Intel’s Q3 Catalyst: Doubling Down on Dominance (Earnings Preview)

Published: October 18, 2025 | America/Detroit

Intel heads into its Q3 2025 print with more eyes on it than at any point in years. The setup is a paradox: a stock that ripped higher in 2025, controversial analyst calls warning it’s run “too far, too fast,” a freshly inked strategic partnership with Nvidia that rewrites industry alliances, and a PC replacement cycle that refuses to die quietly. Add a still-loss-making foundry business that management promises to discipline, and you’ve got one of the most consequential semiconductor earnings events of the season. Barron's+1

Below, I’ll break down what matters most for the quarter—what Intel must show, what the Street actually expects, how to frame the Nvidia partnership, and the bull/bear scenarios across Client Computing (PC), Data Center & AI (DCAI), and Intel Foundry. I’ll end with a scorecard you can keep beside you during the call.

The near-term scoreboard: what the Street expects

When: After the close Thursday, October 23, 2025. Yahoo Finance+1

Consensus snapshot (as of this weekend):

Revenue: Roughly $12.6B–$13.6B guided by Intel last quarter; current outside estimates cluster around ~$13.1B. FT Markets+1
EPS: Ranges from slightly negative to roughly breakeven on a non-GAAP basis (some sources cite < $0.01 positive; others see ~–$0.12 GAAP). The spread reflects uncertainty around one-offs and depreciation linked to the foundry build-out. TipRanks+1

Why the dispersion? Q2 results revealed solid top-line but pressured profitability, plus explicit Q3 revenue guidance and “~$0.00 non-GAAP EPS” commentary—leaving room for model variance on gross margin and opex cadence. Intel Corporation+1

The strategic backdrop in one page

PC cycle tailwinds are real. Within the last two weeks, both IDC and Omdia reported Q3 PC shipment growth as the Windows 10 end-of-support deadline forces refreshes, particularly in desktops and commercial notebooks. That’s a natural tailwind for Intel’s Client Computing revenue mix into the print. IDC+1
The Nvidia shock. In mid-September, Nvidia disclosed a $5B investment in Intel and a plan to co-develop multiple generations of custom data-center and PC products, including x86 SoCs integrating NVIDIA RTX GPU chiplets and custom x86 CPUs for Nvidia’s AI platforms. This isn’t a press-release footnote—it recasts Intel’s relevance in both AI infrastructure and the next wave of “AI PCs.” Ars Technica+3NVIDIA Newsroom+3Reuters+3
Foundry discipline, finally. Intel’s Q2 materials and post-call commentary emphasized tighter spending control and no more “blank checks” for Intel Foundry while the company works to land anchor customers and improve unit economics. Investors will be laser-focused on loss trajectory, external wafer starts, and any fresh customer disclosures. Intel Corporation+1
AI hardware roadmap: Gaudi → Crescent Island. Gaudi 3 expanded availability this year (Dell AI Factory; IBM Cloud regions), but Intel is repositioning to emphasize AI inference efficiency and a yearly cadence for data-center AI silicon, including a new GPU code-named “Crescent Island” set to sample next year for customers (H2’26 timing). Don’t expect Q3 revenue fireworks here—do expect roadmap color that supports 2026–2027 TAM capture. Newsroom+2HPCwire+2
Sentiment check: After a big YTD rally, BofA downgraded Intel to Underperform this week, arguing valuation got ahead of fundamentals and questioning AI positioning and foundry sustainability. This frames the “prove-it” bar for Thursday. Barron's+1

Segment deep dive: what to watch and why it matters

1) Client Computing (PC)

What’s priced in: A modest sequential lift tied to the commercial refresh and desktop strength as Windows 10 sunsets, with upside risk if OEM sell-in held up better than feared. Both IDC and Omdia called out Q3 shipment expansion—especially desktop’s double-digit growth—suggesting channel health. IDC+1

Key KPIs to listen for:

Mix & ASPs: Are AI PC attach rates lifting blended ASPs?
Commercial vs. Consumer: Is enterprise lagging or catching up?
Inventory posture: Distributor and OEM channel inventories exiting Q3.
NPU and local inference: Concrete attach on next-gen mobile/desktop parts and app ecosystem traction.

