Marvell ships 3.2T retimer and 3nm 1.6T AEC DSP for 224G AI scale-up fabrics

Marvell Alaska E 3.2T retimer and second-gen Alaska A 1.6T AEC DSP are sampling, targeting 224G PAM4 links for AI rack-scale fabrics.

Published Updated 5 min read ✓ AI-ready Share

TL;DR — three sentences

  • Marvell announced the Alaska E 3.2T retimer (16 × 224G PAM4, 45 dB+ reach) for on-board and backplane AI fabrics. (S1)
  • Second-generation Alaska A 1.6T AEC DSP is built on 3nm, fits 12 mm × 14 mm, draws 12.8 W, and supports known-good-die for AEC assemblies. (S1)
  • Both parts support Ethernet, UALink and ESUN, giving one retimer family across switch-NIC, XPU-XPU and AEC links. (S1)

Taken from key_takeaways (real field, max 3 entries). Fact and analysis are labelled separately.

Marvell ships 3.2T retimer and 3nm 1.6T AEC DSP for 224G AI scale-up fabricsEditorial cover for a component sourcing brief. Factory motif; headline "Marvell ships 3.2T retimer and 3nm 1.6T AEC DSP for 224G AI scale-up fabrics". No reported key numbers on this brief, so no number strip is drawn. 1 cited source. A one-time entrance animation stops under reduced motion.Marvell ships 3.2T retimer and 3nm 1.6TAEC DSP for 224G AI scale-up fabricsXAIPART
Cover image generated from data already shown on this page: the category motif, the headline, and reported key numbers. No decorative illustration; no third-party image.

You arrived from an AI answer — this is the verifiable original

An AI summary may have dropped the basis for its claims. The original forecast, the affected parts and every citation are on this page, each traceable to its source.

XAIPART forecast

This brief carries a falsifiable forecast.

Jump to the full forecast and its falsification rule ›
Affected parts

No catalog match for this story — nothing is shown rather than inferring a match.

See how that is decided ›
Sourcing a part? Send a request → RFQ

Reported by Marvell Newsroom on 2026-10-13. Summarized and analyzed by XAIPART Editorial.

What happened

Marvell Technology (NASDAQ: MRVL) announced two additions to its Alaska connectivity portfolio for AI scale-up infrastructure on .S1 The new devices are the Alaska E 3.2T protocol-aware retimer and the second-generation Alaska A 1.6T AEC DSP, both aimed at 224G-per-lane signaling for rack-scale and multi-rack AI fabrics.S1

The Alaska E 3.2T retimer delivers 3.2 Tbps aggregate bandwidth across 16 lanes of 224G PAM4 SerDes, with reported channel reach greater than 45 dB insertion loss compensation, and supports Ethernet, UALink and ESUN protocols.S1 It targets on-board and backplane applications where signal integrity at 224G is the main design constraint.S1

The second-generation Alaska A 1.6T AEC DSP moves to a 3nm process node, integrates eight lanes of 224G PAM4, fits a 12.8 W, 12 mm × 14 mm package and adds a known-good-die format for dense AEC assemblies.S1 Marvell reports copper reach exceeding six meters, which it states doubles the first-generation product, and lists 8 W of power savings versus 5nm designs with ILT support, plus retimer modes from 1.6T down to 50G and break-out configurations including Y-cable.S1

Both devices are available now for sampling to customers and are being showcased at the OCP Global Summit 2026.S1 Marvell positions the launch to support the transition to 102.4T switch fabrics and broader rack-scale AI deployments, and Xi Wang, SVP and GM of Marvell's Connectivity Business Group, framed the portfolio as targeting signal integrity, latency and power efficiency for hyperscalers.S1 LightCounting's Vladimir Kozlov was quoted projecting that the AI scale-up networking market will exceed $50 billion by 2030.S1

MetricReported valueSource
Alaska E 3.2T aggregate bandwidth3.2 Tbps across 16 × 224G PAM4 lanes[S1]
Alaska E 3.2T channel reachGreater than 45 dB insertion loss compensation[S1]
Alaska A 1.6T SerDes lanes8 × 224G PAM4[S1]
Alaska A 1.6T process node / power / package3nm, 12.8 W, 12 mm × 14 mm[S1]
Alaska A 1.6T copper reachExceeding six meters, stated as 2× first generation[S1]
Power savings vs 5nm (with ILT)8 W[S1]
Target switch fabric102.4T[S1]
AI scale-up networking TAM (LightCounting, quoted)Exceeding $50 billion by 2030[S1]
PriceNot disclosed[S1]
Production (non-sampling) availabilityNot disclosed[S1]

Background timeline

  1. - Marvell announced the Alaska E 3.2T protocol-aware retimer and second-generation Alaska A 1.6T AEC DSP, with sampling to customers and a showcase at OCP Global Summit 2026.[S1]
  2. - Marvell stated the second-generation Alaska A 1.6T reaches copper distances exceeding six meters, characterized as double the first-generation product.[S1]

XAIPART take

Technical impact

The Alaska E 3.2T retimer absorbs 16 lanes of 224G PAM4 and 45 dB+ of insertion-loss compensation, which is the headline constraint for board and backplane designers moving from 112G to 224G signaling. The protocol-aware Ethernet/UALink/ESUN triple stack means a single retimer can sit across switch ASIC-to-NIC and XPU-to-XPU links, which matters for rack designers that previously needed different retimers per protocol.

