nvidia), manufacturing cost inflation (running lagging-edge fabs at high cost), and the investment burden of IDM 2.0 (enormous capex and R&D).

Intel (INTC) Income Statement

Financial data sourced from Intel SEC Filings.

Intel (INTC) Key Financial Metrics

• Gross Margin: 33.2% — The most alarming single metric. Intel’s historical gross margin was 55–65%, reflecting the pricing power of its CPU monopoly and the cost efficiency of its leading-edge manufacturing. The collapse to 33.2% reflects three compounding factors: (1) manufacturing costs are high because Intel’s fabs are running lagging-edge processes with low yields and high per-wafer costs; (2) competitive pressure from AMD has forced Intel to cut CPU prices; (3) the Foundry segment’s massive losses are dragging on the blended margin. Recovery to 40%+ gross margins is possible only with successful 18A ramping and competitive product launches in Xeon

• Operating Margin: -4.8% — Intel spent more on R&D ($16.5B) and other operating expenses than its gross profit ($18.0B) covered. This is not entirely irrational — the R&D investment funds future process nodes and products — but it is unsustainable indefinitely. Intel has announced significant restructuring (15,000+ layoffs in 2024) to reduce the operating expense base

• Capital Expenditures: ~$25B — Intel is one of the most capital-intensive companies in the world, building new fabs in Arizona (Chandler), Ohio (New Albany), Germany (Magdeburg), and Israel. The $25B annual capex requires significant external funding support: US CHIPS Act grants ($8.5B), CHIPS Act loans ($11B), and international government subsidies from Germany and Israel. Without government support, the investment programme would be financially untenable at current profitability levels

• R&D Spending: $16.5B (30.4% of revenue) — Extraordinarily high as a percentage of revenue. This reflects the dual investment in product design (new CPU architectures, Gaudi AI, Mobileye) and process technology (18A development, future nodes). Intel cannot cut R&D without sacrificing the competitiveness of future products — creating a challenging dilemma as losses mount

• Restructuring: ~$2.2B in FY2024 — Intel announced the elimination of 15,000+ positions (approximately 15% of its global workforce) in August 2024, accompanied by capex reduction plans. These restructuring charges are one-time costs but reflect permanent workforce reduction aimed at right-sizing the cost base for the near-term revenue reality

• Net Debt — Intel carries significant debt on its balance sheet, used to fund the fab construction programme. As operating losses continue and capex remains high, debt management is a genuine concern. The CHIPS Act funding ($19.5B in combined grants and loans) provides critical relief but does not eliminate the capital risk

• Free Cash Flow: deeply negative — With $25B in capex and operating losses, Intel’s free cash flow is substantially negative. The company has suspended its dividend and is prioritising capital preservation and investment programme funding over shareholder returns

The 18A Process Node: The Make-or-Break Milestone

Intel 18A is the most important technology development in Intel’s history since the introduction of its original x86 microprocessor — and the outcome of its development will define whether the current turnaround is a success or a failure.

What Intel 18A is: A manufacturing process node targeting density, power efficiency, and performance comparable to (or exceeding) TSMC’s N2 and Samsung’s SF2 nodes. Intel 18A incorporates two significant architectural innovations:

• RibbonFET: Intel’s implementation of gate-all-around (GAA) transistor architecture, which provides better electrostatic control of the transistor channel than FinFET and enables better performance at lower power. TSMC is introducing its equivalent (GAAFET) in its N2 node. Both represent the same fundamental architectural transition happening industry-wide
• PowerVia: A backside power delivery network — routing power to transistors from beneath the silicon, rather than through the same interconnect layers as signals. This frees up routing resources on the top of the chip and reduces power delivery resistance, improving both performance and efficiency. PowerVia is an Intel-specific innovation; TSMC has its own backside power delivery in development

Why it matters commercially: If Intel 18A delivers on its density and performance specifications, it validates Intel’s entire foundry thesis:

• Intel’s own CPU products (Panther Lake for client, Clearwater Forest for servers) would be manufactured on a leading-edge process, restoring competitive positioning against AMD’s TSMC-made chips
• External foundry customers (fabless chip companies currently using TSMC) would have a credible alternative for future tape-outs, justifying Intel Foundry as a real business
• The US government’s $19.5B CHIPS Act investment would be vindicated as building genuine domestic semiconductor manufacturing capacity

If 18A disappoints: A failure or significant delay of Intel 18A would be catastrophic. It would mean Intel has spent $25B+ annually for four years without achieving manufacturing parity. The foundry business model would be untenable. Intel’s product teams would remain dependent on lagging manufacturing processes, and CPU competitiveness against AMD would not recover. The strategic options would narrow to either a full outsourcing model (abandoning IDM) or an asset sale scenario.

Current status: Intel has stated that 18A is on track for risk production in late 2025 and volume production in 2026. Microsoft has been named as an early 18A customer (designing an internal chip on Intel’s process). External validation — a second major external customer announcement — is the key pending milestone.

Is Intel Profitable?

