Record Revenue $2.006B (+58% YoY), Non-GAAP EPS $0.67 (+123%), AI Day Delivered $94B 2028 TAM + 18 Custom Design Wins + 20% Market Share Goal, Custom Lumpy in Q3 (Q4 Stronger), Auto Ethernet Sale Closed for $2.5B — Maintaining Hold

Coverage Update from Q1

We initiated coverage at Hold three months ago at ~$67 with a $60–$85 fair value range, citing the Trainium 3 multi-vendor question and the pending June 17 AI Day as the two binary catalysts. The June 17 event delivered cleanly: $94B 2028 TAM (vs prior $75B), explicit 18-win design count, 20% market share goal (vs prior 15%). Stock rallied through June and July to ~$75. The Q2 print confirms the fundamentals are excellent — record revenue, +123% EPS, AI Day framework intact, auto Ethernet sale closed early. The new wrinkle is the Q3 custom lumpiness disclosure: customers digesting Q2's heavy build, then ramping again in Q4. This is normal early-stage custom dynamics but creates near-term modeling noise. We maintain Hold because the long-term thesis improved but the near-term setup got slightly noisier; we want Q4 validation before upgrading.

Results vs. Consensus

Q2 Scorecard

Year-over-Year & Q1-to-Q2 Comparison

Quality-of-Beat Callout

Revenue Assessment

Revenue +58% YoY to $2.006B — first quarter above the $2B threshold. Data center +69% YoY, +3% QoQ. The QoQ data center growth slowed from Q1's +5% as custom shipments concentrated in Q1/Q2 with some digestion entering Q3. Enterprise networking + carrier combined revenue +43% YoY (vs +14% QoQ Q1, +2% QoQ Q2) — recovery decelerating sequentially but YoY comparison remains strong. Consumer +84% QoQ (seasonal rebound, gaming). Auto/industrial flat. Q3 guide: $2.06B (+3% QoQ at midpoint, +36% YoY) — implies sequential growth continuing but at lower pace as custom digests.

Margin Assessment

Non-GAAP gross margin 59.4% — at the low end of guidance range. Mix-driven from continued custom expansion + strong consumer rebound (consumer typically below-average gross margin). Per Meintjes Q2 guide framework: Q3 gross margin expected 59.5%–60% — modest expansion as custom mix moderates seasonally. Operating margin 34.8% (+60bp QoQ, +870bp YoY) — operating leverage continues. Non-transaction OpEx came in slightly below guide ($493M vs ~$495M); Q3 guided to $485M (down sequentially).

EPS / Cash Assessment

Non-GAAP EPS $0.67 (+123% YoY). $462M operating cash flow demonstrates strong conversion. Capital structure: $1B note offering in June used to repay existing debt (net debt unchanged in spirit). Cash $1.2B; gross debt-to-EBITDA 1.63x; net debt-to-EBITDA 1.19x (both improving). Buyback $200M in Q2; $540M YTD; $2B remaining authorization. Auto Ethernet $2.5B cash receipt at quarter-end (closing was early Q3) — additional $2.5B redeployable. The capital position is strong and getting stronger.

The AI Day Framework — Refresh of the Multi-Year Outlook

Marvell's June 17 Custom Silicon Investor Event was the principal catalyst of the past quarter, delivering substantial expansion of the multi-year framework from AI Day '24. The key disclosures:

Assessment: The framework expansion is substantial and credible. The 18 wins disclosed include several already in volume production, with the remainder ramping over the next two years. The pipeline of 50+ new opportunities — with $75B aggregate lifetime revenue potential — is materially larger than the market had anticipated. Per Chris Koopmans on the call: "Since our event in June, our team has won additional sockets, adding to the 18 sockets we had already discussed. Collectively, these new wins represent multibillion-dollar lifetime revenue potential." Net: AI Day '25 firmly upgraded the long-term opportunity. The Hold rating reflects near-term execution (Q3 lumpiness) rather than long-term framework concern.

