Don’t let Intel (INTC) remain merely a “CPU company”: assessing the dual engine of products and manufacturing—and whether its restructuring will succeed

* This report is prepared based on data as of 2026-01-07.

What does Intel do? (for middle schoolers)

Intel designs, manufactures, and sells the “brains” inside computers and servers—computing components like CPUs. For decades, it has supplied the core compute silicon used in Windows PCs and enterprise servers.

Today, Intel is also taking on a second big mission: leveraging its massive in-house fabs to grow into a “foundry (contract manufacturer)” that builds chips designed by other companies. Put simply, Intel is both a “company that sells chips” and a “company trying to become a chip factory.”

Analogy: a restaurant and a central kitchen

If you think of Intel as a restaurant, it used to make money mainly by selling its own signature dishes (PC/server chips). Now it’s also trying to grow by becoming a “central kitchen (factory)” that cooks for other restaurants, too.

Who does it create value for? (three faces of the customer)

1) PC makers and the enterprise users beyond them

Intel sells CPUs to PC manufacturers, but the practical end customers are enterprises, schools, government agencies, and others that deploy PCs at scale. Intel has long been the “standard brain of Windows PCs,” and more recently it has been leaning into next-generation chips aligned with the shift toward on-device AI (AI PCs).

2) Companies that operate data centers

Intel supplies server CPUs to cloud providers, large enterprise IT organizations, telecom operators, and others. Deal sizes are large, and wins here can become a major profit pillar. At the same time, as AI use cases expand, competition often shifts from “CPU performance” to a broader contest spanning power efficiency, operations, and procurement diversification, which changes how vendors win in this market.

3) Companies that design chips (foundry customers)

Potential customers include fabless semiconductor companies and in-house design teams at large enterprises. Intel is clearly positioning this as a growth area, with the goal of making Intel Foundry (the manufacturing business) a major pillar over time.

How does it make money? (two-part revenue model)

(A) Selling its own product chips (product model)

The core model is to manufacture and sell chips such as CPUs. Scale can drive strong profitability, but profits are also highly sensitive to process transitions, product cycles, and competitive dynamics.

(B) Manufacturing other companies’ chips as a factory (contract manufacturing + related services)

Intel Foundry is aiming to deliver an end-to-end offering that includes not just manufacturing, but also “tooling such as design kits and IP,” plus advanced packaging (advanced assembly) that integrates multiple chips. Intel is also reshaping its internal structure so the product business and the factory business operate “as if they were separate companies,” moving toward a model where the fabs run closer to a stand-alone P&L with tighter cost discipline and faster improvement cycles.

Today’s profit pillars / tomorrow’s pillars (including future direction)

Current core businesses

• PC (Client): strengthening next-generation chips to capture the AI PC wave.
• Server (Data Center): high ASPs, but in the AI era the competitive set expands from “chips” to “operations, peripherals, and TCO,” making the battleground structurally tougher.
• Intel Foundry: still in incubation today, but strategically important enough to shape the company’s future.

Potential future pillars (three)

• Full-scale expansion of Intel Foundry: if the 18A volume ramp and customer pull-in for next-generation 14A progress, the center of gravity could shift toward a business that captures external growth.
• AI PC platform (on-device AI): a market where integration with the OS and applications becomes critical, increasing the odds of meaningful differentiation in PC chips.
• “Whole-rack” orientation for AI data centers: rather than competing on AI chips alone, Intel has signaled a shift toward end-to-end solutions that include surrounding components; if executed well it could be a starting point for a comeback, but the degree of difficulty is high.

Outside the business lines: internal infrastructure that will shape future competitiveness

Beyond rebuilding manufacturing processes (such as 18A), Intel is also changing how the fabs themselves are run. Product groups pay the fabs “as if at market prices,” while the fab organization pushes harder on cost discipline and improvement velocity. If this starts to work as intended, it could lift both product competitiveness and foundry competitiveness.

Why has it been chosen? (the core of its value proposition)

• Design capability and accumulated software/compatibility: in PCs and servers, fit with the surrounding software stack matters as much as the chip itself, which helps enterprises adopt with confidence.
• Keeping design and fabs close: tight coupling between process and product can translate into better performance and power efficiency. Intel is working to rebuild this advantage.
• Integrated offering of fabs + design enablement + advanced packaging: as a foundry, Intel is positioning itself as more than “just manufacturing,” aiming to reduce customer burden.

