Fifth Straight Beat, First Real Guide Miss: The 17% Drop Creates the Entry Point the Stock Denied Us in Q3

Results vs. Consensus

Full-Year 2025

• Revenue: $52.9B (approximately in line with post-Q3 expectations)
• OpEx reduction: -15% vs. FY2024 — met the $17B target Tan set on the Q1 call
• Custom ASIC: +50% revenue growth YoY — a new disclosed metric
• Beat cadence: Five consecutive quarters of exceeding guidance on revenue, margin, and EPS

Quality of Beat/Miss

• Revenue: The +2.2% beat is the smallest beat since Q1, and is tighter because demand exceeded Intel's ability to supply — a different phenomenon from pull-forward or seasonality. Underlying demand was strong enough that supply-constrained the print.
• Margins: 37.9% non-GAAP GM is ~200 bps below Q3's outsized 40% print but still above guidance. Mix tilt toward server on the wafer allocation explains the sequential step-down.
• EPS: $0.15 nearly doubled consensus. Operational quality, not tax or below-the-line help. This is the cleanest high-quality EPS beat Intel has delivered in the Tan era.
• FCF: Q4 directionally positive on cash from operations; full-year adjusted FCF still negative but materially narrower than entering 2025. Zinsner's commitment to positive adjusted FCF in FY2026 is now the key external target.

Segment Performance

Client Computing — Strong Demand Meets Wafer Allocation Reality

CCG at $8.2B down 6.6% YoY looks weak on the comp, but the reality is more nuanced: PC demand in Q4 was robust (client AI PC units +16% sequentially), but Intel made a deliberate Q3 decision to shift wafer allocation toward server (where unit ASPs are higher). CCG was the explicit loser in that trade. Panther Lake's launch with three SKUs is the positive offset.

"Our client franchise showed the strongest unit momentum we've seen in years, but we made the deliberate call to allocate limited wafer capacity to server in Q3 — a decision that shows up in Q4 client revenue and will show up in Q1 as well." — David Zinsner, CFO

Assessment: This is not a franchise-weakness story — it's a capacity-allocation story. The right wafer allocation was chosen (higher-ASP server over lower-ASP client), and the supply recovers from Q2 onward. The Q1 guide pain is the cost of that Q3 decision. Our read: positive strategic trade-off, short-term optics pain.

Data Center and AI — Custom ASIC Is the New Growth Engine

DCAI at $4.7B up 8.9% YoY is the segment with the cleanest acceleration in the quarter. The disclosed standout: custom ASIC revenue grew 26% sequentially to exceed a $1B annualized run-rate, with full-year growth above 50%. Management framed this as the base for a $100B TAM opportunity.

"Our custom ASIC business exited the year at more than $1 billion annualized and grew 50% in 2025. We see this as the first meaningful chapter of a much larger opportunity — the AI silicon market is heterogeneous, and custom is where the next decade of volume will be." — Lip-Bu Tan, CEO

Assessment: This is the disclosure that matters most for the long-term thesis. Custom ASIC is the category where Intel can plausibly compete: it doesn't require discrete accelerator architectural advantages over Nvidia; it requires manufacturing, packaging, and co-design capability — which Intel has. If the $1B run-rate doubles by end of 2026, DCAI becomes a sustainable double-digit grower.

Intel Foundry — 18A in Production, 14A Still the Binary

IFS at $4.5B up 3.8% YoY is the quietest segment of the print, but the operational milestones were significant:

• 18A HVM achieved October 2025 — the first Intel leading-edge node at high-volume in three years
• Yields improving 7–8% monthly — a textbook ramp curve
• EUV wafer mix 10% of 2025 production (up from <1% in 2023)
• Panther Lake first external-process product shipping at volume

Assessment: Every operational 18A milestone that Intel has committed to over the past 18 months has been hit. Panther Lake launching with three SKUs vs. one committed is genuine positive surprise — Tan's team over-delivered on the most visible execution test of the turnaround. The 14A question remains open; see Key Topics.

Guidance & Outlook

The Q1 guide is the first meaningful miss of the Tan era, and the market reaction (-17%) reflects a stock that had been pricing continued beat-and-raise cadence. The mechanical drivers are disclosed:

• Buffer inventory depleted by strong H2 2025 demand
• Q3 2025 wafer mix decision to prioritize server reduced client supply
• Q1 is the trough; sequential improvement begins Q2

FY2026 Framing (Reiterated):

1. Exit 2026 at ~40% non-GAAP gross margin
2. Positive adjusted FCF for the full year (first since 2022)
3. Continued revenue growth beyond Q1 trough

Implied ramp: If Q1 is $12.2B and FY2026 ends at a ~$14.5B run-rate supporting ~40% GM, quarterly progression is roughly Q1 $12.2B / Q2 $13.0B / Q3 $13.8B / Q4 $14.5B — implying FY2026 revenue of roughly $53.5B. That's flat-to-modest growth, but with materially better margin profile exiting the year.

