Credo Technology Group Holding Ltd

NASDAQ:CRDO

Earnings Call Analysis

Credo sees revenue inflection point driven by AI demand and strong product performance.

In Q2, Credo reported record revenue of $72 million, up 64% year-over-year, fueled by a 21% sequential increase. Their product line, especially AECs for AI applications, drove much of this growth. The company anticipates revenue for Q3 between $115 million and $125 million, reflecting a potential sequential growth of 67%. With a strong design pipeline and increased demand, they project a year-over-year revenue growth of over 100% from FY24 to FY25. Credo maintains a non-GAAP gross margin forecast of 61-63% for Q3 and expects to drive operating leverage as expenses grow at a slower rate than revenue.

Significant Revenue Growth Reflects Strong Demand

In the second quarter of fiscal 2025, Credo achieved remarkable financial results, reporting revenue of $72 million. This represents a 21% sequential increase and an impressive 64% increase year-over-year, significantly exceeding previous guidance. The product business alone contributed $69.1 million, showcasing a 21% sequential growth and an 88% year-over-year increase. This growth indicates strong customer demand and an uptick in product shipments, positioning Credo for a promising future.

Diverse Product Offerings Drive Growth

Credo's success stems from its diverse product offerings, particularly in Ethernet connectivity solutions, which include active electrical cables (AECs), optical DSPs, and line card retimers. The company's focus on solutions tailored for AI applications has notably driven sales, with AECs demonstrating lower power usage and higher reliability compared to traditional optics. Moreover, the launch of the 800-gig ZeroFlaps AEC for AI back-end networks marks Credo’s commitment to innovation in high-speed, reliable connectivity.

Robust Margins and Operational Efficiency

Credo's non-GAAP gross margin for Q2 stood at 63.6%, reflecting improved operational efficiency, with product gross margins reaching 62.2%. The notable increase of 943 basis points year-over-year indicates that the company is benefiting from economies of scale and a higher product contribution margin. With a non-GAAP operating margin of 11.5% for the quarter, up from 3.7% in the prior quarter, Credo is effectively managing its costs while increasing profitability.

Strategic Customer Base and Guidance Outlook

The company’s top three customers accounted for over 10% of revenue each in Q2, with a total of seven customers contributing more than 5%. This diversification of the customer base is crucial in mitigating risks associated with reliance on a few key accounts. Looking ahead, Credo anticipates Q3 revenue to be between $115 million and $125 million, representing a 67% sequential increase at the midpoint. With expectations of year-over-year growth exceeding 100% from fiscal year 2024 to 2025, Credo is positioned for continued robust performance.

Focus on Innovation and Future Growth Drivers

As Credo enters the second half of fiscal 2025, the leadership emphasizes the company is at a crucial inflection point of revenue growth, with significant momentum driven by demand for AECs, particularly in AI server configurations. The company’s ongoing innovations, such as the anticipated 200-gig per lane DSP solutions expected to roll out in fiscal 2026, indicate a strong pipeline that aligns with industry demands for enhanced connectivity. Furthermore, the expectation of ongoing design activity across multiple customers fosters optimism for continued growth beyond the current fiscal year.

Earnings Call Transcript 2025-Q2

[Operator]

Good day, and thank you for standing by. Welcome to the Credo Second Quarter Fiscal Year 2025 Earnings Conference Call. [Operator Instructions] Please be advised that today's conference is being recorded. I would now like to turn the conference over to your speaker for today, Dan O'Neill. Please go ahead.

[Daniel O'Neil, executive]

Good afternoon. Thank you for joining our earnings call for the second quarter of fiscal 2025. Today, I am joined by Credo's Chief Executive Officer, Bill Brennan, and our Chief Financial Officer, Dan Fleming. During this call, we will make certain forward-looking statements. These forward-looking statements are subject to risks and uncertainties discussed in detail in our documents filed with the SEC, which can be found in the Investor Relations section of the company's website.

It is not possible for the company's management to predict all risks nor can the company assess the impact of all factors on its business or the extent to which any factor or combination of factors may cause actual results to differ materially from those contained in any forward-looking statements. Given these risks, uncertainties and assumptions, the forward-looking events discussed during this call may not occur, and actual results could differ materially and adversely from those anticipated or implied.