Why it matters: Client Computing remains Intel’s largest revenue engine and cash seedcorn for the turnaround. A firmer PC floor makes opex discipline and foundry investments survivable.

Upside swing factor: If management quantifies AI PC momentum—either via premium tier mix or early enterprise pilots—Street models will start nudging FY26 gross margin upward. (External work suggests AI PC units can step-change in 2026; investors want Intel-specific attach data now.) IO Fund

2) Data Center & AI (DCAI)

What’s priced in: Mid-single digit Y/Y growth against a muddied competitive backdrop. Market share erosion versus AMD Genoa/Bergamo remains a sticking point, and the AI accelerator narrative has favored Nvidia. Intel’s counter is twofold: Sierra Forest/Granite Rapids progress for CPU leadership on density and performance, and Gaudi 3 availability targeted largely at inference-heavy, cost-sensitive deployments. Add “Crescent Island” as the longer-dated proof of consistency. ServeTheHome+1

What to listen for:

Sierra Forest & Granite Rapids adoption milestones with cloud and enterprise.
Gaudi 3: order book, win-rates, and total solutions price/perf stories against Nvidia and AMD.
Crescent Island cadence and software stack plans—annual releases are a credibility plank. Reuters
Nvidia partnership traction: any early engagements on custom x86 for DGX-class systems or Intel x86 RTX SoCs for PCs. NVIDIA Newsroom

Why it matters: The market will forgive a smaller slice of AI training if Intel can own large swaths of inference (both in the data center and on the PC) at favorable gross margins. A credible story here reframes DCAI from “share donor” to “profit engine.”

3) Intel Foundry (IFS)

Baseline reality: IFS remains loss-making, with Q2 showing billions in operating losses despite revenue recognition from internal and external wafers. Management telegraphed tighter capital discipline into 2026 and needs to prove slope of improvement—especially if the U.S. government’s strategic stake and incentives have already been digested by the market. More Than Moore+1

What to watch:

External wafer wins: Named customers, tape-outs, and 18A/14A pipeline.
Capex/opex glidepath: Concrete numbers on 2025–2026 capex and opex reductions.
Utilization and internal load: How much is the Intel Products side filling the fabs at improving contribution margins?
Process readiness: 18A in production for Clearwater Forest, momentum into 14A. (Any customer add would be a headline.) ServeTheHome

Why it matters: If IFS can narrow losses quarter-on-quarter while maintaining process cadence, the market will begin to value it as an option with real upside rather than a perpetual drag.

The Nvidia partnership: how to underwrite it into models

In September, Nvidia disclosed a $5B investment in Intel at $23.28/share, and both companies said they’ll co-develop multiple generations of custom data-center and PC products. For PCs, Intel will deliver x86 SoCs with integrated NVIDIA RTX GPU chiplets; for data centers, custom x86 CPUs that fit directly into Nvidia’s AI platforms and interconnect fabric (NVLink). This is not a one-off procurement MOU; it sets an architecture lane where Intel’s CPU design IP and manufacturing know-how marry Nvidia’s system-level dominance. Tom's Hardware+3NVIDIA Newsroom+3Reuters+3

How to think about it:

TAM unlock, not cannibalization. Nvidia gains CPU optionality within its systems; Intel gains distribution and relevance where spend is exploding.
PC flywheel: If x86 RTX SoCs hit windows in late 2026/2027, that’s an attach-rate story with platform premiums—good for Intel’s blended client margins. NVIDIA Newsroom
Foundry pull-through: Even absent explicit foundry outsourcing from Nvidia, tighter co-development increases the odds Intel fabs some portion of partner-adjacent silicon longer-term.

What to listen for on Thursday: Any milestone dates, design center collaborations, or software enablement targets. Even modest detail moves the multiple.