Guidance style: The Q1 guide looks conservative in the context of the supply trough framing. The 40% GM exit target gives Tan room to beat through the year as yields improve. We read this as reset-and-reload setup rather than a pattern break.

Key Topics & Management Commentary

Overall Management Tone: Steady, almost deliberately unflashy. Tan and Zinsner spent the call talking execution mechanics (wafer allocation, yield curve, supply ramp) rather than strategic positioning. This is the tone of a team that has shifted from "prove the transformation" to "execute the playbook." The contrast with Q1's "there are no quick fixes" and Q2's "no more blank checks" is striking — the posture is now that of an operator, not a restructurer.

The Q1 2026 Supply-Constrained Trough

Zinsner laid out the mechanics in detail:

• H2 2025 demand exceeded plan; buffer inventory drew down
• Q3 2025 wafer allocation decision: prioritize server (higher ASP) over client
• Q1 2026: combined effect of buffer depletion + allocation lag shows up as a revenue trough
• Quantified as "a few hundred million dollars" of revenue deferred, not lost
• Sequential improvement from Q2 as 18A capacity ramps and inventory rebuilds

"Q1 2026 will be our most constrained quarter. The combination of strong demand and our decision to shift wafer mix toward servers in Q3 depleted our buffer inventory. We expect sequential improvement beginning in Q2." — David Zinsner, CFO

Assessment: The supply constraint story is operationally credible and strategically favorable — Intel is capacity-constrained on the products it wants to sell. The 17% stock drop reflects the optics (first miss, weaker margins, lower EPS) but not the substance (demand-limited, not demand-weak). This is the kind of miss that historically reverses within 2–3 quarters when the supply rebuilds.

18A HVM and Panther Lake Over-Delivery

Intel 18A entered high-volume manufacturing in October 2025. Yields improving 7–8% monthly — textbook ramp. Panther Lake launched with three SKUs vs. the one-SKU commitment. Fab 52 Arizona fully operational. EUV wafer mix at 10% of production.

"We exceeded our commitment by delivering three SKUs of Core Ultra Series 3 by year-end 2025, built on Intel 18A. This is the strongest manufacturing milestone in Intel's transformation." — Lip-Bu Tan, CEO

Assessment: This is the operational achievement that most validates the entire Tan thesis. For 18 months the market priced Intel for a likely node slip (following the pattern of TSMC and Samsung this cycle); Intel did not slip, and over-delivered on the most visible external commitment. Combined with 7–8% monthly yield improvement — an exceptional ramp pace — the manufacturing credibility rebuild is now genuine, not hypothetical.

Custom ASIC — The New Growth Leg Emerges

Custom ASIC revenue grew +26% sequentially in Q4, +50% for the full year, and exited 2025 at a $1B+ annualized run-rate. Tan framed this as the base for a $100B TAM opportunity.

Assessment: This is the most strategically important new disclosure of the quarter. Custom ASIC is the category where:

1. Intel has natural right-to-win (manufacturing, packaging, x86 integration, Altera IP legacy)
2. The market structurally favors hyperscaler-specific silicon over general-purpose GPUs
3. Nvidia partnership positioning is directly applicable (custom Xeons for AI platforms are conceptually custom ASIC)
4. Margins are typically healthier than commodity server CPUs

If the $1B run-rate grows to $2–3B by end of 2026 and $5B+ by 2028, this single product line contributes 100–200 bps of annual revenue growth plus margin support. The market has not yet modeled this.

14A — Same Framework, 2H 2026 Decision Window Approaching

No change in the framework established in Q2 and reaffirmed in Q3. Key updates:

• Two prospective external customers currently evaluating test chips
• Firm capacity commitments expected H2 2026 / early 2027
• 14A risk production: late 2027
• 14A volume production: 2028
• Capex paced to firm commitments — no speculative build

Tan's phrase — "going big time into 14A" — signals the commitment if customers commit, but the "if" is still the binary.

Assessment: The 14A decision window is now 6–12 months out. That's close enough to be a catalyst-rich period but far enough that pricing in a specific outcome would be premature. Our probability estimate for landing at least one anchor customer is 60–70% — up modestly from the 55–65% we assigned after Q3 — reflecting Panther Lake's execution confirming Intel's 18A capability and strengthening the 14A pitch to prospective customers.

