Intel's Q1 2026 will be summarized everywhere as a clean beat: revenue, gross margin, and EPS all above the high end of guidance, the sixth consecutive quarter of exceeding expectations, and a sequential narrative that reads like the company has finally turned. The numbers are not in dispute. What is worth examining is what those numbers actually describe — and what management put on the table in framing them.
This note walks through the ASP-versus-volume mechanics that produced the beat, the Q2 input-cost guide that bookends the celebration, three claims from the call that deserve to be pressure-tested against the rest of the market, and a set of items in the 10-Q that didn't make most of the day-of coverage.
The headline results, briefly
Q1 2026 revenue of $13.6B (+7% YoY) came in $1.4B above the midpoint of the prior guide. Non-GAAP gross margin was 41.0% — roughly 650bps above the company's own guide — and non-GAAP EPS was $0.29 versus a guide of break-even, with about 6 cents of that beat attributed to a one-time gain in interest and other.
GAAP results were a different picture: a $3,728M net loss attributable to Intel, or $(0.73) per diluted share, driven almost entirely by $4,070M of restructuring and other charges that include the $3.9B Mobileye goodwill impairment (more on this below).
Segment revenue:
| Segment | Q1 2026 | YoY | Operating income |
|---|---|---|---|
| Client Computing Group (CCG) | $7.7B | +1% | $2.5B (33% margin) |
| Data Center and AI (DCAI) | $5.1B | +22% | $1.5B (31% margin) |
| Intel Foundry | $5.4B | +16% | $(2.4)B |
| All Other (incl. Mobileye) | $0.6B | -33% | $0.1B |
The Q2 2026 guide is $13.8–14.8B in revenue, 39.0% non-GAAP gross margin (down from 41.0%), and $0.20 non-GAAP EPS.
The ASP story is the real story
The most informative passage in the entire reporting package is two paragraphs deep in the 10-Q's MD&A. From the source, with our emphasis:
In Client Computing, "client revenue (collectively notebook and desktop) was $6.6 billion in Q1 2026, up $54 million from Q1 2025, primarily driven by a 16% increase in ASPs. The increase in ASPs was driven by a higher mix of premium products sold in Q1 2026 and demand-based pricing actions in part to offset higher input costs. This increase in client revenue was partially offset by a 13% decrease in volume compared to Q1 2025. Market demand exceeded our available product supply..."
In Data Center and AI, "DCAI revenue increased $926 million from Q1 2025, primarily driven by $696 million of higher server revenue due to a 27% increase in server ASPs... This increase in server revenue was partially offset by a 5% decrease in volume compared to Q1 2025."
Read those carefully and the picture changes. Intel did not sell more silicon in Q1 2026 than it did in Q1 2025. It sold less. It made more revenue per unit because it shipped a richer mix of premium parts and because, by its own admission, it raised prices to offset rising input costs. The supply-constrained framing — "market demand exceeded our available product supply" — is genuine, but it is also the precondition that allowed pricing actions to stick.
ASP did the heavy lifting in both segments — units shrank, revenue grew.
Year-over-year change in average selling price vs. unit volume, Q1 2026 vs. Q1 2025
Source: Intel 10-Q for the quarter ended March 28, 2026 (MD&A discussion of CCG and DCAI segments).
The Q2 gross margin guide is what tells you to take this seriously as a forward signal. The non-GAAP gross margin guide of 39.0% is a 200bps sequential step-down from 41.0%, which CFO Zinsner attributed in the prepared remarks to a meaningfully larger Intel 18A contribution still early in its ramp, the absence of one-off Q1 inventory benefits, and — the part procurement teams should highlight — "rising input costs, especially in memory" presenting "growing headwinds in the second half." The CFO went further: "constraints and rising prices around key components like memory, wafers, and substrates are driving higher costs that could impact demand for our products at some point in the year."
Translation for buyers: Intel is signaling that the 2H 2026 cost stack for client and server CPUs is structurally higher. Some of that gets passed through. Procurement plans built around 2025 unit pricing should be re-baselined.
The Q2 step-down also needs context. Read against Q1 2025's actual non-GAAP gross margin of 39.2%, the Q2 2026 guide of 39.0% is essentially a return to the prior-year baseline — Q1 2026's 41.0% looks like the outlier, not Q2. The structural read is that the input-cost wave shows up cleanly in Q2 and that the Q1 beat came from a combination of one-time inventory benefits and a temporarily favorable mix.