AI accelerators: setting the right bar

It’s tempting to demand Gaudi 3 chart-toppers versus H200/GB200. That’s the wrong bar for this quarter. Intel telegraphed an inference-centric position and a new annual cadence culminating in Crescent Island (customer sampling H2’26). The right lens is cost-per-token and throughput per dollar in real-world enterprise inference, plus total solution cost with networking and storage. Look for named wins, reference architectures, and availability via large channels (Dell, IBM Cloud) that can scale deployments without bespoke integration. Reuters+2Newsroom+2

One caution: Reports last year indicated Gaudi 3 shipment targets for 2025 were trimmed—so don’t expect management to trumpet unit numbers. Instead, expect case studies and software ecosystem updates (open frameworks, container recipes, observability). Nasdaq

The PC cycle: more than a one-quarter sugar high

Both IDC and Omdia point to Q3 growth with a tilt toward desktops and commercial, driven by the ticking Windows 10 clock. The world doesn’t re-image overnight, which implies multi-quarter refresh momentum into 2026. That gives Intel oxygen to compress foundry losses while funding DCAI bets. The margin swing factor is AI PC premium mix—NPUs, on-device inference, and RTX-integrated SoCs later on can all raise ASPs. IDC+1

The bear case (what could disappoint Thursday)

Gross margin squish. Higher depreciation and under-absorption from foundry, plus mix (client over data center) could pin margins closer to the low 30s than bulls want.
IFS opacity. If Intel soft-pedals external wins and ducks precise loss containment steps, the market will see “more of the same.”
DCAI growth stalls. Any sign cloud CPU ramps are sliding right—or that Gaudi 3 traction is limited—will reignite “Intel vs. AMD/Nvidia” share-loss narratives.
Guide only meets the midpoint. After the rally, merely in-line outlooks may not be enough, especially with a prominent downgrade fresh in memory. Barron's

The bull case (how Intel can change the conversation)

Client over-delivers with tight channels. Clean inventory, premium mix, and firm enterprise demand would justify multiple support. IDC+1
DCAI specifics. If management gives hard milestones for Sierra Forest/Granite Rapids adoption and Crescent Island’s software stack and partner plan, it resets 2026 expectations. Reuters
Nvidia partnership proof points. Even a small, tangible milestone (joint dev kits, internal test platforms, or software enablement timelines) would be viewed as high-quality forward indicators. NVIDIA Newsroom
IFS loss trajectory bends. Any quantification like “sequential operating loss improvement” and capex discipline for 2026 backs the turnaround narrative. Investopedia

What I’ll have on my notepad during the call

Revenue & EPS vs. the range (do they beat the $12.6–$13.6B guardrails and land near ~$13.1B?) FT Markets+1
Gross margin guide (ex-one-timers; path back to mid-30s?)
Client Computing: channel inventory; AI PC attach; enterprise momentum. Omdia
DCAI: CPU adoption milestones; Gaudi 3 wins; Crescent Island cadence and annual release commitment. Reuters
IFS: sequential loss change; named external wafer customers; 18A/14A timing breadcrumbs; capex cadence. More Than Moore
Nvidia partnership: any concrete 2026 checkpoints on x86 RTX SoCs or custom x86 for Nvidia platforms. NVIDIA Newsroom

Valuation & sentiment: walking the tightrope

BofA’s Underperform call crystallized a broader worry: the stock has run ahead of fundamentals while profitability remains fragile. At the same time, the Nvidia investment and co-development plan isn’t a garden-variety partnership—it’s a structural tie-in that could pull Intel into the fastest-growing spend buckets in tech. The market will toggle between these poles until Intel prints clean execution for a few consecutive quarters. Thursday is the chance to take the first step. Barron's+1

“Doubling down on dominance” — what would that actually look like?

Dominance in PCs doesn’t mean 1999 all over again; it means Intel shaping the AI PC premium with attach, ASPs, and platform features that bend mix toward higher margins.
Dominance in data center doesn’t require winning every piece of AI training; it means owning enough of CPU density (Sierra/Granite) and inference economics (Gaudi 3 now, Crescent Island later) to be a must-buy in AI stacks. ServeTheHome+1
Dominance in manufacturing isn’t overnight profit; it’s relentless loss compression plus external proof points that validate 18A/14A as viable alternatives—setting the table for future foundry margins.
Dominance in ecosystem is where the Nvidia alliance matters most: software, interconnects, and platform-level integrations that ensure Intel silicon sits at the heart of AI systems and next-gen PCs. NVIDIA Newsroom