The company undertakes no obligation to publicly update forward-looking statements for any reason after the date of this call to conform these statements to actual results or to changes in the company's expectations except as required by law.

Also during this call, we will refer to certain non-GAAP financial measures, which we consider to be important measures of the company's performance. These non-GAAP financial measures are provided in addition to and not as a substitute for or superior to financial performance prepared in accordance with U.S. GAAP. A discussion of why we use non-GAAP financial measures and reconciliations between our GAAP and non-GAAP financial measures is available in the earnings release we issued today, which can be accessed using the Investor Relations portion of our website.

I will now turn the call over to our CEO. Bill?

[William Brennan, executive]

Welcome to our Q2 fiscal '25 earnings call. I'll begin by discussing our results for the second quarter, and then I'll provide comments on our outlook. After my remarks, Dan Fleming, our CFO, will provide details on Q2 and guidance. In the second quarter, Credo generated revenue of $72 million, up 21% sequentially and up 64% year-over-year. Q2 non-GAAP gross margin was 63.6%. The second quarter was our most successful to date. We marked record revenue across our 3 main product lines and began to see the uptick in shipments that marks the beginning of the revenue inflection point we've discussed in past quarters.

Today, Credo offers a suite of innovative and differentiated Ethernet connectivity solutions, including active electrical cables or ADCs, optical DSPs, line card retimers, SerDes chiplets and SerDes IP licenses for port speeds ranging from 100 gig up to 1.6 terabits per second. We are also developing a suite of PCIe connectivity solutions, including retimers and [ ADCs ], which we announced and demonstrated at the OCP Summit in October. AI cluster architectures are evolving rapidly, driven by advancements in processing, cooling and power sourcing. These advancements enable increasingly dense and scalable clusters where back-end network reliability has become paramount.

In this environment, addressing or eliminating link lapse, which are momentary disruptions in network links as well as managing power and costs are essential to achieving operational reliability. This rapid evolution is creating significant opportunities for high-speed connectivity solutions, which are crucial in supporting the seamless implementation and operation of these cutting-edge architectures.

Credo's AECs represent a market solution that highlights the company's fundamental differentiation in addressing customer challenges. Built around Credo expertise in SerDes technology. AEC excel in maintaining signal integrity, optimizing power efficiency and delivering unparalleled reliability as data speeds continue to increase. This innovation exemplifies Credo's ability to deliver substantial ROI for customers by solving critical pain points. In Q2, Credo participated in a number of industry conferences, where we demonstrated a broad range of existing and next-generation connectivity solutions that we believe will fuel our continued growth.

For several quarters, we've anticipated an inflection point in our revenues during the second half of fiscal '25. I'm pleased to share that this turning point has arrived, and we are seeing even greater momentum than initially projected. I will now review our businesses in more detail. First, regarding our AEC product line. In Q2, we achieved another record quarter for AEC revenue, driven by strong demand from our top 2 customers and an emerging hyperscaler.

While we're proud of our achievements, we are also optimistic about the future of our AEC business. We believe that we are still in the early stages of widespread market adoption, and we are well positioned as the market leader. AI-driven demand for high-speed, power efficient and reliable connectivity is accelerating. AECs outperform laser-based optics, offering lower power, reduced cost and maybe most importantly, greater reliability. We recently launched our 800-gig ZeroFlaps AECs for AI back-end networks, supporting lengths of up to 7 meters.

With increasing rack power density and the shift to liquid cooling, shorter physical lengths for back-end connections are now possible. This enables AECs to displace optics in certain GPU to switch applications. Optical Linkflaps and AI clusters have become increasingly costly, causing significant downtime and loss of productivity for training clusters. Our AECs offer unparalleled hardware reliability and have demonstrated billions of hours of operations without link flaps. Our system level focus and collaborative approach to solving critical challenges have positioned us as a trusted partner, embedded in the network architecture plans of our customers.

Building on the adoption and engagements highlighted earlier, our second half revenue growth will be driven by our AEC products. Furthermore, we anticipate continued growth beyond fiscal '25 as market adoption expands across the data center ecosystem. Now I'll turn to our optical DSP business. In the second quarter, we achieved record optical DSP revenues, marking another significant milestone for Credo. True to our core values, we continue to innovate alongside our customers delivering enhanced value products backed by exceptional support.

