Intel Q1 26: Growth Trap

Executive Summary

We initiate coverage of Intel with a SELL call and a FY28 price target of USD 65, representing -42% downside from today’s price of USD 113.

The stock has rallied +466% in 12 months and now embeds an outcome where Intel Foundry approaches breakeven and external revenue scales materially, neither supported by management's own roadmap nor by the bottom-up segment math.

While operational execution has improved materially, we believe the current valuation already running ahead of an optimistic Foundry outcome.. The market is extrapolating a future financial profile that still lacks sufficient operational evidence

What has happened since our last update

Looking back at our June and August 2025 notes, the operational call has aged well, but the price call missed badly. We held a USD 25 Neutral target while the stock rallied from ~USD 21 to USD 113, a +438% move, underestimating how aggressively the market would re-rate Intel on Lip-Bu Tan's six consecutive EPS beats, the USD 5 billion NVIDIA equity stake, and the AI-driven CPU narrative.

FY25 EPS came in at USD 0.42 versus our USD 0.13–0.32 forecast, and our 24x Fair P/E proved too conservative for the post-rally environment, which is why we now apply 30x. On the operational side, however, our cautious view has held up well: Foundry remains a structural revenue drag, gross margin recovery is tracking the 2026+ story we flagged (Q1A 41% / Q2 guide 39%), 18A hit high-volume manufacturing on schedule, external Foundry customer wins remain vague, and AMD share gains and ARM threats continue to play out as previewed.

Operational concerns regarding Foundry profitability, margin recovery, and competitive positioning have broadly aligned with our earlier expectations. However, market re-rating driven by AI enthusiasm and improving sentiment materially exceeded our prior assumptions..

The same caution that justified Neutral at USD 21 now justifies Underweight at USD 113, because the stock is pricing in optimistic Foundry success while the underlying business continues to track our pre-rally base case.

Figure 1: Intel share price

Q1 2026 earnings summary

Intel delivered a clear beat across every line, revenue, gross margin and EPS all printed above the high end of guidance, marking the sixth consecutive quarter of exceeding financial expectations. The print sent shares up as much as 20% in after-hours trading and resulted in the second-largest single-day re-rating Intel has seen in the past five years. CFO Zinsner's framing on the call captured the shift: "A year ago the conversation about Intel was about whether we could survive. Today it's about how quickly we can add manufacturing capacity and scale our supply to meet enormous demand."

Looking the above, while sentiment has improved materially, we note that earnings quality remains heavily dependent on internal Foundry accounting, AI-related demand assumptions, and margin normalisation that is still in early stages.

Table 1: Q1 2026 earnings summary

(USD bil)	Q1 2026A	Notes
Revenue	13.58	AI-driven businesses now ~60% of revenue, +40% YoY
Client Computing Group (CCG)	7.7	AI PC mix >60% of client CPUs; AI PC revenue +8% QoQ
Datacenter & AI (DCAI)	5.1	Fastest-growing segment; sustained CPU demand
Intel Foundry (gross)	5.4	Mostly internal wafer revenue; better-than-expected yields
All Other (Mobileye, IMS)	0.63	Altera deconsolidated Sep-2025 (one-off)
Non-GAAP EPS	0.29	Includes ~$2.1bn restructuring
Non-GAAP Gross Margin	41%	Include one off inventory boost
Operating Cash Flow	0.5
Adjusted FCF	-2

Source: Bloomberg Finance L.P., Intel, iFAST Compilation, Data as of 6 May 2026.

Intel is the world's largest integrated device manufacturer (IDM) by revenue, designing and manufacturing x86 microprocessors, server CPUs, AI accelerators (Gaudi), client SoCs, and fab services for external customers. The company operates two reporting groups, Intel Products (CCG, DCAI, plus All Other, including Mobileye and IMS) and Intel Foundry (the manufacturing arm, restructured into a separate operating segment in 2024 with its own P&L). The dual structure was designed to enable external customer wafer business while the internal product business remains the anchor demand.

Following the Q1 2026 segment reorganisation, Network & Edge (NEX) revenue has been folded into CCG and DCAI, lifting the historical revenue base by roughly USD 5.5 bil. The reorganisation has not yet produced material external Foundry revenue, Intel Foundry is largely internal wafer revenue today (~USD 5.4 bil in Q1 2026, of which we estimate <$0.5bn is external).