Segment / Product Detail

Data Center — $1.49B (+69% YoY) — Custom Lumpiness in Q3

The Custom Q3 Lumpiness

Custom revenue grew strongly Q2 with first half of FY26 well above prior expectations. Q3 expected to decline sequentially as customers digest Q2's heavy build. Q4 substantially stronger than Q3 — custom revenue resumes growth. "Custom growth next fiscal year to be higher in the second half ... fourth quarter substantially stronger than the third."

"I think you captured the right phrase, which is lumpiness. I think this is normal to see, particularly with the large hyperscale builds that happen and especially as you ramp them into production, which we've done this year on a number of programs. So this is not unusual. Fortunately, our optics business is quite strong in the coming quarter, and that's growing double digits. And then as we said — as I said in the prepared remarks, we see a demand increase again in custom. So yes, there's nothing unique there, Ross, other than we've spent the last couple of years ramping these into production, and we've got kind of a 1 quarter digestion with the recovery in Q4."
— Matt Murphy, CEO

Assessment: The Q3 custom lumpiness is the central debate. Two readings: (1) operationally normal — large hyperscale build cycles produce digestion quarters; the Q4 rebound + 2H>1H framing supports this; (2) competitive concern — Q3 weakness may reflect not just timing but share concession. The Q4 rebound (which will be visible in the December print) is the test. We assume Reading (1) but acknowledge the residual question.

18 Custom Design Wins — The Diversification Story

Per the AI Day '25 disclosure, Marvell has 18 multi-generational XPU and XPU-attach socket design wins. Several already in volume production; remainder expected to ramp over the next two years. Per Chris Koopmans: "Since June, we've added more wins ... these new wins represent multibillion-dollar lifetime revenue potential." The diversification expanding beyond the initial 3-4 hyperscalers into:

• Existing major hyperscalers (AWS, Microsoft, Google, Meta — Marvell + competitors split)
• Next-wave hyperscale-class customers (sovereign data centers, enterprise hyperscale)
• XPU attach (CXL, retimers, NICs, AECs) — entirely incremental category not in AI Day '24 framework

Assessment: 18 wins (vs ~3 in AI Day '24) is the meaningful disclosure. The composition matters: not just additional XPU wins but XPU-attach attaches that ride with every XPU in the network. Materially more durable revenue base than dependence on a single XPU socket at a single customer.

Scale-Up Networking — The Emerging Vector

Murphy detailed the scale-up networking opportunity:

• Scale-up switches supporting Ethernet + UALink fabrics
• Marvell investing to develop scale-up switches tailored to each customer's protocol of choice
• AEC + AOC DSPs already in market
• Retimers in customer evaluation
• 6.4T silicon photonics light engines demonstrated for NPO + CPO
• AECs design wins at 2 Tier-1 US hyperscalers + multiple emerging hyperscalers

Murphy framed this as "a massive scale-up opportunity for Marvell over time."

Assessment: Scale-up is the third pillar (after custom + electro-optics). Scale-up TAM emerging from $1-2B today toward $5-10B+ by 2028-2030. Marvell's positioning (DSPs + photonics + switches) is genuinely best-in-class. Near-term contribution modest; multi-year contribution material.

Electro-Optics — Strong Q3 (Double-Digit Sequential Growth)

1.6T PAM DSPs entering volume production at multiple customers. 800G demand remains very strong with long life cycle ahead. 400 gig per lane demonstrated at OFC (path to 3.2T optical).

Q3 outlook: electro-optics +double-digit sequential growth — offsets custom Q3 decline. Optics + custom now >75% of total data center revenue.

Data Center Storage + Switch — Solid Progress

Data center storage revenue significantly improved (both SSD + HDD markets returning to health). Switching: 12.8T high volume; 51.2T ramping. Microsoft Azure HSM (hardware security module) collaboration expanded.

Enterprise Networking + Carrier Infrastructure — +43% YoY (Combined)

Q3 outlook: combined +30% sequential. Strong inventory normalization + new product cycle adoption. Enterprise networking annual revenue run rate approaching $1B (vs $900M trough in Q1 FY25); carrier roughly doubling YoY in Q3 implied.

