Marvell Technology shares hit a fresh all-time high on June 19, 2026, jumping 7.27% to $310.58 and pushing the AI networking specialist’s market value above $271 billion, according to Yahoo Finance market data. The Santa Clara-based chip designer is now one of the largest US semiconductor names by capitalization, and the rally was triggered by three converging catalysts: a new flagship AI data-center switch, an executive transition that imports a heavyweight finance chief from Adobe, and renewed Wall Street focus on Marvell’s custom silicon work for Amazon Web Services.
Teralynx T100: The 102.4 Tbps Switch Built for AI Clusters
On June 1, 2026, Marvell announced availability of the Teralynx T100, which it calls the industry’s first 102.4 Tbps purpose-built AI data-center switch. The chip is fabricated on a 3-nanometer process, consumes under 1,000 watts in typical configurations — about 25% less than competing devices — and supports up to a 512-port radix for scale-out AI clusters. Marvell ships it in three packaging variants, including conventional ball-grid array, co-packaged copper, and co-packaged optics, and the silicon supports both the Ultra Ethernet Consortium and Ethernet Scale-Up Networking protocols that hyperscalers are converging on for the next generation of AI fabrics.
“The Teralynx T100 was purpose-built for AI — designed without the legacy baggage that inflates power,” said Rishi Chugh, Marvell’s VP and General Manager of the Data Center Switch Business Unit, in the announcement. Independent analyst Alan Weckel of 650 Group called the architecture’s “advantages in latency, power efficiency, radix scalability and overall TCO” significant. Sampling began in the second calendar quarter of 2026, putting volume shipments in the back half of fiscal 2027 — right when hyperscalers are slated to ramp their next-generation AI training clusters.
CFO Transition: Adobe’s Dan Durn Takes the Books
Four days before the Teralynx launch hit the wires, Marvell announced a top-level finance change. On June 11, the company said Dan Durn would join as Chief Financial Officer effective June 15, replacing Willem Meintjes. Durn arrived from Adobe, where he served as CFO and Executive Vice President from October 2021 through June 2026, steering the software giant through its AI-product transition and a roughly tripling of its market cap during his tenure.
Chairman and CEO Matt Murphy framed the handoff in the press release: “Willem has been an integral part of Marvell, and his contributions have been instrumental to our strong financial and operational positions,” before adding that “Dan brings outstanding credentials and a strong understanding of this business that will serve Marvell well.” Durn, in his own statement, said he was “excited to join the Marvell management team at such a dynamic moment for the Company and the industry.” The market read the appointment as a vote of confidence in the AI cycle — senior CFOs at peak-of-cycle software companies do not usually move sideways to a smaller chip designer unless the runway is unusually long.
The Amazon Trainium Spillover
The third leg of Tuesday’s rally was a less concrete but more consequential catalyst. Reports surfaced overnight that Amazon Web Services is preparing to sell its in-house Trainium AI accelerators to external customers, a move that would broaden the addressable market for the entire custom-silicon supply chain that AWS has built. Marvell is one of the publicly disclosed design partners on AWS custom silicon, and its growing custom-XPU and XPU-attach franchise is precisely the revenue line that benefits if Trainium becomes a merchant chip rather than just an internal one. KeyBanc raised its Marvell price target to $385 from $260 — a 48% bump — and reiterated an Overweight rating on the back of those custom-silicon and optical-networking dynamics, according to Yahoo Finance.
Q1 FY27 Already Showed the AI Inflection
The June news landed against a financial backdrop that was already running hot. In its Q1 FY2027 results reported May 27, Marvell posted total net revenue of $2.418 billion, up 28% year-over-year and $18 million above the midpoint of its prior guidance. The data center segment delivered $1.833 billion, fully 76% of total revenue, up 27% YoY. Non-GAAP diluted EPS came in at $0.80 on a 58.9% non-GAAP gross margin.
The forward number was the real headline. Q2 FY27 guidance: revenue of $2.7 billion at the midpoint — representing 35% YoY growth — with non-GAAP EPS of $0.93 plus or minus $0.05. CEO Matt Murphy went out of his way on the call to flag the booking trajectory: “We are seeing exceptional AI-related bookings, and as a result, we are significantly raising Marvell’s revenue outlook for both fiscal 2027 and fiscal 2028 compared with the guidance we provided last quarter.” He cited “800G and 1.6T scale-out optics, 51.2T Ethernet scale-out switches” and “custom XPU and XPU-attach solutions” as the drivers.
| Metric | Q1 FY27 actual | Q2 FY27 guide (mid) | YoY growth |
|---|---|---|---|
| Total revenue | $2.418B | $2.700B | +35% |
| Data center revenue | $1.833B | n/a | +27% (Q1) |
| Non-GAAP gross margin | 58.9% | 58.75% (mid) | — |
| Non-GAAP diluted EPS | $0.80 | $0.93 | — |
The Optical and Switch Stack Marvell Is Building
To see why the Teralynx launch matters to MRVL, it helps to look at the broader Marvell franchise. The company sells three product blocks into the AI training cluster: (1) electro-optical PAM4 DSPs that drive the 800G and emerging 1.6T pluggable optics linking GPU racks together, (2) Ethernet switch silicon — previously 51.2T, now 102.4T with the Teralynx T100 — that powers the top-of-rack and spine layers, and (3) custom XPU and XPU-attach ASICs co-designed with hyperscalers like AWS. The first two compete head-on with Broadcom, the largest incumbent in AI networking silicon, while the third puts Marvell next to Broadcom and a handful of much smaller players at the negotiating table for hyperscaler design wins.
What to Watch From Here
The bull case stitches together as follows. Marvell’s Q1 FY27 backlog already reflected “exceptional AI-related bookings” by Murphy’s own description, and Q2 guidance implies a step-up in growth to 35% YoY. The Teralynx T100 launch broadens the data-center attach beyond optics into switching just as hyperscalers begin specifying their next-generation AI fabrics. The Durn hire signals operating discipline as the company scales. And any expansion of AWS Trainium beyond captive use adds incremental dollars to the custom-silicon book without altering Marvell’s existing design relationships.
The bear case is the same one that haunts every AI infrastructure name: competition. Broadcom is the deeply entrenched incumbent in both AI networking switches and hyperscaler ASICs, and a slip in either franchise — an OEM losing share, a pricing concession on a renewed contract, or simply a softer AI capex cycle — would hit Marvell’s 27%-plus growth rate hard. Investors paying for $0.93 of Q2 EPS at a $310 share price are pricing in continued execution, not just continued demand.
Sources
- Marvell Technology — “Marvell Announces Availability of Industry’s First 102.4 Tbps Switch Purpose-Built for AI and Cloud Data Center Infrastructure,” June 1, 2026
- Marvell Technology — “Marvell Announces CFO Transition,” June 11, 2026
- Marvell Technology — Q1 FY2027 results press release, May 27, 2026
- Yahoo Finance — MRVL quote and news (price action, KeyBanc PT, all-time high coverage)
Disclosure: This article was produced with AI assistance and reviewed before publication. It is for informational purposes only and is not investment advice.