Figure 2: Intel’s revenue segmentation as of Q1 2026

Intel’s Q2 2026 guidance

Table 2: Intel’s Q2 2026 management guidance

(USD mil)	Q2 guidance	Q1 actual	y/y change	Notes
Revenue	13,800-14,800	13,600	up 7.2%	14,000 above Jan outlook
Gross Margin	39%	41%	up 1.8 bps	6.5 ppts above Jan outlook
EPS	$0.2	$0.29	up $0.16	$0.29 above Jan outlook

Source: Intel, iFAST Compilation, Data as of 7 May 2026.

Management guided Q2 2026 revenue of USD 13.8–14.8 billion, non-GAAP EPS of $0.20. Gross margin guide of 39% (vs Q1A 41%) is the single most important number in the release. It tells us: (i) Q1 41% included one-time inventory benefit, not a step-change; (ii) the structural FY26 GM corridor is 38–40%; and (iii) the FY27 step-up to 44% in our model carries genuine forecast risk if Foundry losses do not narrow as scheduled. The H2 2026 trajectory remains the swing factor; management framed factory output as 'continuing to increase' but at a 'more measured pace than we anticipated 90 days ago'.

We expect gross margin to hover at ~40%

Intel is structurally the most capital-intensive business in the semiconductor stack ex-foundry. Capex runs at ~28–32% of revenue (vs <5% for fabless peers). Depreciation runs at ~USD 13 bil p.a. through our forecast, a permanent drag on gross margin that fabless competitors simply do not carry. The hybrid IDM-plus-foundry construct adds a second margin headwind: Intel Foundry currently runs at structural operating losses we estimate at –USD 8.5 bil FY26E narrowing to –USD 3.0 bil FY28E. Until the Foundry breaks even, which management has guided is multiple quarters beyond our forecast horizon, Intel's blended gross margin is structurally capped in the mid-40s.

CPU/GPU ratio compression, the bull case argument

The thesis circulating among AI hyperscale watchers is straightforward: as AI training and inference clusters scale, the historical CPU-to-GPU ratio of roughly 1:8 (one host CPU shepherding eight accelerators) is compressing toward 1:4 or even 1:2. The drivers are real. Inference workloads, which now dominate hyperscaler capex versus training, are chattier, they need more pre-processing, vector database lookups, agentic orchestration, and retrieval-augmented generation steps that all run on CPU, not GPU. NVIDIA's own GB200 NVL72 rack architecture pairs 36 Grace CPUs with 72 Blackwell GPUs (a 1:2 ratio), and the next-generation Vera Rubin platform reportedly tightens this further. If hyperscalers are buying USD 400B+ of GPU capacity in 2026 (NVIDIA's data centre revenue trajectory implies this), and every GPU now needs roughly 2x more CPU compute behind it than two years ago, that's an enormous tailwind for the CPU TAM and Intel, as the dominant x86 server CPU vendor with ~70% share, is the structurally largest beneficiary.

However, the key debate is not whether AI infrastructure increases CPU demand, but whether Intel can monetise that demand sustainably amid competitive share pressures..

Even if AI server CPU TAM doubles from USD 25B to USD 50B over FY26-FY28, Intel needs three things to convert: (i) hold or take share against AMD EPYC, which has been steadily winning hyperscale sockets for five consecutive years and now sits at ~30% share; (ii) match Granite Rapids and Diamond Rapids product roadmap to AMD Turin and Venice on a per-socket performance basis; (iii) avoid being designed-out by ARM-based custom silicon (AWS Graviton, Google Axion, Microsoft Cobalt) which is the genuine long-term threat in hyperscale. Our Base case DCAI forecast (USD 19.8B FY26, USD 22.5B FY27, USD 25.5B FY28), implying ~14% CAGR, already bakes in meaningful CPU/GPU ratio compression.

To put that in context: Intel's reported FY25 DCAI was USD 16.9B, so we're modelling ~usd 8.6B of cumulative growth over three years, which is roughly Intel capturing one-third of the incremental hyperscale CPU TAM expansion. That's a credible base case but not a heroic one.

Figure 3: Intel DCAI Revenue, actual and forecast

Figure 4: Intel’s Disaggregation Journey

Source: Intel, iFAST Compilation, Data as of 2025.