The 3nm Alaska A 1.6T AEC DSP hitting 12.8 W in a 12 mm × 14 mm package with known-good-die support is the part that most affects AEC cable and tray design. Multi-rate gearboxing from 1.6T down to 50G inside one device also reduces the cable-bill-of-materials variants a hyperscaler platform team has to qualify.XAIPART analysis

Market impact

Marvell is positioning two generations of 224G silicon against a LightCounting-quoted $50 billion-by-2030 AI scale-up TAM. The combination of 16-lane on-board retiming and 3nm AEC DSP keeps Marvell competitive against Astera Labs and Broadcom on both backplane and AEC fronts, which is the BOM-relevant question for hyperscaler buyers.

For procurement, the immediate signal is sampling-only status: any 224G retimer or AEC with Marvell silicon on the AVL will be in qualification, not volume, through at least the next two quarters. Lead-time expectations cannot be set from a sampling announcement; component buyers should plan dual-source for 224G retimer sockets and confirm production timing directly with Marvell FAEs.XAIPART analysis

Action

Design engineers should pull the Alaska E 3.2T datasheet and the second-generation Alaska A 1.6T datasheet for any 224G backplane or AEC designs targeting 2027 platform refresh, and verify ESUN and UALink protocol coverage against their scale-up fabric choice. Procurement teams should open NCNRs on sampling parts now if 2027 production slots are a constraint, and request a stated MP date from Marvell rather than assume a sampling-to-production gap.

Buyers sourcing Alaska A AEC assemblies should confirm the known-good-die option availability at their cable vendor, since that directly affects assembly yield and field failure rates at 224G.XAIPART analysis

XAIPART Forecast

XAIPART expects that, by Q2 2027, if Marvell Alaska E 3.2T and second-generation Alaska A 1.6T parts have not been added to at least one hyperscaler's 224G AVL or AEC reference design, this sampling-to-design-in trajectory will fall behind Astera Labs and Broadcom equivalents, and the 102.4T switch / 224G AEC program timeline for 2027 rack-scale AI platforms will slip by at least one quarter. XAIPART analysis

Facts are as reported by the cited source. The design-in, sourcing and forward-looking statements above are XAIPART's analysis, not the source's claims. Uncertainty: pricing, production-volume timing, and end-customer design wins are not disclosed; the $50 billion 2030 figure is a LightCounting quote within the Marvell press release, not a Marvell-provided number.

Corrections: none at publication. This brief will be updated with a timestamp if the source corrects its report.

Alternatives and lifecycle

Alternatives on file — catalog scope, not a market ranking
PartRelationLifecycleBasis
— No alternative on file NEEDS DATA — Requires match_type before any predecessor label may appear

“Predecessor” and “alternative” labels come only from a recorded match type, a lifecycle status, or the source text. Where no match is recorded, no label is shown. Our own listings are not a market reference.

Affected parts

No catalog match. This story does not map to any part currently listed on XAIPART. We do not infer matches, and we do not show a placeholder list.

Cost pressure by dimension

Source: FRED · Generated · reference output z and composite are algorithm outputs, not a fetch snapshot

Cost pressure by dimension — composite stateDiverging bars for three dimensions around a zero midpoint. Input Cost has no reading; Supply Demand z -1.4711, notable; Fx Energy z 0.4541, neutral. Composite index -7.6, band neutral.0Input costno readingSupply/demandz −1.4711 · notable46 dFX & energyz +0.4541 · neutral6 dcomposite −7.6 · neutral−100 · −40 · −15 · 0 · +15 · +40 · +100
Underlying values — copyable for verification.
DimensionFRED series IDUnitLagAs of
Input costPCOPPUSDM, PALUMUSDM, PNICKUSDMUSD/mt96 d
Supply/demandPCU334413334413, IPG3344Sindex(1982=100) / index(2017=100)46 d
FX & energyDEXCHUS, DCOILBRENTEU, DHHNGSPrate / USD/bbl / USD/MMBtu6 d

No free authoritative interface for: gold · silver · rare earth · polysilicon · lithium carbonate · market share — listed rather than approximated.

How this was produced

Process evidence — fetch, source sanity, model and gate
Audit trail for this brief
Source fetched Marvell Newsroom · www.marvell.com · 5,561 chars
Source sanity (S1) ok — passed before any model call (no s1_ block recorded)
Model / prompt MiniMax-M3 · prompt Blog Agent - News Summary (v2.8-final) · 7,874 tokens · 42.9s · run completed
Gate verdict hold · 6/6 applicable checks passed · spec v2.9 · C10 ratio 0.015664 · C10 rounds 0
Time contract 3 <time datetime> values in this brief — 0 violations (+08:00 or date-only)

Every row above is produced by the pipeline, not written by hand.

Editorial score

Deterministic gate result — pure PHP, no model involvement in the verdict
Gate score breakdown 6 applicable checks, 6 passed: 100 percent. Hold items: C13 (falsifiable forecast) — sent to human review. Gate spec: v2.9 (S0 + S1 + C1–C20) · Source affinity (C20): 0.752. Applicable checks passed 6 / 6 = 100% · verdict: hold Hold items: C13 (falsifiable forecast) — sent to human review Gate spec: v2.9 (S0 + S1 + C1–C20) · Source affinity (C20): 0.752 The gate decides publication. The independent audit score is advisory only.

Related coverage

Found an error? Tell us what is wrong — corrections are published alongside the original text and logged. We do not silently edit articles.

Comments

Be the first to comment.

Sources

  1. Marvell ships 3.2T Alaska E retimers and 3nm 1.6T AEC DSPs for 224G-per-lane AI scale-up fabrics — https://www.marvell.com/company/newsroom/marvell-connectivity-portfolio-3-2t-retimer-second-generation-1-6t-aec-dsp.html · rss · accessed

Facts are marked S1/S2 and trace to the linked source. Passages labelled analysis are XAIPART's own reasoning, kept separate from reported facts. Source links are followed rather than nofollow — attribution is intentional.