Growth drivers (structural tailwinds)

• AI PCs as a new replacement catalyst: the more workloads shift toward running AI on the PC, the more demand can build for next-generation chips.
• Need for fab diversification / value of U.S.-based manufacturing: efforts to reduce geographic concentration risk can become a tailwind.
• Expanding foundry customers anchored on leading-edge manufacturing (18A): whether external customers truly ramp is repeatedly highlighted as the key determinant of success.

That covers the “business understanding.” Next, we lay out the unavoidable “numbers narrative” for investors in the order of long term → short term → financials.

Long-term fundamentals: what is Intel’s “company type”?

Using Peter Lynch’s six categories as a lens, Intel is most naturally viewed as a hybrid of “Cyclicals (economic cycle)” × “Asset Plays”. That’s because two realities coexist: profits and cash flow can swing meaningfully, and the stock is often viewed through an asset lens given a PBR below 1x.

Revenue: modest contraction over the long term

Revenue 5-year CAGR (FY) is -5.90%, and 10-year CAGR (FY) is -0.51%. From the prior peak (2021: 79.0B), revenue fell to 53.1B in 2024.

EPS and FCF: long-term CAGR is difficult to assess (mix of losses and negative FCF)

EPS 5-year and 10-year CAGR (FY) cannot be calculated as a growth rate because the recent period includes a loss year (FY2024 EPS -4.38). This doesn’t mean “there is no growth,” but rather that the CAGR calculation isn’t valid when its underlying assumptions aren’t met.

Likewise, because FCF has been negative since 2022, 5-year and 10-year FCF CAGR cannot be calculated. In level terms, FCF declined from 20.9B in FY2020 to -15.7B in FY2024.

Profitability: sharp deterioration on an FY basis, with multiple negatives in 2024

• ROE (FY2024): -18.89%
• Operating margin (FY2024): -21.99%
• Net margin (FY2024): -35.32%
• FCF margin (FY2024): -29.48%

In FY2024, both accounting profit and free cash flow are negative, and long-term earnings power is in a weak phase.

What is driving the change in EPS? (Growth Attribution)

Looking across the FY time series, as revenue shifted into negative growth, margin deterioration became the dominant factor. The change in EPS is best described as a period where “margin factors” contributed more than “revenue factors.” Share count has trended down over time, but it is secondary relative to the scale of the recent P&L swings.

Explicit Lynch classification: why “Cyclicals × Asset Plays”?

Rationale as a Cyclicals

• FY EPS swings materially: 2021 4.86 → 2023 0.40 → 2024 -4.38
• TTM EPS growth rate (YoY) deteriorated sharply to -101.18%
• Inventory turnover (FY2024) is 2.93, consistent with a profile where efficiency can swing across the cycle

Rationale as an Asset Plays

• PBR (latest FY) is 0.86x, below 1x
• Cash ratio (latest FY) is 0.62, so it’s hard to argue short-term liquidity is severely depleted
• ROE is low/negative (FY2024 -18.89%), the opposite of a “high-ROE growth stock” profile

Where are we in the cycle? closer to bottom to early recovery (no definitive call)

On an FY basis, losses, negative FCF, and negative ROE overlap in 2022–2024—conditions that often show up near a cyclical bottom. Meanwhile, on a TTM basis, net income has recovered to slightly positive (TTM net income 0.198B), but TTM FCF is -8.42B, pointing to an uneven recovery path.

Short-term momentum: is the long-term “type” still intact near-term?

The latest assessment is Decelerating. This matters even for long-term investors because it tests whether the long-term “type” is starting to fray in the near term.

Last 1 year (TTM): revenue slightly down, but profit and cash are weak

• EPS (TTM) YoY: -101.18% (TTM EPS declined to 0.0437)
• Revenue (TTM) YoY: -1.49%
• FCF (TTM) YoY: -44.10% (TTM FCF -8.42B)

The defining near-term feature for Intel is “revenue isn’t collapsing, but EPS and FCF are clearly weak.”

Last 2 years (~8 quarters): direction of weakness is relatively clear, though cash includes pockets of improvement

• 2-year CAGR of EPS (TTM): -66.80%
• 2-year CAGR of revenue (TTM): -0.73%
• 2-year CAGR of net income (TTM): -65.76%

Over the last two years, EPS and revenue show a clear downward bias. FCF shows some signs of improvement at points, but it remains negative on a TTM basis, making it less likely to be the primary momentum driver right now.