Assessment: The communications recovery is the underappreciated tailwind. Off the Q1 FY25 $900M trough, the business is now at a ~$1.5B annualized run rate (Q2 + Q3 implied). Long-term target ~$2B combined.

Consumer + Auto / Industrial

Consumer Q2 $116M (+84% QoQ, +30% YoY). Auto/industrial $76M (flat). Q3 consumer expected modest sequential decline (small seasonal step). Auto/industrial Q3 ~$35M (post divestiture; includes $5M of stub auto Ethernet revenue prior to August close).

Key Topics & Management Commentary

1. AI Day Refresh — $94B TAM, 18 Wins, 20% Share, $75B Pipeline

"During the quarter, we hosted a highly successful custom silicon investor event in June, where we outlined an expanded $94 billion data center TAM for calendar 2028, a 26% increase from our prior view. We also unveiled a new fast-growing custom silicon product category of XPU attach, updated our custom design win board to 18 multigenerational XPU and XPU attached sockets and highlighted over 50 new pipeline opportunities with an estimated $75 billion of lifetime revenue potential."
— Matt Murphy, CEO

Assessment: The framework expansion validates the multi-year thesis. The 20% market share target on a $94B TAM implies ~$19B data center revenue by 2028 — roughly 3x Q2 FY26's $1.49B run rate.

2. Q3 Custom Lumpiness — Operational Normal or Competitive Concession?

The Q3 custom decline is the central debate. Multiple analysts pressed on it. Murphy's framing: digestion of Q2 heavy builds, customer schedule timing. Q4 substantially stronger. "Custom growth ... 2H over 1H."

Tom O'Malley pushed: "is that one project winding down while another is then winding back up?" Murphy: "At a high level, these are existing programs, and it's really just a timing issue in terms of how we deliver the product and when the customers' builds are occurring."

Assessment: The operational explanation is plausible but the timing of the lumpiness narrative (post a quarter of intense investor focus on Trainium 3 competition) creates skeptical priors. We give the company benefit of the doubt — but want to see Q4 sequential recovery materialize as guided.

3. Leadership Reorganization — Koopmans + Bharathi Promoted

Chris Koopmans promoted to President & COO (overseeing sales + non-data-center businesses + corporate development). Sandeep Bharathi promoted to President, Data Center Group (full ownership including engineering + product + customer engagement).

"This unifies full ownership of our largest and most important business under a single leader, spanning the entire product life cycle from technology platform, IP and road map to customer engagement, product definition and chip development."
— Matt Murphy, CEO

Assessment: Strategically important. The Bharathi consolidation gives the data center business genuine single-leader accountability — eliminates handoffs between engineering and customer engagement. The Koopmans expansion gives the non-data-center businesses dedicated executive attention as they shrink (in % terms) within the company portfolio.

4. Scale-Up Networking — Emerging Multi-Billion Opportunity

Murphy detailed Marvell's scale-up positioning across switches (Ethernet + UALink) + interconnect (AECs, AOCs, retimers, silicon photonics CPO). Scale-up TAM ~$5-10B by 2028-2030.

Sandeep on UALink + Ethernet products: "Product introductions in the UALink and Ethernet space for scale-up specifically will be in the next 2 years."

Assessment: A third growth pillar emerges. Initial AEC wins at 2 Tier-1 US hyperscalers + multiple emerging hyperscalers is concrete progress. Scale-up will be the multi-year tailwind through 2027-2030.

5. Auto Ethernet Sale Closed Ahead of Schedule

$2.5B all-cash transaction closed at the beginning of Q3 (vs original calendar 2025 timing). Adds materially to capital allocation flexibility.

"The auto divestiture aligns with our strategy to focus the company on what we expect to continue to be a massive AI opportunity in front of us by purposely redirecting our investments towards data center relative to our other end markets. That strategy has been very successful with data center alone now driving 3/4 of our total revenue."
— Matt Murphy, CEO

End-market reporting simplification: starting Q4, "data center" + "communications and other" (auto/industrial + enterprise networking + carrier + consumer).