Our optical module customers and end users are increasingly realizing tangible benefits from our optical DSP solutions. This quarter, revenue was fueled by a broad range of 50-gig and 100-gig per lane solutions, including AOCs and transceivers, supporting port speeds from 100 gig to 800 gig. Our unique combination of performance, power efficiency, cost effectiveness and innovation continues to resonate with customers, driving increased adoption across both domestic and international markets.

We believe the market for 50-gig and 100-gig per lane DSP solutions presents significant long-term growth opportunities for Credo. This is supported by robust industry demand and our close collaboration with leading optical module manufacturers and their end customers. Looking ahead, we're enthusiastic about the potential of our 200-gig per lane solutions. We've recently completed the tape-out of our 3-nanometer 200-gig per lane designs with a strong focus on power efficiency.

In calendar '25, we expect to deliver industry-leading full DSP solutions operating at approximately 10 watts alongside LRO solutions at half that power. As always, we remain agnostic to our customers' preferences and are committed to providing tailored high-performance options for full DSP and LRO solutions.

Credo is actively sampling both solutions with leading optical module manufacturers. As architectures continue to evolve, we anticipate growth from both full DSP and LRO deployments, further cementing Credo's role as a disruptive innovator in the optical connectivity market. Now regarding our line card retimer business. In Q2, our line card retimer business also added to our positive overall momentum. During the quarter, we generated record quarterly revenue driven by 400-gig and 800-gig applications.

Our customers included global networking OEMs, hyperscalers and AI appliance manufacturers. In this market, Credo is a leader with solutions for traditional retiming and gearboxing as well as for [ MACsec ] encryption. Across our product set, we have substantial power consumption and cost advantages against competitive products. Going forward, we see continued growth opportunities as line card retimers are adopted for scale app networks in AI appliances and for 100-gig per lane switching applications.

At OCP in October, we introduced our first PCIe retimers, showcasing our Gen 6 and Gen 7 silicon in several demos, including a 1 terabits per second PCIe AEC. Built with our own SerDes technology, we expect to enter the PCIe market with advantages in latency, reach, power and development tools. Customer feedback has been encouraging as they've expressed a need for better performance and development support as they plan the migration to PCIe Gen 6 for scaleup networks. We plan to sample these PCIe products to customers in calendar '25 with production revenues expected in calendar '26.

To sum up, Q2 showcased outstanding execution marked by record revenues, continued innovation and increased customer traction. We are seeing the anticipated revenue inflection point in second half fiscal '25, and we are seeing even greater momentum than initially projected. With demand fueled by AI deployments and deepening customer relationships, Credo is well positioned for sustained growth and increasing profitability.

Now Dan Fleming, our CFO, will provide additional details.

[Daniel Fleming, executive]

Thank you, Bill, and good afternoon. I will first review our Q2 results and then discuss our outlook for Q3 of fiscal year '25. In Q2, we reported revenue of $72 million, up 21% sequentially and up 64% year-over-year and well above the high end of our guidance range. Our product business generated $69.1 million of revenue in Q2, up 21% sequentially and up 88% year-over-year. Notably, our 3 main product lines grew double digits sequentially to achieve new, record revenue levels. Our product business, excluding product engineering services generated another record $64.4 million of revenue in Q2, 20% higher than our previous product record in the prior quarter.

Our IT business generated $3 million of revenue in Q2. As demonstrated by our product revenue ramp, we are seeing substantial opportunities with customer programs on the product side, which we are prioritizing. Given our current product growth trajectory, we do not expect IP to contribute 10% or more of our revenue this year or in future years. However, IP remains a strategic part of our business, and we will pursue select opportunities where the combination of strategic relevance and ROI align.

This prioritization does not impact our long-term model for company-wide non-GAAP gross margin of 63% to 65% or non-GAAP operating margin of 30% to 35%. Our top 3 end customers were each greater than 10% of revenue in Q2. We had another 4 end customers that were each between 5% and 10% of revenue. While customer mix will vary from quarter to quarter, the fact that Credo had 7 end customers at greater than 5% of revenue in the second quarter strongly reinforces how we are diversifying our customer base.