Profitability momentum (FY): operating margin fell sharply over three years

• FY2022：3.70%
• FY2023：0.17%
• FY2024：-21.99%

The FY deterioration is substantial, consistent with the weakness in near-term EPS.

Differences between FY and TTM (important)

FY results show pronounced losses and negative profitability, while TTM net income has recovered to slightly positive. This isn’t a contradiction—it’s a difference in what each measurement window captures—and it’s reasonable to treat it as evidence that the recovery is not uniform.

Financial soundness: how to frame bankruptcy risk (no definitive call; structural view)

Intel is in a period of heavy investment, and with profits and cash under pressure, the debt optics have become more challenging. Rather than making a bullish or bearish call, it’s more useful to frame the situation across three dimensions: interest-paying capacity, cash cushion, and leverage.

• Debt/Equity (latest FY): 0.50 (not extreme as a ratio, but it can feel heavy when earnings are weak)
• Interest coverage (latest FY): -12.35 (consistent with weak profitability, and numerically implies very tight interest-paying capacity)
• Cash ratio (latest FY): 0.62 (not enough to conclude cash is depleted, but also not a strong comfort factor)
• Net Debt / EBITDA (latest FY): 23.23x (may reflect EBITDA compression in the denominator, but it screens as meaningful debt pressure)

These figures don’t justify an immediate conclusion about bankruptcy risk, but they do describe a phase where “investment burden × weak cash generation × weak interest-paying capacity” overlap, making close monitoring important.

CapEx burden: why cash is hard to retain

In this manufacturing rebuild phase, investment is heavy relative to operating cash flow. CapEx/OCF rose from 1.62 in FY2022 → 2.24 in FY2023 → 2.89 in FY2024, extending a period where “investment tends to exceed cash earned.”

Dividends and capital allocation: should it be viewed as an income stock?

Dividends are not the “main character,” but one element of capital allocation

Dividend yield (TTM) is approximately 1.05%. Dividends aren’t zero, but this isn’t typically a yield level that anchors an income thesis. That said, Intel has paid dividends for 33 consecutive years, so “continuity” and “near-term sustainability” remain central questions.

Versus historical averages: yield is lower than the past 5-year and 10-year averages

• Past 5-year average yield: approximately 2.59%
• Past 10-year average yield: approximately 3.42%

The latest yield of 1.05% is below the company’s own historical average (this is not a market or peer comparison).

Dividend growth: the decline trend continues

• DPS 5-year CAGR: -21.42%
• DPS 10-year CAGR: -8.13%
• Latest 1 year (TTM) YoY: -28.76%

Based on the data, it’s hard to frame Intel as a “designed to keep increasing” dividend growth story.

Dividend safety: weak earnings and FCF make the ratios look harsh

• Earnings-based payout ratio (TTM): approximately 807.6% (the ratio looks extreme because TTM EPS is very small)
• FCF (TTM): -8.42B, so the FCF-based payout ratio is negative by definition
• FCF coverage (TTM): -5.26x, indicating dividends are not covered by free cash flow

In short, the data makes it difficult to argue dividend safety is high; the profile skews toward instability. The issue is not “because the dividend is large,” but because Intel is in a period of weak earnings and weak FCF, which makes the ratios look severe.

Track record: long payment history, but no streak of dividend increases, and a recent dividend cut

• Consecutive dividend years: 33 years
• Consecutive dividend increase years: 0 years
• Record of a dividend reduction/cut in 2024 (fact)

On peer comparison (within what can be said from this material)

Because specific numerical peer data isn’t provided, a definitive ranking can’t be made. In general, semiconductors aren’t necessarily a high-dividend sector; Intel’s latest yield is about 1% and below its own historical average, and FCF is negative. As a result, even in a peer comparison, the core issue is likely dividend sustainability (alignment with earnings power and cash generation) rather than headline yield.

Fit with investor types (Investor Fit)

• For income investors, it’s hard to argue the yield is high, and this isn’t a period with strong earnings/cash capacity to support it, so there is limited basis to make dividends the primary objective.
• For total-return investors, dividends are present, but business recovery and improved cash generation are more likely to be what ultimately supports dividend stability.