6. NVLink Fusion + Custom Complementarity

NVLink Fusion partnership announced in Q1 is moving forward. Customers showing strong interest. Sandeep on engagement: chiplet-based integration model with broad applicability across customer base.

7. AEC + Retimer + LPO Emerging Categories

AEC wins at 2 Tier-1 US hyperscalers. Retimers in customer evaluation. Several LPO sockets secured (Marvell leads emerging LPO category even as it remains a small fraction of overall transceiver market). PCIe Gen 6 retimers gaining broad traction with multiple hyperscalers and partners.

8. Capital Allocation — Buyback + M&A Optionality

"This additional capital really gives us a lot of flexibility around being opportunistic on doing more buybacks. But at the same time, we're at this historic moment in terms of the size of this AI market and where we do see tuck-ins that can accelerate our road map towards addressing that, we'll take advantage of that."
— Willem Meintjes, CFO

Assessment: M&A optionality remains on the table — potential targets include CXL, photonics, networking IP companies. Buyback acceleration continuing; the auto-sale proceeds enable substantial buyback capacity through FY27.

Analyst Q&A

Q3 custom lumpiness drivers + Q4 rebound

Q: "I want to dive into the guidance for the custom business, Matt. I appreciate the lumpiness of it, but could you give any more color on what the headwinds are in the third quarter? And then what gives you the confidence and any sort of magnitude on the increase in the fiscal fourth quarter?"
— Ross Seymore, Deutsche Bank

A: "You captured the right phrase, which is lumpiness. I think this is normal to see, particularly with the large hyperscale builds that happen and especially as you ramp them into production, which we've done this year on a number of programs. So this is not unusual. Fortunately, our optics business is quite strong in the coming quarter, and that's growing double digits. … we see a demand increase again in custom. … we've spent the last couple of years ramping these into production, and we've got kind of a 1 quarter digestion with the recovery in Q4. I will say that overall, we expect custom to be up in the second half over the first half. And so you should expect a strong fourth quarter for custom."
— Matt Murphy, CEO

Design win momentum since June 17

Q: "How much of your custom products revenue that you expect in the second half is coming from some of these new programs? And how much is coming from some of your existing design wins?"
— Jeremy Lobyen Kwan (for Tore Svanberg), Stifel

A: "It's truly an exciting time to be in the custom silicon business for data center. We have a tremendous amount of design activity, more than I've ever seen in my 9 years at Marvell. … even since our event in June, where we said that the XPU attach opportunities were in the sort of several hundred million dollar design win lifetime, that's grown from there. Some of the ones we're chasing now are much, much larger than that. … since June, the design wins that we've added, these are very meaningful, thinking the billions of dollars for the new design wins. If you put it all together, it just gives us even more confidence in our 20% share target."
— Chris Koopmans, President & COO

Customer concentration + lumpiness timing

Q: "Kind of building on the earlier question, just to level set, as we think about the lumpiness in the custom XPU business, I'm curious, you've had obviously a very talked about lead customer. I'm curious, as you're looking at the business today, how concentrated are you amongst your lead customer? And if we look out, let's say, 6 months or even 12 months from now, how do we expect to see some of these additional design wins start to fold into the XPU revenue stream?"
— Aaron Rakers, Wells Fargo

A: "We had started a few — just a few years back a couple of AI days ago, really talking about a handful of sockets that were going to be kind of our initial lead and those have now ramped and are ramping, albeit lumpiness we're seeing in the short term. … then on top of it, the 18 we talked about just a couple of months ago at the AI Day, those are all either starting now, next year, really in the next sort of, I'd call it, between now and the next 18 to 24 months, those will all start to layer in. … we've actually secured some incremental wins. So think of it as kind of 18 plus. … And then as Chris mentioned, it's almost episodic right now."
— Matt Murphy, CEO

Scale-across DCI opportunity

Q: "I know NVIDIA this week talked about scale out or scale across networks. I'm curious how Marvell sees this opportunity moving from just not scale out and scale up but scale across DCI. Any framing of how big of an opportunity that might represent for Marvell?"
— Aaron Rakers, Wells Fargo