No. Intel reported a net loss of $1.6 billion on $54.2 billion in revenue in FY2024 — a -2.9% net margin and the second consecutive year of declining profitability (FY2023 net income was $1.7B; FY2022 was $8.0B).

The losses are not primarily from revenue collapse — $54.2B is still a very large revenue base. They reflect the combination of: compressed gross margins (manufacturing inefficiency and competitive price pressure), massive R&D investment ($16.5B) that exceeds the gross profit available after paying for the cost of products sold at current margins, and restructuring charges.

Intel is not a company in financial distress — it has access to CHIPS Act funding, significant asset value in its fabs and IP, and sufficient liquidity to fund the turnaround programme. But it is a company whose path to profitability depends on execution of a multi-year technology programme (18A) that has no guaranteed outcome.

The profitability recovery thesis: If 18A succeeds, Intel’s manufacturing costs decline (better yields, more chips per wafer), Xeon regains competitive positioning (allowing price recovery), and Intel Foundry begins generating external revenue that spreads fixed costs — all of which expand gross margins back toward 40%+. With R&D flat or declining and restructuring complete, operating income would recover. This is a 2026–2028 scenario at the earliest under an optimistic case.

Intel (INTC): What to Watch

1. Intel 18A process node delivery — The single most critical milestone. Watch for: (a) Intel’s own Panther Lake client chip taping out and yielding successfully on 18A; (b) external customer announcements beyond Microsoft; (c) third-party validation from chip designers or analysts confirming 18A’s density and performance characteristics meet Intel’s claims. Any delay beyond H2 2025 for risk production would reset the entire recovery timeline

2. Intel Foundry external customer wins — Beyond Microsoft’s single disclosed 18A design, Intel needs multiple major fabless customers committing to Intel Foundry for future chip generations. Each win validates the foundry model and adds revenue that spreads fixed fab costs. Watch for any Foundry customer announcements at Intel Innovation, investor days, or in quarterly earnings calls. A Tier 1 fabless customer (Qualcomm, MediaTek, or a hyperscaler custom silicon team) would be a transformational signal

3. Lip-Bu Tan’s strategic direction — New CEO Tan has signalled potential structural changes including more formal separation of Products and Foundry businesses. Watch for any strategic announcements at Intel’s investor day (expected in 2025) about organisational structure, capital allocation priorities, and whether Tan maintains, modifies, or abandons elements of IDM 2.0. A decision to explore a Foundry spin-off or partial sale would be major news

4. Gross margin recovery trajectory — The clearest financial health indicator. Each quarter’s gross margin reveals whether manufacturing efficiency is improving (better yields on Intel 4/3 nodes), whether product pricing is recovering (less competitive pressure from AMD in client CPUs and Xeon), and whether the Foundry drag is stabilising. A sustained return to 40%+ gross margin would signal the turnaround is working

5. Mobileye strategic options — Intel has been exploring a full sale or further reduction of its ~88% Mobileye stake. A sale of Mobileye at a reasonable valuation would provide Intel with significant capital to fund the foundry programme without additional debt or equity dilution. Watch for any M&A activity around Mobileye and Intel’s public statements about its intentions for the stake

6. ARM/Snapdragon penetration in Windows PCs — If Qualcomm’s Snapdragon X chips (or future ARM-based Windows chips from MediaTek or Apple-inspired designs) take meaningful market share in Windows laptops, Intel’s most stable revenue base faces structural erosion. Watch quarterly PC market share data from IDC and Gartner for x86 vs. ARM unit splits, and OEM design win announcements for Snapdragon X vs. Intel Core Ultra laptops

7. CHIPS Act funding flow and conditions — Intel’s $8.5B in CHIPS Act grants and $11B in loans are subject to performance milestones and conditions set by the US Department of Commerce. Delays in fab construction, process node milestones, or changes in US government policy could affect the timing and amount of funding received. Any revision to CHIPS Act commitments would be significant for Intel’s capital plan

Intel (INTC) Financial Summary

Intel (INTC) is a semiconductors company that generated $54.2 billion in total revenue in fiscal year 2024 — flat year-over-year — while reporting a net loss of $1.6 billion and a gross margin of 33.2%. The company is executing the most complex turnaround in semiconductor history: simultaneously regaining manufacturing leadership (18A process node), rebuilding product competitiveness (Xeon, AI accelerators, AI PC), and launching a commercial foundry business (Intel Foundry) — all while reporting operating losses and investing $25B+ annually in capital expenditures.

The CEO change from Pat Gelsinger to Lip-Bu Tan in late 2024/early 2025 introduces strategic uncertainty but also fresh perspective from one of the semiconductor industry’s most respected executives. The outcome of Intel’s turnaround — full manufacturing recovery, partial recovery with strategic restructuring, or significant retrenchment — will determine whether INTC is a value opportunity or a value trap at current prices.

For broader semiconductor context, see How Nvidia Makes its Money, How AMD Makes its Money, and How TSMC Makes its Money.