Where valuation stands today (positioning vs. its own history)

Here, without comparing to the market or peers, we place Intel’s current six metrics within its own historical distribution (primarily the past 5 years, with the past 10 years as a supplement). We do not draw a conclusion (definitively cheap/expensive).

PEG: negative, making it difficult to compare to the historical “positive range”

PEG is -8.90. Because the latest EPS growth rate (TTM YoY) is negative, PEG is negative, which makes it hard to judge positioning versus the “positive PEG range” that dominated the past 5 and 10 years.

P/E: far above the typical 5-year and 10-year range (but reflects extremely small earnings)

P/E (TTM) is 900.92x, far above the typical range over the past 5 and 10 years. This is driven by extremely small TTM EPS of 0.0437, which effectively makes P/E a poor metric in the current setup.

Free cash flow yield: within the 5-year range, but below the 10-year lower bound

FCF yield (TTM) is -4.48%. It sits within the past 5-year range, but over the past 10 years it is below the typical lower bound. Over the last two years, the trajectory shows the negative magnitude narrowing (moving upward), but it remains negative today.

ROE: below the typical 5-year and 10-year range

ROE (FY2024) is -18.89%, below the typical range over the past 5 and 10 years. Over the last several FY periods, the trend has been downward.

FCF margin: on the lower side within the 5-year range, and also low within the 10-year range

FCF margin (TTM) is -15.75%. It is within the past 5-year range but toward the weaker end, and it is also low versus the typical range over the past 10 years. The last two years show an improving (upward) trend.

Net Debt / EBITDA: far above the typical 5-year and 10-year range

Net Debt / EBITDA is an inverse metric where the smaller (the more negative), the greater the financial flexibility. Against that backdrop, the latest FY level of 23.23x is far above the typical range over the past 5 and 10 years. Over the last two years, the trend is upward (i.e., moving toward what mathematically screens as higher debt pressure).

Cash flow tendencies: are EPS and FCF consistent?

Recently, accounting profit (TTM net income recovering to slightly positive) and free cash flow (TTM -8.42B) have not improved in tandem. In other words, this is a period where “profits may start to return, but cash still doesn’t stick” due to investment burden and working capital dynamics.

FY FCF has been negative since 2022, which can be framed as the overlap of heavy investment and weak profitability. As a result, the near-term deceleration isn’t necessarily explained only by “business deterioration,” but also reflects a phase where cash pressure from rebuilding (manufacturing turnaround and investment) can be pronounced.

Success story: why Intel has won (the essence)

Intel’s historical success wasn’t just about “building a good CPU.” Over many years, it became deeply embedded in the computing foundations of PCs and servers, accumulating compatibility, operational know-how, and peripheral optimization—creating indirect network effects.

In addition, owning both design and manufacturing (fabs) can—when executed well—create a reinforcing loop where products drive utilization, utilization drives the learning curve, and manufacturing improvements feed back into product competitiveness. Intel’s structural value is that it holds this two-layer stack of computing foundation × manufacturing foundation at the same time.

Continuity of the story: are recent strategies consistent with the success story?

In recent years, Intel’s narrative has shifted from “king of CPUs” to “a rebuild story that includes manufacturing.” Expectations also tend to move away from “near-term profits” and toward conditions for a successful ramp—18A volume production, 14A customer pull-in, yields, and external customer wins.

In AI data centers, Intel has also signaled a shift from chip-only competition toward a rack/system orientation (including the decision not to productize an AI chip that had initially been planned). This aligns with the historical success story—owning the foundation—as an attempt to change how Intel wins as “the unit of selection” changes. However, the execution bar is higher.

Invisible Fragility: points that can look strong yet still break

Without claiming anything “breaks tomorrow,” this section highlights structural weaknesses that can be easy to miss but potentially high impact.