A: "In terms of scale-up opportunities, there is certainly aside from the lead GPU player who has its own proprietary scale-up fabric, there's a huge demand for Ethernet and purpose-built fabrics such as the UALink. And we see a lot of traction over the next couple of years for the scale-up requirements. And we are investing heavily to bring our scale-up switches to the market, and we see momentum in the next couple of years."
— Sandeep Bharathi, President Data Center Group

Q4 acceleration + 2026 visibility

Q: "Do you think Q4, your data center growth can accelerate year-on-year from the Q3 levels that you gave, just so that we kind of levels that are our models. And then as we look at 2026, one of your XPU competitors has suggested their business can grow 60%. … Do you think Marvell has the visibility today around timing and magnitude of your large projects to kind of say that your business can sort of grow in line with what the industry expectations are?"
— Vivek Arya, Bank of America

A: "We don't do an annual guide, and we typically just guide a quarter at a time. … on the annual stuff, we've only done that very, very rarely, and that's typically been later in the year as we have more visibility. … the overall momentum in the business has been very strong for several quarters now. … if you look at our optics performance, Q2, Q3, especially the Q3 up double digits. … the very strong recovery in the core business in enterprise networking and carrier. … For reference on that business, we hit a low point during the inventory recovery cycle at about a $900 million annualized run rate. And implied in our Q3 guide, this business goes back up at like a 17 run rate. So very, very strong recovery."
— Matt Murphy, CEO

Q3 custom decline — project winding down?

Q: "I'm going to hit on the ASIC topic again. So apologies, I just want to dive in for a little more clarity. But when you look at what the digestion that's occurring in the October quarter is, is that one project winding down while another is then winding back up in the fourth quarter? Is that just a temporary pause?"
— Tom O'Malley, Barclays

A: "At a high level, these are existing programs, and it's really just a timing issue in terms of how we deliver the product and when the customers' builds are occurring and when they want the product from us. So given that this is — as I said earlier, we're in the early stages of custom. This is really our first big first year with the handful of sockets that will translate over to many more. It's really just a timing issue between the quarters. So it's just more apparent."
— Matt Murphy, CEO

Optics supply chain + laser constraints

Q: "You're guiding to double-digit growth in the October quarter. You've heard others during this earnings period talk about supply constraints, particularly on the laser side. You're obviously a component provider that's going into these modules. But in terms of the ecosystem, are you seeing any stops and starts there in terms of product ramps as well?"
— Tom O'Malley, Barclays

A: "We've ramped this optics business just massively, okay, over the last few years and very successfully. … We have deep partnerships up and down the supply chain and with the key module companies to really plan our business together. So there seems like there's always something going on, but I think we've been able to just manage through it and continue to grow quite dramatically."
— Matt Murphy, CEO

Capital allocation framework — buyback vs M&A

Q: "If you look at your automotive Ethernet business, that's a very attractive price you're able to get from that. So maybe you can maybe talk a little bit about the intended use of the proceeds, whether your bias is more towards tuck-in acquisitions that will allow you to pursue the AI strategy even faster or buybacks? And then more broadly, are you open to potential sale of other components of the business at the right price, whether that be carrier, consumer or otherwise?"
— Jim Schneider, Goldman Sachs

A: "We have run since August 2016 … we implemented a strategy process, which was really our capital allocation framework on how we think about investing our R&D dollars primarily. … And what you've hopefully seen is that over time, we've continued to evolve the company from really a consumer enterprise kind of focused company to a data center AI-first company. … this additional capital really gives us a lot of flexibility around being opportunistic on doing more buybacks. But at the same time, we're at this historic moment in terms of the size of this AI market and where we do see tuck-ins that can accelerate our road map towards addressing that, we'll take advantage of that."
— Matt Murphy, CEO & Willem Meintjes, CFO

Lead customer 3nm follow-on

Q: "The noise level out of Asia on your lead customers' follow-on 3-nanometer XPU program continues at this deafening pace, right? With your Asia competitor, they're essentially claiming victory on 3-nanometer. So what's the update with Marvell's 3-nanometer XPU follow-on program with your lead customer?"
— Harlan Sur, JPMorgan