Our team delivered Q2 non-GAAP gross margin of 63.6%, near the high end of our guidance range and up 73 basis points sequentially as a result of improved top line leverage and higher product contribution margin in the quarter. Our product non-GAAP gross margin was 62.2% in the quarter, up 66 basis points sequentially and up 943 basis points year-over-year, primarily due to increasing scale. Total non-GAAP operating expenses in the second quarter were $37.6 million, above the midpoint of our guidance range and up 6% sequentially due to higher R&D project-related spend in the quarter.

Our non-GAAP operating income was $8.3 million in Q2 compared to non-GAAP operating income of $2.2 million in Q1. Our non-GAAP operating margin was 11.5% in the quarter compared to a non-GAAP operating margin of 3.7% in the prior quarter, a sequential increase of 780 basis points. We reported non-GAAP net income of $12.3 million in Q2 compared to non-GAAP net income of $7 million in Q1. Cash flow from operations in the second quarter was $10.3 million, up sequentially, primarily due to our increased profitability driven by leverage from our ramp in product shipments.

CapEx was $21.9 million in the quarter, driven largely by production 5-nanometer tape-outs. And free cash flow was negative $11.7 million, an improvement of $1.4 million from the first quarter. We ended the quarter with cash and equivalents of $383 million, a decrease of $15.7 million from the first quarter. We remain well capitalized to continue investing in our growth opportunities while maintaining a substantial cash buffer. Our Q2 ending inventory was $36.3 million, up $4.8 million sequentially.

Now turning to our guidance. We currently expect revenue in Q3 of fiscal '25 to be between $115 million and $125 million, up 67% sequentially at the midpoint. We expect Q3 non-GAAP gross margin to be within a range of 61% to 63%. We expect Q3 non-GAAP operating expenses to be between $42 million and $44 million. We expect Q3 diluted weighted average share count to be approximately 184 million shares.

As we enter the second half of fiscal year '25, we expect double-digit sequential revenue growth from Q3 to Q4. And from fiscal year '24 to fiscal year '25, we expect year-over-year revenue growth of 100% or more. We expect non-GAAP operating expenses to grow at less than half the rate of revenue from fiscal year '24 to fiscal year '25. As a result, we look forward to driving operating leverage throughout the year.

And with that, I will open it up for questions.

[Operator]

[Operator Instructions] Our first question today will be coming from the line of Toshiya Hari of Goldman Sachs

[Toshiya Hari, analyst]

Bill, maybe my first question is on the outlook for fiscal Q3. Obviously, you're guiding to a level that's well above, what most of us had in our models. Curious if it's really the size of certain projects that's coming in above what you were expecting 90 days ago or a couple of quarters ago? Is it the timing of certain projects? Is it both? What's driving the significant upside in the quarter? You talked about AECs, but if you can expand on that, that would be really helpful.

[William Brennan, executive]

Sure. Yes. We've talked in the past about wanting to have better visibility as as we look at the new projects that we're engaged on and as we're going through qualifications and as we start to see forecasts come together. Really, in the last 90 days, we've seen a couple of projects come together in even a stronger -- with the stronger forecast than we expected. I think this speaks to the overall opportunity that we're addressing, especially with the AEC products. And so I think it's a nice thing as it's come together. As we look towards the quarters that will follow. I think we're encouraged as well, as Dan mentioned about Q4.

And as we think about FY '26, we've been talking about this inflection point for a number of quarters. And we're seeing really healthy design activity across several customers and across all of our products. And so we expect to see continued growth over time. Although we're not able to really give specific guidance for fiscal '26, we do expect the growth to be kind of on the order of what expectations have been in the past and that being on kind of the order of 50% annually.

[Toshiya Hari, analyst]

That's really helpful. And then as my follow-up, One on the competitive landscape in AECs, I think since the time of the IPO, you've always said that competition would be kind of an endorsement of your vision and what you're going after. Many companies have spoken to the AEC opportunity. It doesn't feel like or seem like they've made significant traction, but curious what you're seeing in the marketplace today? And any early feedback on the ZeroFlap AEC product you announced that OCP would be super helpful as well.

[William Brennan, executive]

Yes. So I think that the competitive landscape hasn't changed significantly as we think about over the last 90 days. There's no question that the market wants multiple suppliers in this space. And I don't think there's any question from our perspective that the market is large enough to support multiple players. With that said, I think that we've been clear with the way that we're approaching the market in a sense that we're trying to position ourselves to be first.