Q2 guide of 39.0% returns gross margin to the Q1 2025 baseline — Q1 2026 was the outlier.
Non-GAAP gross margin, Q1 2025 actual vs. Q1 2026 actual vs. Q2 2026 guide (midpoint)
Source: Intel Q1 2026 Earnings Release (Apr 23, 2026); Q2 2026 figure is the company's own non-GAAP guide at the midpoint of the $13.8-14.8B revenue range.
Claim 1: The "CPU renaissance"
CEO Lip-Bu Tan spent meaningful airtime on a thesis that "the CPU is reasserting itself as the indispensable foundation of the AI era," that the CPU is "the orchestration layer and critical control plane for the entire AI stack," and that customers are "deploying server CPUs along accelerators in a ratio that is moving back towards the CPU."
The strongest single proof point in the prepared remarks is that Xeon 6 was selected as the host CPU for NVIDIA's DGX Rubin NVL8 systems. Read straight, that win is real. NVL8 is an 8-GPU rack with 2 host Xeon CPUs — a 1:4 CPU-to-GPU ratio. That ratio is twice as accelerator-heavy as NVIDIA's flagship NVL72 architecture (36 Vera CPUs to 72 Rubin GPUs is 1:2, and the previous-generation GB200 NVL72 was 36 Grace to 72 Blackwell — both reflecting accelerator-led economics). The NVL8 win is a meaningful socket and a meaningful margin contribution for Intel; it is not by itself evidence that the broader system architecture is moving back toward the CPU.
NVL8 is twice as accelerator-heavy per host CPU as the flagship NVL72 — the win doesn't validate the broader CPU thesis.
GPU-to-host-CPU ratio across NVIDIA rack architectures (lower = more CPU per GPU)
Source: NVIDIA system architecture disclosures (GTC 2024, GTC 2026). NVL8 ratio derived from the 8-GPU / 2-CPU host configuration cited in Intel's Q1 2026 prepared remarks. Rubin NVL8 uses Intel Xeon 6 hosts.
What would actually validate Tan's claim is a measurable shift in server capex mix — accelerators flat or declining as a share of bill-of-materials while host CPUs grow. That is not what the public data shows today. Hyperscaler 2026 capex is being led by accelerators (a large fraction of the roughly $650B aggregate hyperscaler capex consensus for 2026 sits in AI infrastructure), and AMD's Q4 2025 server CPU revenue share reached 41.3%, a record — meaning the x86 server-CPU pool itself is being split, not concentrated.
Three datapoints to watch over the next 30 days:
- AMD Q1 2026 (~May 5) — server CPU growth, EPYC ASP commentary, and any callout of CPU-share gains in hyperscaler deployments.
- Microsoft, Alphabet, Meta (April 29–30) — comments on the CPU-versus-accelerator split in 2026 capex.
- Amazon (around April 30) — AWS commentary on Graviton, EPYC and Xeon mix.
If those prints show CPUs growing as a share of incremental data center spend, Tan's framing earns weight. If they show accelerators continuing to absorb the marginal dollar, the NVL8 win remains a real win for Intel without confirming the broader thesis. Both can be true.
Claim 2: Foundry external traction
The Intel Foundry segment posted $5.4B of revenue, up 16% YoY. That number is worth looking through. Of the $5.4B, external foundry revenue was $174M in Q1 2026, up from $31M in Q1 2025 — an absolute increase of $143M. Most of that increase reflects the Altera supply relationship that became "external" after Intel deconsolidated Altera following the September 2025 sale of 51%. That is, the larger external number is structurally driven by an accounting boundary moving, not by net-new third-party design wins ramping into volume.
External foundry is 3.2% of segment revenue — and most of the YoY gain is Altera transitioning, not net-new design wins.
Intel Foundry segment revenue split, Q1 2026 vs. Q1 2025 ($M)
Source: Intel 10-Q for the quarter ended March 28, 2026 (Note 14: Operating Segments). External revenue includes third-party foundry and assembly/test, with the Q1 2026 step-up largely attributable to Altera supply post-deconsolidation (Sep 12, 2025). 'Internal' is intersegment revenue from Intel Products and other internal businesses.