A: "We're at a point where the initial programs and wins we have are ramping. We've increased our opportunity set pretty significantly to — from a handful of sockets to this 18-plus. And we're really driving to the market share targets in the future. … commenting on just the individual sockets at this point is only probably increasing the noise level. And what we're really focused on is winning incremental designs, executing on the ones we've got and driving the business forward and ultimately trying to get 20% of the $90-plus billion TAM in the future."
— Matt Murphy, CEO

Scale-up switch fabric timing

Q: "Just kind of on the scale-up switch fabric opportunity. … Just wondering, can you give us a sense when Marvell may have its first products ramping to revenue? Is that a calendar '26 event? Or is it going to be more UALink based and more likely a calendar '27 event?"
— Quinn Bolton, Needham & Company

A: "We are investing to bring UALink and Ethernet-based products as we engage with our customers and working very closely with our customers' time line, what I would say is product introductions in the UALink and Ethernet space for scale-up specifically will be in the next 2 years. And certainly, with the assets that we have, not only are we looking at UALink-based products in interconnect, we're already starting to see the use of AECs in the near term and AOCs all for — which is active electrical cables and active optical cables, positioning us to participate in these markets."
— Sandeep Bharathi, President Data Center Group

What They're Not Saying

• Specific XPU custom revenue breakout. Marvell continues to bundle custom + electro-optics + storage within data center; no explicit revenue line for custom alone.
• No quantification of Q4 custom magnitude. The "substantially stronger" framing is the only direction; no dollar magnitude.
• Lead customer not named. Industry consensus assumes AWS Trainium; Marvell continues confidentiality.
• Second hyperscaler not confirmed. Industry presumed Microsoft; not confirmed.
• No FY27 framework yet. Murphy maintains quarter-at-a-time guidance discipline.
• Limited disclosure on LPO competitive positioning. "Several LPO sockets" but no specific share characterization.

Market Reaction

• Pre-print (Aug 27 close): ~$74. Stock had rallied from May ~$67 levels through June (AI Day) and July (positive sell-side notes). Some Q3 lumpiness fears built in pre-print.
• Day-of (Aug 28): Print landed after-hours. Initial reaction modestly negative on the Q3 custom decline disclosure. As call progressed and AI Day framework reinforcement + Q4 rebound framing came through, stock recovered. After-hours roughly flat.
• Day-after (Aug 29): Stock closed ~$74 (flat). Volume ~30M shares (~1.8x 30-day average). High-conviction flow.
• Peers: AVGO −0.3%; AMD +0.2%; NVDA +0.1%. AI-semis mixed without clear differentiation.
• Sell-side flow: Mixed reaction. Bulls highlighted AI Day framework + 18 wins + auto sale executed. Bears highlighted Q3 custom decline + decelerating QoQ in data center. Several PT raises ($80–$95 range); some downgrades to Hold from Buy.

Interpretive read: The market processed Q2 as "fundamentals strong, but Q3 setup creates near-term uncertainty." Stock at ~$74 reflects the trade-off — AI Day improved the long-term framework + auto sale executed cleanly, but Q3 custom lumpiness defers the operational validation by one quarter. Hold rating reflects this balance.

Street Perspective

Debate 1: Is Q3 custom decline operational normalcy or competitive concession?

Bull view: Hyperscale build cycles produce digestion quarters. The 2H>1H + Q4 rebound framing is consistent with normal operational lumpiness. Marvell's lead-XPU program is mature; Q3 is timing-only. The 18-win pipeline + strong electro-optics offset support the operational narrative.

Bear view: The Q3 decline coincides with the multi-quarter Trainium 3 controversy. The framing of "lumpiness" may mask a structural share concession at the lead customer. The Q4 rebound needs to be validated.

Our take: We side with the bull case (operational lumpiness), but acknowledge that Q4 validation is required. The Q4 December print is the key inflection. If Q4 delivers the promised custom rebound + total revenue acceleration, the bear case is invalidated. If Q4 disappoints, the share-concession narrative becomes the consensus.