• Customer concentration on the PC side: dependence on large OEMs can be meaningful, and shifts in customer design preferences can quickly impact volume and mix.
• Normalization of competitive procurement in data centers: even without a sharp revenue decline, margins and cash generation can be gradually eroded (consistent with today’s profile of “revenue not collapsing, but profit and FCF weak”).
• More differentiation requirements: as competition becomes a broader contest spanning power, supply, compatibility, and operations—not just standalone performance—paths to winning get more complex and the cost of missteps rises.
• Supply chain dependence (single-source): EUV lithography tools are effectively dependent on a specific company, and supply delays can disrupt node ramps. Intel can also face “dual dependence,” where some leading-edge products rely on external fabs and materials.
• Organizational culture side effects: large-scale layoffs and restructuring can create near-term friction, with potential lagged effects on technical accumulation, quality, and execution speed.
• Profitability deterioration makes recovery calls harder: if investment burden persists while ROE, operating margin, and FCF remain weak, “flat revenue” can coexist with “weak profit and cash,” making deterioration easier to recognize late.
• Deteriorating interest-paying capacity can bite with a lag: if investment continues through a period of weak profit and cash, the company may be forced into choices around capital allocation and investment pacing.
• Leading-edge nodes are a customer-acquisition game: beyond technology, landing emblematic large external customers can pull in an ecosystem; failing to do so raises the difficulty of sustaining leading-edge investment.

Competitive landscape: Intel is fighting “three games” at once

In practice, Intel is competing across three parallel arenas within semiconductors.

• PC CPUs: a multi-factor battle across refresh demand, power efficiency, AI processing (NPU, etc.), and price
• Server CPUs: a comprehensive procurement contest that includes performance, power efficiency, operations, and diversification of supply
• Foundry (contract manufacturing): a “trust game” that spans process technology, design kits, EDA, IP, advanced packaging, volume-production track record, and repeat orders

These three arenas are interconnected. If internal manufacturing is proven through Intel’s own products, it becomes a track record; but if external customers don’t materialize, fab utilization and the learning curve are less likely to scale.

Key competitors (where they collide)

• AMD: direct competition in PC CPUs and server CPUs
• NVIDIA: builds a GPU-centered platform in AI data centers and can shift the unit of competition toward racks/systems
• Qualcomm: can become a comparison point in parts of Windows PCs (thin-and-light, power efficiency, always-connected, etc.)
• Apple: a representative example of in-house SoC design in PCs (not an Intel customer, but a symbol of substitution)
• TSMC / Samsung Foundry: competitors to Intel Foundry (standard options for customers)
• Arm ecosystem (Ampere, cloud providers’ in-house CPUs, etc.): encourages procurement diversification as an alternative data-center architecture

What customers can readily value (Top 3)

• Confidence from compatibility and stable operations (especially enterprise PCs)
• Comprehensive proposals across design × manufacturing × advanced packaging
• Potential option for U.S.-based manufacturing and procurement diversification

What customers are likely to be dissatisfied with (Top 3)

• Roadmap uncertainty (hard to predict the timing of generational transitions)
• Harsher comparisons in data centers as alternative options increase
• Time required to build trust in foundry (EDA, IP, yield, volume scale, consistency of commitments)

Moat (barriers to entry) and durability: strong, but “conditional”

Leading-edge CPUs and leading-edge manufacturing come with extremely high barriers to entry, and Intel has real assets—fabs, process know-how, and supply networks—that can underpin a moat.

However, in the AI era, the GPU/accelerator ecosystem is more likely to become the center of gravity, and differentiation in CPUs alone can narrow. The most natural framing is that Intel’s moat “exists, but its durability depends heavily on execution (stable volume production, winning external customers, and roadmap consistency).”

Structural positioning in the AI era: tailwinds exist, but not automatic

Intel’s AI-era positioning is fundamentally a rebuild aimed at capturing two layers at once: “middle (computing foundation) × middle (manufacturing foundation)”.

• Computing foundation: as AI PCs (on-device inference) spread, there is room to deepen integration on the PC side and build adoption.
• Manufacturing foundation: as demand for supply-chain diversification rises, successful 18A volume production and 14A customer pull-in could open a path to external sales.

On the other hand, AI data centers are often led by “racks/platforms” rather than “chips,” and as value shifts across the stack, Intel may be more exposed to pricing pressure—not so much being replaced by AI, but facing tougher competitive procurement.

Leadership and culture: will the speed to “finish the rebuild” increase?

CEO Lip-Bu Tan (assumed office in March 2025): objectives

CEO Lip-Bu Tan is emphasizing less bureaucracy and faster decision-making, consistent with Intel’s rebuild agenda (design × manufacturing × volume execution). The direction is to cut unnecessary meetings and internal administrative work, speed decisions through live dashboards, and return time to engineering and customer value.