First, to deliver unique, differentiated products, first to be qualified with any given customer and ultimately, the first to ramp production qualification. I can't say strongly enough that we view this market and this product family as really a system-level solution. And that means taking ownership of the product from a design all the way through being accountable and responsible in production. It's really a system-level product from our perspective that's highly complex. It's not as simple as putting a retimer into cable. That's under calling the difficulty significantly.

And I think more and more as we go forward and we show to our customers that we can work on unique customized SKUs and turn them around quickly and deliver very high quality and reliable product. I think that's going to continue to be an advantage, and that's going to continue to put us in kind of the pole position on the different hyperscalers that we're working with.

As it relates to the the ZeroFlap AECs, we alluded to this on the last call, but we're seeing that more and more the network reliability for AI clusters is a super high priority, very important thing that -- like if we go back 6 to 9 months ago, this is not something that was talked about so much. As it relates to the announcement that we made in October and what we demonstrated, we're looking at, and we're talking with customers about the advances that are being made with liquid cooling and power sourcing, in particular, as enabling better network reliability in a sense that you can do -- if you go from maximum rack density that's enabled by cooling and power sourcing, you can actually make shorter connections, connections that can be covered with AECs. And AECS are fundamentally more reliable than laser-based optics. And regarding link flaps, we've talked about the effect of a single momentary disruption in the connection in, say, [indiscernible] of AI appliances can basically shut down an entire cluster and it's on the order of 30-plus minutes and also losing productivity in a sense of going back to the start of whatever training that was being done.

And so we're seeing network engineers taking a different approach and looking at different architectures and really trying to build out a ZeroFlap capability. And as it relates to the AECs, we've shipped hundreds of thousands of AECs over the last multiple years. And we have yet to have any customers come back to us and report failures through the Link Flaps. And so that's -- I mean, we're looking at billions of hours in an operating environment. And that's why we're seeing this momentum related to kind of [ row scale ] designs versus what we've previously been focused on, which is rack scale. So hopefully, that gives you more color.

[Operator]

Your next question will be coming from the line of Thomas O'Malley of Barclays.

[Thomas O'Malley, analyst]

I thought the stats that you kind of just mentioned, Bill, the 50% growth into the next fiscal year was the most impressive, and I think that if you look at the growth trajectory for you guys into next year, could you maybe spend a little time on splitting out those growth drivers? Because, obviously, you're seeing a really large ramp in the AEC product already. But in the preamble, you mentioned that you have taped out that 200-gig per lane DSP solution as well. And you mentioned LRO there. Could you maybe comment on what those growth drivers are into the next fiscal year? And to the extent that you can, will that DSP opportunity start to become more impactful in the next fiscal year?

[William Brennan, executive]

Yes. I would say that the way we've talked about the different product lines that we're developing, we've talked about them based on overall market size. And we've talked about the AEC market being the largest market that we're addressing is our largest product line, and we expect that to continue. So with that said, we view optical DSP as kind of being the second largest market that we're addressing.

And this is for Ethernet, of course. And then the copper solutions, so linecard retimers, chiplets and so forth. So we see the growth over time being driven by all of our product lines. And I think that over time, you'll see our revenues almost match the markets that we're addressing. But with that said, for sure, the main driver in FY '26 based on the market size will be AECs. And so I don't think there's any change in the way that we're thinking about Optical being a growth driver, maybe optical from a percentage perspective will be a faster growth product lines than AECs.

But of course, AECs were working from a much larger base. And I'll say, generally, what we're seeing is great design activity across the board. I think the advantages that we bring to the market are clear. They're playing out very well. And so basically, it's kind of the same messaging that we've been giving.

[Thomas O'Malley, analyst]

Super helpful. And then this is just a 30,000-foot view question here. You're seeing the Ethernet market expand into certain different applications. You're also seeing the PCIe market expand. You talked about moving into that market as well. Obviously, playing in both markets allows you to reach the benefits of growth across both. But could you maybe just give your commentary on where do you think the back end moves towards longer term? Is it more towards the Ethernet part of the world or more towards the PCIe part of the world. It's just a constant debate that we kind of run in too daily and would love to hear your opinion on it.