Process node roadmap and positioning vs foundry peers

Intel's process roadmap pivots on three nodes: Intel 18A (HVM as of Q1 2026), Intel 14A (HVM 2028-29), and the in-between Intel 3 / Intel 4 nodes that anchor the current cost structure. 18A is the first node combining RibbonFET gate-all-around transistors with PowerVia backside power delivery. Intel is first to market on BSPD, ahead of TSMC's analogous offering on N2P/A16 (2026-27). PowerVia delivers ~5-8% performance uplift at iso-power and frees up frontside routing density. RibbonFET, Intel's nanosheet GAA architecture, is in line with TSMC N2 and Samsung SF2 in approach but earlier in production maturity. Panther Lake (Core Ultra Series 3) shipped on 18A at CES January 2026; Clearwater Forest (E-core server, 288 cores) and Diamond Rapids (P-core flagship server, up to 256 cores) launch H2 2026.

Figure 5: Intel Diamond Rapids Architecture Changes

Source: Semianalysis, iFAST Compilation, Data as of Feb 2026

14A is the strategic bet. It will be the first Intel node to use High-NA EUV lithography in production, TSMC is holding off on High-NA until later nodes. Intel has indicated 'multiple external customer evaluations' on 14A PDKs. Two prospective customers are expected to make binding commitments in H2 2026; we treat these as binary outcomes that materially shape the FY27+ external Foundry trajectory. If both convert at meaningful volume, the Foundry breakeven timeline pulls forward. Yet if they slip, the segment continues to lose money through the explicit forecast horizon.

Intel's advanced packaging franchise is increasingly the underrated asset. EMIB (Embedded Multi-Die Interconnect Bridge) competes directly with TSMC CoWoS-L for AI accelerator integration; second-generation EMIB 3.5D delivers a 45-micron bump pitch at competitive density. Foveros Direct 3D enables vertical die stacking with sub-25-micron bond pitches. Management upgraded the 2026 packaging revenue outlook from 'hundreds of millions' to 'billions of dollars per year' on the Q1 call, a meaningful repositioning. The Google/EMIB partnership disclosed in Q1 (with TPU advanced packaging) is the marquee external validation. We model packaging revenue scaling from <USD <0.5 bil FY25 to ~USD 2 bil FY28E, embedded inside the Foundry segment.

Intel has narrowed the process gap meaningfully. PowerVia is genuinely a year ahead of TSMC. RibbonFET 18A appears to be performing in line with Intel's own commitments. But process leadership in the foundry business is necessary, not sufficient: TSMC's structural advantage is in customer ecosystem, PDK quality, IP library breadth, yield consistency at scale, and 25 years of fabless co-design relationships. Intel cannot close that gap in two product cycles. The realistic best case for Intel Foundry is a credible #2 player with a defensible niche in BSPD-sensitive workloads (AI accelerators, dense networking, mobile high-end), not parity with TSMC.

Acknowledging the recovery path, management has been explicit that Foundry breakeven is 'multiple, multiple quarters'. This is what our Base case prices in, and why we view current spot as already capturing the upside.

Figure 6: Intel 18A with RibbonFET and PowerVia

Source: Intel, iFAST Compilation, Data as of 2025

Table 3: Competitive positioning vs TSMC/Samsung

Dimension	Intel	TSMC	Samsung
Leading-edge node in HVM (May 2026)	18A (Q1 2026 HVM)	N2 (H2 2025 HVM)	SF2 (2025 ramp, behind plan)
Backside power delivery	Yes (PowerVia, 18A) — first to market	From N2P/A16 (2026-27)	Roadmap — undisclosed timeline
High-NA EUV	First on 14A (FY28-29)	Holding off until later nodes	No public commitment
Tier-1 external foundry customers	None at volume	Apple, NVIDIA, AMD, Qualcomm, MediaTek, Broadcom	Tesla, Ambarella, NVIDIA (legacy)

Source: Claude, Company reports, iFAST Compilation, Data as of 6 May 2026.

Figure 7: Intel Foundry Revenue, actual and forecast

FY28 EPS justify earnings upside, but not share price

The single most important reason not to own Intel at USD 113 is that the equity has rallied ~+240% in 12 months on operational improvement that, while real, has not yet translated into a financial profile that justifies a 144x NTM multiple. The market has priced in Foundry success that management has explicitly described as multi-year. Six EPS beats do not equal a Foundry win.