Profile (organized across four axes)

• Vision: transition to a leaner, faster Intel
• Personality tendencies: operations-focused, speed-focused, inclined toward integration and flattening
• Values: customer-centric, engineer-centric, execution and visibility
• Priorities: prioritize speed, execution, customers, and engineer productivity; reduce excessive meetings and formalistic work

How it may show up in culture, and potential friction

Fewer meetings and a return-to-office posture align with the goal of increasing collaboration density and execution speed. At the same time, reforms that include reorganization and layoffs can reduce psychological safety and raise fatigue in the short term; culture often shows friction before it improves, making this an important monitoring item.

Adaptability to technology and industry change (organizational design)

Intel’s challenge isn’t only technology—it’s the execution intensity required to run technology, manufacturing, and customer acquisition simultaneously. Building a central engineering organization and a structure to drive custom silicon for external customers aligns with the goal of winning foundry customers through a value proposition that includes design enablement.

Fit with long-term investors (culture and governance lens)

• Potential positives: the weaker profits and cash are, the more “can they execute through it” becomes the core question, and Tan’s actions are aimed at building execution consistency.
• Cautions: the more reforms involve personnel changes, the more culture can destabilize in the short term, and the board–CEO relationship becomes an important transition-phase variable.
• Culture KPIs (qualitative monitoring items): effectiveness of meeting reductions, cross-functional execution bottlenecks in foundry/custom engagements, signs of key-person attrition after reorganization.

“Two-minute” long-term investment skeleton (Two-minute Drill)

The long-term way to underwrite Intel isn’t just “did it ship a good CPU.” The central question is whether Intel can rebuild both the computing foundation (PCs/servers) and the manufacturing foundation (leading-edge foundry) at the same time—and get the two engines working together.

• Product-side hypothesis: even in a tougher competitive procurement environment, can Intel create reasons for adoption to compound across power, operations, compatibility, and supply?
• Manufacturing-side hypothesis: does external customer acquisition move from “headline” to “volume production and repeat orders” (test chips and sustained volume are not the same)?
• Organization-side hypothesis: do execution speed and prioritization improve, reducing roadmap volatility?

In the numbers, FY shows overlapping losses and negative FCF, while TTM shows net income recovering to slightly positive but FCF still negative—an “uneven recovery.” As a result, what long-term investors should emphasize is less “is the story right” and more is the process advancing (stable volume production, external customer continuity, and simultaneous improvement in profits and cash).

Anchor with a KPI tree: what to watch to capture Intel’s change early

Ultimate outcomes (Outcome)

• Profit generation capacity (return to profitability and sustainability)
• Cash generation capacity (stabilization of free cash flow)
• Capital efficiency (ROE, etc.)
• Financial stability (resilience to investment burden and debt burden)
• Long-term competitive durability (whether it continues to be chosen in both products and manufacturing)

Intermediate KPIs (Value Drivers)

• Revenue (volume × ASP × mix) and fixed-cost absorption
• Gross profit (aggregation point of generation, yield, and mix)
• Operating margin (absorption of R&D, SG&A, and manufacturing fixed costs)
• Cash conversion (linkage from earnings → operating CF → FCF)
• CapEx burden (manufacturing rebuild and leading-edge investment)
• Working capital efficiency (inventory directly links to profitability and cash)
• Manufacturing execution (yield, utilization, quality)
• External customer acquisition and retention (Foundry utilization and learning curve)
• Adaptation to changes in how it is chosen (chip → system/operations/supply network)

Constraints and bottleneck hypotheses (Monitoring Points)

• Investment burden is heavy, making profits and cash more likely to diverge
• Supply constraints in leading-edge manufacturing (dependence on EUV, etc.) can affect ramps
• Demand cyclicality and competitive procurement can pressure margins
• Foundry has a time lag in trust-building (adoption → volume production → repeat orders)
• Friction from organizational reform can affect execution speed

• Whether volume-production ramps are accumulating as planned
• Whether external foundry customers are increasing on a “repeat” basis
• Whether fab utilization and fixed-cost absorption are improving
• Whether profit improvement and cash improvement progress simultaneously
• Whether distortions are emerging in inventory and supply planning
• Whether adoption patterns in data centers shift toward full refresh or partial substitution
• Whether Intel can expand the unit of proposal beyond CPUs alone
• Whether meeting reductions and shorter decision cycles are cascading into frontline execution