[William Brennan, executive]

Yes. So it's kind of a hot topic for sure. We're a bit agnostic as it relates to ultimately where the market is heading. We know that the market is heading towards faster speeds, and that really plays to our advantage. When we think about our PCI solutions, we think about the advantages that played out in the Ethernet space. And those same advantages that we're bringing to the market will play out in the PCIe space as speeds go to 64 gig and 128 gig per lane.

With that said, Ethernet is already pushing towards 224-gig per lane and there's a lot of conversation in the market about scale-up networks, in particular, moving to faster lane speeds. And UA Link is a topic that is well discussed. And it feels like if that moves towards 224-gig per lane that surely, we're going to be well positioned for that. So we really look at it as really well addressing both Ethernet as well as PCIe long term. And I think the trend towards faster lanes, that's clearly an advantage for us .

[Operator]

And our next question will be coming from the line of Quinn Bolton of Needham & Company.

[Quinn Bolton, analyst]

Congratulations on the nice results and outlook. I guess, maybe to start, Bill, the guidance is well, well above Street expectations as you look into the back half of the year for AEC. You guys, I think, have [ 3 10% ] customers that are driving a lot of that growth. Can you give us some sense as to the number of AECs that ship sort of [ per NIC ] card or per accelerator and how that ratio may be changing as you look from fiscal '25 into fiscal '26. I think in past conversations, you might have implied that you could have one, AEC that aggregates the bandwidth from 2 NIC cards, but I'm wondering if that attach ratio of effectively half an AEC to one GPU, if that may now be a number that ends up being too low? Just trying to get some sense from you guys, what gives you the confidence that the business continues to grow into fiscal '26?

[William Brennan, executive]

Sure. As it relates to architecture, what we're seeing is -- we're really seeing different architectures based on the different deployments at each customer. Now as it relates to attach rates for GPUs, I think we've talked about a trend moving from [ 1 per ] to multiple per GPU. And it's really based on the simple calculation of lane speeds and number of connections. And how does that map to a given GPU strategy? But I think it's one of these things where we're seeing different approaches, and I don't think there's any kind of one size fits all. I do think the trend towards a higher ratio of AECs per GPU will continue to play itself out. .

I don't know if I can tie that directly to our expectations for growth in fiscal '26. What I really tie that to is the fact that we've got great design activity. We've got great qualification activity across multiple customers. And we do see that playing out, again, across all of the different products that we're building.

[Quinn Bolton, analyst]

Maybe, Bill, as a follow-up, as you look into the second half of the year, do you see the business -- is it mostly driven by just -- you guys are seeing an acceleration in the number of GPUs or XPUs deployed? Is it driven by a higher attach rate of some of these new projects go to market with more AECs per GPU or per rack? Or is it a mix shift towards higher ASP devices. You talked about 400 gig in the market now. You've got the 800-gig ZeroFlap AECs launching -- or is it a combination of all 3? Just any color would be appreciated.

[William Brennan, executive]

Sure. I think over time, long term, I think all of those factors will play into our growth. As it relates to what we've seen in the last 90 days, and what we expect through the end of this fiscal year, it's more related to the sheer number of GPUs that are being deployed. That's our perception. These are architectures that we're decided on several quarters ago, and they've been going through the qualification process. I think this -- I think it reflects that the overall opportunity out there for our end customers is big. And as it relates to just the sheer number of GPUs being deployed, what we saw in the last 90 days is that number is increasing.

[Operator]

[Operator Instructions] And our next question will be coming from the line of Tore Svanberg of Stifel.

[Tore Svanberg, analyst]

Congratulations on the strong results. Bill, I have to ask this question just because of history, right, that when we look at this type of growth, what gives you the visibility that there's not some inventory build here, especially given the geopolitical climate and so on and so forth.

[William Brennan, executive]

I think that there's different ways that we look at things. Of course, we've talked about the supply chain several times, about -- from the standpoint of when one of our end customers puts together a build plan, ultimately, how inventory is staged, how sets of products are put together by their different partners. And we've got a pretty good view into actual deployments, what is consumed. And at any given time, you can see that number fluctuate as our customers kind of surge with deployments or maybe [indiscernible] on deployments.