This is not to dismiss the foundry business. Tan flagged that Intel 14A maturity, yield, and performance are outpacing Intel 18A at a similar point in time, that PDKs continue to develop with multiple customers actively evaluating, and that "early design commitments emerge beginning in the second half of 2026 and expanding into the first half of 2027." Those are the timestamps to put in your calendar. Until those design commitments are announced — by name, with dollar size and target nodes — the external foundry book remains a story about future ramps, not current revenue.
There is also a risk disclosure in the 10-Q that mainstream coverage is unlikely to highlight. The risk-factor language is explicit: "if we are unable to secure sufficient committed demand for Intel 14A through product design wins with potential significant external customers and our Intel products roadmap, we face the prospect that it will not be economical to develop and manufacture Intel 14A and successor leading-edge nodes on a go-forward basis. In such event, we may pause or discontinue our pursuit of Intel 14A and successor nodes." That sentence has been in the filings for a few quarters now, but it is the conditional that the 2H 2026 / 1H 2027 design-win window is meant to resolve. If the wins don't materialize on that timeline, the 14A discussion changes shape.
For context on how 14A would have to perform to compete, see our TSMC 1Q26 earnings analysis — TSMC just announced three new N3 sites and pulled forward N2 ramps, which is the relevant external benchmark.
Claim 3: Mobileye — the operating beat versus the structural pressure
Mobileye's Q1 2026 was, on its own terms, a strong quarter. Revenue grew 27% YoY to $558M (the company's own report), the full-year guide was raised, and Mobileye announced a $250M buyback. Inside Intel's reporting, the All Other segment posted $628M of revenue and $102M of operating income, with management attributing the sequential strength to Mobileye.
And then Intel impaired $3.9B of Mobileye goodwill.
The 10-Q's language on what drove the impairment is the part to read carefully. The quantitative test was triggered by the "sustained decline in Mobileye's market capitalization since our most recent assessment date in the fourth quarter of 2025, as well as increased uncertainty in the broader macroeconomic and geopolitical environment in which Mobileye operates." The discount rate increase that drove the write-down was attributed to "higher market-based and Mobileye-specific risk premiums arising from changes in global macroeconomic conditions, including heightened geopolitical risks associated with operations in the Middle East, including the current conflict, and increased uncertainty related to Mobileye's evolving competitive landscape." The filing then names the specific concern: "the long-term competitive positioning of certain Mobileye technologies, particularly with regard to increasingly advanced autonomous driving solutions."
That is Intel's auditors and management writing down the long-tail value of Mobileye's competitive position. After the charge, Mobileye reporting unit goodwill stands at $4.3B (down from $8.2B at year-end 2025). The 10-Q also discloses a sensitivity that should give pause: an additional 1% increase in the discount rate would have produced approximately $682M of further impairment in this same period.
Q1 2026 cut goodwill by $3.9B — and the 10-Q says another +1% discount rate would clip another ~$682M.
Mobileye reporting unit goodwill: actual movements and disclosed sensitivity ($B)
Source: Intel 10-Q for the quarter ended March 28, 2026 (Note 10: Goodwill). $682M sensitivity disclosed in the same note for an additional 1% increase in the discount rate.
The "evolving competitive landscape" language is consistent with what is publicly visible across Mobileye's competitive set. NVIDIA Drive Thor is now in production wins across multiple OEMs (Lucid dual-Thor, Mercedes CLA, BYD, Li Auto and others). Qualcomm disclosed an automotive design pipeline of more than $45B at its most recent fiscal-quarter print. Horizon Robotics is reported to hold approximately 45.8% share of the China ADAS market. Tesla taped out its AI5 inference chip in April 2026.
A near-term operating beat at Mobileye is not in tension with a long-term competitive impairment, and that is the read the 10-Q is endorsing. If you cover the AV stack, this is the quarter the auditor took a position.
What's hidden in the 10-Q
Three items that materially affect Intel's risk profile didn't make most of the day-of coverage.
Risks from the conflict with Iran (10-Q risk factor disclosure). On February 28, 2026, the U.S. and Israel initiated coordinated strikes against Iran, with retaliatory action across the broader region. Two items from the disclosure matter for semiconductor procurement: first, "Iranian strikes on two energy fields in Qatar that supply a meaningful percentage of the world supply of helium have resulted in a global shortage of this gas that is essential to the semiconductor manufacturing" process. Helium is consumed across multiple front-end steps; a sustained Qatar disruption is industry-wide, not Intel-specific. Second, "in late March 2026, Iran published a list of U.S. companies with operations in the Middle East whose facilities they indicated they would target in retaliation, with Intel being near the top of that list. A significant portion of our revenues are generated from products on Intel 7 manufactured at our fabrication facility in Israel." Read directly: a meaningful slice of Intel's revenue is currently tied to a single fab inside an active conflict zone.