AI Strategy — ASIC-Centric, Gaudi Winding Down

The quarter confirmed what has been implicit for three calls: Intel's AI accelerator strategy is custom-silicon centric rather than discrete-accelerator centric. Gaudi 3 was the last discrete accelerator; no Gaudi 4 roadmap disclosed. Custom ASIC and Nvidia-partnership Xeons are the forward strategy.

Assessment: This is honest. Intel is not going to out-Nvidia Nvidia in training. But custom silicon for hyperscalers, combined with the x86 Xeon roadmap and the Nvidia-partnered NVLink Xeons, is a genuinely differentiated position. The AI strategy has crystallized into something coherent for the first time in three years.

Analyst Q&A Highlights

Q1 Supply Constraint Magnitude

• Vivek Arya, Bank of America: Asked for quantification of the supply constraint in dollar terms. Zinsner: "A few hundred million dollars of revenue deferred from Q1 into later quarters."
  Assessment: "Deferred, not lost" is the key framing. If accurate, Q2–Q4 2026 each get a ~$100M boost from the Q1 constraint unwind — supportive of the sequential-ramp narrative.

2026 Full-Year Revenue Trajectory

• Stacey Rasgon, Bernstein: Pushed for formal FY2026 revenue guidance. Zinsner declined, instead directing analysts to the sequential improvement pattern and the 40% GM exit target.
  Assessment: Refusing to provide a full-year number is consistent with Tan's quarterly-only disclosure discipline. Implied FY2026 revenue: ~$53.5B based on reconstructed quarterly ramp.

Custom ASIC Pipeline

• Joseph Moore, Morgan Stanley: Asked about the customer concentration and pipeline depth of the $1B+ custom ASIC business. Tan: declined to disclose customer names but confirmed "multiple top hyperscalers" and described the pipeline as "early but growing faster than we modeled."
  Assessment: The $100B TAM framing is aspirational; the execution marker is whether the pipeline converts to disclosed wins over the next 2–4 quarters. This is a new thing to watch.

14A Customer Commitment Probability

• Timothy Arcuri, UBS: Pressed for a probability weighting on 14A landing an anchor customer. Tan: "We have active engagements. I'm not going to handicap probabilities publicly."
  Assessment: Appropriate refusal. The substance — that Tan is describing engagements as "active" rather than exploratory — is incrementally positive vs. Q2 and Q3 language.

2026 Adjusted FCF Bridge

• Ross Seymore, Deutsche Bank: Asked for the bridge from 2025's negative FCF to 2026's positive adjusted FCF target. Zinsner: operating cash flow improvement from margin expansion, capex flat-to-down, working capital neutral.
  Assessment: Credible if 18A yields continue ramping and Q1 supply constraints resolve on schedule. Is a $2–5B swing in FCF, which if delivered materially improves the investability of the stock.

What They're NOT Saying

1. Formal FY2026 revenue guide: Declined to provide. Implied trajectory ~$53–55B but left to analyst reconstruction from Q1 plus sequential-improvement language. This preserves beat-and-raise optionality but reduces transparency.
2. Custom ASIC customer names: $1B+ run-rate disclosed but customer-level visibility not provided. Industry expectation: AWS, Microsoft, Meta, Google are candidates. Confirmation of specific wins would be materially positive.
3. 14A anchor customer identity or probability: Still two prospective customers unnamed. Decision window 6–12 months. The absence of forward signaling is consistent with negotiation discipline but keeps the binary unresolved.
4. Q2 2026 supply recovery quantification: "Sequential improvement" is the language; no specific dollar magnitude given. Street to model $800M–$1B of sequential uplift.
5. Foundry breakeven year: Still no specific date. Zinsner's Q3 "as we exit 2026" language not revisited on this call. IFS breakeven likely 2027.
6. Gaudi product roadmap: Effectively retired without formal announcement. Gaudi 3 is the last discrete accelerator; no successor disclosed. This is a quiet strategic exit.
7. Specific 2026 capex number: Directional language ("flat-to-down") but no dollar target. Given the capital raised in 2025 this could actually tick higher than Q1-era expectations if 14A customers commit — that would be a positive signal.