Risk checklist heading into the print

Macro/FX: Europe and China PC softness can dent otherwise healthy refresh cycles.
Competitive cadence: AMD’s server roadmap remains aggressive; Nvidia’s platform velocity keeps raising the bar for accelerators.
Execution risk in IFS: Yield, cycle times, and customer onboarding remain heavy-lift.
Accounting & one-offs: Impairments, accelerated depreciation, and reorg costs can skew GAAP vs. non-GAAP optics—watch the reconciliation table closely. (Q2 had plenty of noise.) AlphaSense

A scenario framework for Thursday

Base case (most likely):

Revenue in the upper half of the guide; EPS near breakeven non-GAAP.
Client prints clean (helped by desktop), DCAI steady, IFS still in the red but with better color on discipline.
Stock reaction: muted to slightly positive unless the guide for Q4 edges above consensus.

Upside case:

Client beats with premium mix, DCAI adds tangible Sierra/Granite customer milestones, and IFS quantifies a narrower sequential loss with a 2026 capex path.
Management gives concrete Nvidia partnership milestones for 2026.
Stock reaction: positive follow-through; multiple support resets higher. NVIDIA Newsroom

Downside case:

Gross margin underwhelms on mix and foundry burden; DCAI lacks new wins; IFS disclosure is thin.
Guide sits at the midpoint with cautious language.
Stock reaction: pullback, validating recent Underperform arguments. Barron's

Five questions to ask on the call

What percentage of Q3 client shipments qualified as “AI PCs,” and how did that mix affect ASPs? (Correlate to IDC/Omdia shipment growth.) IDC+1
On Sierra Forest and Granite Rapids, which top-tier cloud or enterprise customers are live in production—and what workloads?
For Gaudi 3 and Crescent Island, what are the performance-per-dollar or token-economics benchmarks you’re using against H200/GB200—especially for enterprise inference? Reuters
What is the sequential change in IFS operating loss and utilization, and what’s your 2026 capex guardrail? Investopedia
Can you provide a 2026 checkpoint for the Nvidia partnership (dev kits, silicon tape-outs, or software milestones) on both the data-center CPU side and the x86 RTX PC SoC? NVIDIA Newsroom

Bottom line

Intel doesn’t need to “win AI” on Thursday. It needs to prove execution on the pieces it can control right now—PC mix and margin discipline, DCAI CPU momentum, a credible inference angle for accelerators, and a clear, measurable plan to bend the IFS loss curve. If it does that and sprinkles in specific milestones on the Nvidia alliance, the market will grant the company the benefit of the doubt for the next leg of the turnaround.

If it doesn’t—if the call leans on high-level aspirations without fresh detail—expect the valuation debate to swing back toward caution. After a powerful year-to-date move, the bar is higher than a simple in-line print. Barron's

Fast reference: facts and sources to keep handy

Earnings date: Thu Oct 23, 2025 (after close). Yahoo Finance+1
Q3 guidance from Q2 release: $12.6B–$13.6B revenue; ~$0.00 non-GAAP EPS. FT Markets
PC shipment momentum in Q3: Growth confirmed by IDC and Omdia (desktop especially strong). IDC+1
Nvidia x Intel: $5B stake; co-develop custom data-center CPUs and x86 RTX SoCs for PCs. NVIDIA Newsroom+2Reuters+2
AI roadmap: Gaudi 3 availability (Dell AI Factory; IBM Cloud); Crescent Island sampling H2’26; emphasis on inference and annual cadence. Newsroom+2HPCwire+2
IFS reality check: Large operating losses in Q2; management signaling spend discipline into 2026. More Than Moore+1
Street sentiment split: BofA Underperform heading into earnings on valuation/strategy skepticism. Barron's

Your quick-hit scorecard for the call

Headline: Rev ≥ $13.1B? Non-GAAP EPS at or above breakeven? TipRanks
Gross margin trajectory improving?
Client Computing: AI PC mix & enterprise strength, clean channels. Omdia
DCAI: Named CPU wins; Gaudi 3 case studies; Crescent Island cadence. Reuters
IFS: Sequential loss improvement + concrete capex guardrails. Investopedia
Nvidia partnership: 2026 milestones—dates, deliverables, dev kits. NVIDIA Newsroom

If Intel checks three or more of those boxes with specifics, it won’t just “meet expectations.” It will re-anchor the narrative—from “hope and hype” to “measurable progress.” That’s the catalyst long-term holders want from Q3.