Fab 34 / Apollo unwind: $14.2B reacquired, partly debt-funded. "In April 2026, we reacquired Apollo's 49% minority ownership interest in Ireland SCIP for aggregate cash consideration of approximately $14.2 billion." CFO Zinsner described the transaction as "highly accretive" and confirmed in prepared remarks that it was funded with approximately $7.7B in cash and $6.5B in new debt. In the call's own words, this allows shareholders "to participate in the full economic benefits from a fab just now hitting its stride." A different framing is also defensible: the original Apollo joint investment was structured to de-risk the Fab 34 build during a period of stretched balance sheet; Intel is now unwinding that hedge while the fab is mid-ramp and adding $6.5B of debt to do it. Both framings can be supported by the same set of facts. Net-new non-controlling interest commentary now points to roughly $250M per quarter for Q2–Q4 2026 and $1.1B in 2027 and 2028 combined, on a GAAP basis.
Mobileye impairment is sensitive to the discount rate. Already noted above, but worth surfacing: a further 1% move in the discount rate triggers approximately $682M of additional impairment. In an environment of rising real rates, this number is unlikely to be the last write-down on this asset.
What to watch in Q2 2026
Three earnings prints in the next 30 days resolve most of the open questions raised this quarter:
- AMD Q1 2026 (~May 5). Server CPU revenue growth and EPYC ASP commentary will tell you whether the +27% server ASP environment is broad-based or Intel-specific. AMD's Q4 2025 server share of 41.3% is the baseline.
- Hyperscalers (Microsoft, Alphabet, Meta, April 29–30; Amazon ~April 30). Capex framing — particularly any commentary on the CPU-versus-accelerator split in 2026 spending — is the most direct test of Tan's "CPU renaissance" framing.
- Marvell Q1 FY27 (late May). Custom ASIC revenue growth and any color on Google MPU / inference TPU programs reads on whether DCAI's "ASIC revenue up more than 30% sequentially and nearly doubling year-over-year" signal holds in the broader merchant ASIC supply chain.
For Intel itself, the items to track over the next two quarters are concrete and named: 14A external design commitments (the 2H 2026 / 1H 2027 window Tan flagged), Intel 18A-P milestones, and the SuperVision Porsche launch from Mobileye in late 2026 against the Drive / Snapdragon Ride / Horizon competitive set.
Live forward calendar: Earnings Tracker.
Methodology and caveats
Every numerical claim in this analysis ties to one of: Intel's Q1 2026 earnings release (Exhibit 99.1, April 23, 2026), the Form 10-Q for the quarter ended March 28, 2026, the prepared CEO/CFO remarks (Lip-Bu Tan and David Zinsner, April 23, 2026), or the Q1 2026 earnings deck. Specific MD&A phrasing on ASP and volume is from the 10-Q's segment discussion. The Mobileye impairment language and the Iran/helium disclosure are from the 10-Q's risk factors and Note 10 (Goodwill). Competitor share figures (AMD Q4 2025 server CPU share, Horizon Robotics China ADAS share, Qualcomm automotive pipeline, Tesla AI5 timing) are sourced from the most recent public earnings filings of each company; references to the AMD Q1 2026 print, hyperscaler April prints, and Mercury / Marvell prints are forward-looking and post-date Intel's April 23 earnings.
References & Sources
- Intel Corporation — Q1 2026 Earnings Release (Exhibit 99.1), April 23, 2026
- Intel Corporation — Form 10-Q for the quarter ended March 28, 2026
- Intel Corporation — CEO/CFO Earnings Call Comments, prepared remarks of Lip-Bu Tan and David Zinsner, April 23, 2026
- Intel Corporation — Q1 2026 Earnings Presentation
- Silicon Analysts — TSMC 1Q26 Earnings: The Capacity-Rule Break Is the Real Story
- Silicon Analysts — Foundry Allocation Status — Q1 2026
- Silicon Analysts — Intel Secures $7B from Rivals to Challenge TSMC Foundry Reign
- Silicon Analysts — NVIDIA AI Accelerator Market Share 2024–2026