Market Reaction

• Pre-earnings close (Jan 22): Stock trading in the mid-$40s zone — up ~84% for calendar 2025 entering the print.
• After-hours move (Jan 22): Initial mixed reaction; sold off as Q1 guide digested.
• Next-day (Jan 23): Shares fell ~17% in regular trading — Intel's worst single day since August 2024 (the post-Q2 2024 Gelsinger-era collapse).
• Volume: Elevated on the sell-off.
• Analyst reactions (within 48 hours):
  • Multiple price target cuts — targets migrating to $38–45 range from $45+.
  • Ratings largely unchanged; most desks maintained their post-Q3 bullish positioning.
  • Some houses moved from Buy to Hold; few outright downgrades.
  • Narrative reset: "turnaround is real but lumpier than modeled."

The 17% drop is sharp but paradoxical against what Q4 actually delivered — beat on revenue, margin, EPS; 18A in HVM; Panther Lake over-delivered; custom ASIC scaled. The reaction is a valuation correction plus a first-miss-after-four-beats pattern-break, not a thesis break. If anything, the post-print level re-creates the entry point the stock denied in Q3 — at roughly 30x normalized 2026 EPS, Intel is back in the range where the Outperform call doesn't require the stock to do anything heroic.

Street Perspective

Debate: Is the Q1 Miss a Pattern Break or a Transitory Wafer-Allocation Decision?

Bull view: The supply constraint is demand-led, not demand-weak. Intel is short on server wafers because hyperscalers are buying more than Intel can make. That's exactly the position semi investors want to see. The Q1 trough resolves by Q2 as 18A scales, and the cadence resumes.

Bear view: Five beats followed by a miss is a pattern. The "supply constraint" explanation is too convenient; maybe CCG demand is actually weaker than management is admitting, and the server prioritization is cover. Intel has historically disappointed after stretches of good execution — don't fight history.

Our take: The bulls have the stronger argument here. CCG unit momentum (+16% sequential on client AI PC) and DCAI +8.9% YoY are not consistent with a demand weakening story. The supply constraint is credible, and the Q1 trough framing sets up beat-and-raise reload through 2026.

Debate: Does the Custom ASIC Growth Justify a Higher Multiple?

Bull view: $1B+ annualized, +50% YoY, into a $100B TAM. If custom ASIC grows to $3–5B by 2028, DCAI becomes a 10%+ grower with better margins than traditional Xeon. This is a multi-year multiple story the Street hasn't priced.

Bear view: Custom ASIC is a crowded and margin-compressed business. TSMC, Broadcom, and Marvell all have stronger relationships with hyperscalers. Intel's $1B is impressive vs. prior years but small vs. Broadcom's $25B+ in custom silicon. Growing from a small base doesn't imply winning the category.

Our take: The bull case is more plausible than the bear, but not yet proven. Our working assumption is that custom ASIC reaches $2.5–3B by end of 2026 — enough to contribute meaningfully to DCAI growth and margin — but we would not yet underwrite a full $100B-TAM-share case.

Debate: Is the Stock Buyable at Post-Drop Levels?

Bull view: The -17% reaction is an opportunity. Fundamentals are intact, 18A in HVM is structurally de-risked, custom ASIC is accelerating, the balance sheet is fixed, and the next binary (14A customers) is still 6–12 months out — which is enough time for the Q1 trough to resolve and the beat-and-raise cadence to reestablish.

Bear view: The stock is still up ~50% YTD 2025 even after the drop. The 14A binary is real, and if the 2H 2026 decision window produces no customer, the leading-edge foundry roadmap is in serious question. Paying 30x 2026 EPS for a stock with that much forward uncertainty is not obviously cheap.

Our take: The bulls have the better 6–12 month case; the bears have the better argument for position sizing. We stay Outperform but size positions accounting for the 14A binary — this is not a conviction-max stock.

Model Update Needed

Valuation impact: On reset FY2026 non-GAAP EPS of $0.85–1.05 and a 30–35x multiple (fair for the stage of the turnaround), fair value band tightens to $28–36. At the post-print level, Intel sits near the midpoint of that range — similar setup to where the stock traded at $34–36 entering Q3 earnings, but with materially more proven execution and de-risked balance sheet behind it. That's why we don't downgrade: the fundamental case is stronger today than it was pre-Q3 despite the lower EPS numbers, and the price gives the thesis room.

Thesis Scorecard Post-Earnings

Overall: Thesis strengthened on execution, unchanged on binary 14A risk, marginally cleaner on valuation. Net: favorable.

Action: Maintain Outperform. Next upgrade trigger (conviction sizing): 14A anchor customer announcement, OR Q2 2026 print confirming the supply-constraint trough resolves as guided. Downgrade triggers: 14A customer decision turns negative in H2 2026, OR Q2/Q3 2026 supply recovery misses the guided sequential ramp.