Until we see (a) a tier-1 14A customer commitment at material volume, (b) a sustained gross margin step-up beyond 42%, and (c) inflection to sustained positive FCF, the risk-reward at spot is asymmetric to the downside.

Putting this together, we expect the CCG segment to remain broadly flat, with weaker PC unit demand offset by improving AI PC mix.. Besides, DCAI is growing mid-teens on AI server CPU demand and CPU/GPU ratio compression; Foundry external scaling but losses persist; All Other resilient on Mobileye. Gross margin reaches 46%, reflecting narrowing Foundry losses but not eliminated.

For more information, kindly find the table below.

Table 4: Intel’s income statement, actual and forecast

(USD mil)	FY24A	FY25A	FY26E	FY27E	FY28E
Total Revenue	53,185	52,800	54,900	59,700	65,200
CCG	33,200	32,200	31,600	32,200	33,500
DCAI	16,100	16,900	19,800	22,500	25,500
Intel Foundry (gross)	17,500	17,800	19,500	21,500	23,500
All Other	3,600	3,600	2,800	3,000	3,200
Intersegment Eliminations	-17,215	-17,700	-18,800	-19,500	-20,500
COGS	-32,709	-33,475	-33,489	-33,432	-35,208
Gross Profit	20,476	19,325	21,411	26,268	29,992
R&D	-16,221	-12,408	-11,529	-11,343	-11,410
SG&A	-4,521	-4,224	-3,843	-4,060	-4,042
Operating Income (EBIT)	-266	2,693	6,039	10,865	14,540
Net Interest Income/(Expense)	-200	-300	-700	-500	-300
Pre-Tax Income	-466	2,393	5,339	10,365	14,240
Income Tax Expense	84	-359	-587	-1,348	-1,994
Non-Controlling Interest	100	-200	-750	-1,100	-1,100
Net Income (Non-GAAP)	-282	1,834	4,002	7,918	11,146
EPS	-0.07	0.36	0.78	1.55	2.17

Source: Bloomberg Finance L.P., iFAST Compilation, Data as of 7 May 2026.

Revised upwardly Fair PE to 30x from 24x

Public statements from the Trump administration in late April 2026 included endorsement of Intel's domestic manufacturing footprint and references to 'strategic sovereign capital' for US chip champions. Press reports of potential government or sovereign-fund participation in Intel, most prominently Apple/Google equity speculation in late April, have contributed materially to the recent rally. Our base case assumes this has contributed to a US premium, where it structurally attracts potential partnerships within the US supply chains.

On top of that, Intel traded at 12-18x NTM P/E during the 2018-2022 'mature IDM' period; the 30x reflects a partial re-rating to a growth multiple, but is materially below where the stock now trades on TTM (144x). This reflects Intel's hybrid status: a recovering IDM with optional Foundry upside, but neither pure-play growth nor a settled mature IDM.

Sell, TP at $65 with -42% downside potential

The catalyst path through 2026 is asymmetrically weighted toward the downside. We acknowledge the earnings are on an ongoing recovery path (both CPU and foundry), and we are positive about that, yet we see the current share price has already reflected way ahead, and any slippage in execution could trigger a material de-rating and pose a pitfall.

Importantly, our negative stance is not based on denying Intel’s operational improvement. Product execution has stabilised, the credibility of the process roadmap has improved, and AI-related demand tailwinds are increasingly visible across both client and server segments. The key issue is that the current valuation now discounts a much stronger future financial profile than what has been operationally demonstrated to date.

In our view, the market is already pricing in successful external Foundry scaling, sustained margin normalisation, and durable hyperscale CPU competitiveness before sufficient evidence has emerged across customer commitments, profitability, and free cash flow generation. Operational recovery alone is not the same as proven economic recovery.

As such, we issue a SELL call for Intel, with a target price of USD 65 (vs today’s price at USD 113), implying a -42% downside potential.

Table 5: Intel’s valuation

	FY25a	FY26e	FY27e	FY28e
EPS	0.36	0.78	1.55	2.17
y/y		116.9%	97.1%	40.5%
PE	312.4	144.0	73.1	52.0
Current Price	113
Fair PE (x)				30
Upside/downside potential				-42%
Target Price				65

Source: Bloomberg Finance L.P., iFAST Compilation, Data as of 7 May 2026.