Debate 2: How much is the AI Day framework worth?

Bull view: The $94B TAM x 20% share = $19B 2028 data center revenue, plus ~$1-2B from comms/other = ~$21B total revenue. At ~40% op margin + ~10% tax = ~$5+ EPS by FY29. At a 25-30x multiple = $125-150 stock — material upside from $74. The framework adds optionality of further TAM expansion.

Bear view: The 20% market share goal requires multi-year execution in a highly competitive market. AVGO has substantially diversified custom franchise; Marvell's diversification is in progress. The framework is plausible but the path is uncertain.

Our take: The framework is credible and worth meaningful premium to the current multiple. Stock at $74 (~22x forward EPS) trades at a discount to AVGO and to Marvell's historical 28-32x range. We see fair value $65-90 today with the upper end requiring Q4 validation.

Debate 3: Is the multiple appropriate given the long-term opportunity?

Bull view: AI-semis with multi-year visibility should trade at premium multiples. AVGO at ~35x, NVDA at ~40x — Marvell at ~22x is conservatively priced. As the design-win diversification materializes through CY2026-CY2027, multiple expands toward AVGO-style ~30x.

Bear view: The discount is rational given Marvell's smaller scale, less diversified customer base, and ongoing Trainium 3 controversy. The multiple shouldn't expand until visible diversification materializes.

Our take: The multiple discount narrows as Marvell delivers multi-quarter execution on the AI Day framework. Q4 validation is the next anchor. Hold rating reflects opportunity but with appropriate near-term caution.

Model Implications & Thesis Scorecard

Model Update

We modestly raise FY26 + FY27 estimates following Q2 + AI Day framework:

• FY26 revenue: $8.0-8.2B (+35-40% YoY) — unchanged from Q1 framing
• FY26 EPS: $2.75-2.90 — unchanged
• FY27 revenue: $10-11B (+30-35%) — modestly raised
• FY27 EPS: $3.80-4.30 — modestly raised
• FY28 revenue: $13-15B; EPS $5.0-6.0
• Long-term framework: $19B+ data center by CY2028 per AI Day '25

Thesis Scorecard

Rating & Action

Maintaining Hold. The Q2 print delivered on what we expected from initiation — record revenue, strong operating leverage, AI Day framework expansion materially upgrading the multi-year opportunity. But the Q3 custom lumpiness disclosure creates near-term uncertainty that warrants holding rather than upgrading. Three factors drive the rating:

1. Q3 sequential custom decline introduces fresh competitive interpretation risk. The "1 quarter digestion" framing is operationally plausible, but the timing aligns awkwardly with the multi-quarter Trainium 3 share-concession debate. Q4 validation is the next anchor.
2. Multi-customer diversification is still early. 18 wins announced, several in production, but the structural diversification across hyperscalers will materialize over the next 12-24 months. The thesis is improving but not fully de-risked.
3. Capital structure repair is in progress. Auto sale proceeds + Q2 buyback step-up + M&A optionality all signal continued aggressive capital deployment — net positive but adds modeling complexity.

Fair value range: $65–$90 (modestly raised from $60–$85). Stock at ~$74 sits in the middle of our range. We upgrade to Outperform on (a) Q4 custom rebound delivered + total data center accelerating, (b) explicit multi-hyperscale custom design-win progression update, (c) FY27 framework formalization. We downgrade to Underperform on (a) Q4 custom decline persisting, (b) Asia supply-chain narrative substantiated through volume data, (c) gross margin compression accelerating beyond model.

Key watch items into Q3:

• Q3 print (early December) — total revenue at $2.06B guide validation; data center composition mix.
• Q4 outlook — does the custom rebound + total acceleration align with promise?
• FY27 framework setup — typically previewed late fiscal year.
• NVLink Fusion customer adoption signals.
• Scale-up switch sampling progress (UALink + Ethernet).
• AEC + retimer revenue ramp.
• Capital allocation — M&A activity, additional buyback acceleration.