What we're seeing right now is there's a pretty high consumption rate within our customers. We don't see a lot of inventory that's being built within their logistics partners. Now with that said, I mean, I think that all of us within the supply chain are constantly looking for better visibility. Any time that we see a surge like we see right now, of course, in -- from one perspective, you're thinking, okay, what's really happening here. And that's why we're very, very closely following the flow-through of the product that we're shipping.

[Tore Svanberg, analyst]

Yes. That's great color. And as my follow-up, and you mentioned some of this, obviously, Credo is very unique with [indiscernible], whether it's copper or optical PCIe or Ethernet. And I'm just wondering if you think about the large GPU vendor, they obviously chose a very unique system architecture for their latest product. I'm just wondering if that has sort of really driven a lot of your customers to focus so much more on using copper. And is that perhaps why you're starting to see these opportunities being greater than originally thought?

[William Brennan, executive]

I personally don't link those two. I think that all of these AI cluster network engineers are striving for the same thing, but I can't comment about the GPU supplier and the architecture that they're pursuing specifically to the scale up versus scale out. It's hard for me to link those things. What we find is that if we can collaborate and really focus each customer on what the best network architecture is for them, that's where we're seeing success. But I wouldn't say, there's any kind of larger trend or market direction based on one guy going out there with a certain solution.

[Operator]

And the next question will be coming from the line of the Vivek Arya of Bank of America Securities.

[Vivek Arya, analyst]

Bill, You said that the second half marks an inflection point. What is driving that inflection now? Because when I look at Blackwell that product has been pushed out somewhat. So is this just a lot more hoppers that are being deployed? Or is it more that there is perhaps more custom ASIC accelerator deployments. I'm just curious to see what is driving this inflection now?

[William Brennan, executive]

Sure. So I think we've been pretty consistent in talking about this for maybe the last, say, 3 quarters. And we don't see that this is like a shift to, say, towards the leading edge, like you're talking about with Blackwell. This is really playing out based on the long-term plan of architectures that have been frozen. And this is really just the playing out of a long-term plan for deployment. Happening a little bit stronger than expected. But there is really nothing to suggest that this is based on going towards, like, I say, a Blackwell type of an architecture.

I will say that for us, we really don't have visibility into what GPU is being used. In fact, at any given customer, they could be deploying any number of GPUs. What we see is an interface with the NIC card. And so it's really -- we're agnostic from a GPU perspective as to the deployments that we're seeing with customers.

[Vivek Arya, analyst]

Just on the 10% customers, how big were these 3, 10% customers and as you go into Q3, is the demand growing across these 3 10% customers? Or is it being driven more by one of them?

[Daniel Fleming, executive]

Yes, Vivek, this is Dan. Let me address that. So as we mentioned, we had 3 10% end customers in Q2, no surprises as to who they were. The first 2 were our first 2 AEC hyperscalers then the emerging hyperscaler that we described last quarter was the third. So when our Q is filed, what you'll see is that first AEC hyperscaler, which everyone knows is Microsoft, they remained a 10% customer during Q2. They were just above that at 11%. And the second AEC hyperscale customer we've talked about with whom we have a warrant, they were a 33% customer.

So if you do the math, they were actually down sequentially from Q1 to Q2 from $31 million to $24 million. And as we mentioned last quarter, we didn't expect the ramp would always be linear. However, we stated previously that this second half inflection point, the single most important driver of that is really that customer. So obviously, they will become a larger customer as a percentage of the total in our Q3.

And then the emerging hyperscale customer. They were our third customer at 14% of revenue. So from my perspective, one of the key takeaways from Q2 is that we really demonstrated significant revenue diversification from both a customer and product perspective. And clearly, there's a broad-based need for our innovative solutions as land speeds increase throughout the hyperscale deployment environment.

[Operator]

And the next question will be coming from the line of Suji Desilva of ROTH Capital.

[Sujeeva De Silva, analyst]

Congrats on the strong guidance. Maybe just a follow-up on the last question about the hyperscale customer. It's going to be important for the second half. Can you talk about the breadth of programs you're penetrating now maybe versus 6 months a year ago? Is it 1 program that's growing? Or penetrating multiple programs that kind of color would be helpful in terms of diversification.

[William Brennan, executive]

Yes. I think that we're, again, seeing design activity with multiple customers. And we see that continuing over time. And so it's hard for us to maybe talk so specifically about this particular quarter or next particular quarter. But we definitely do see that certain customers will surge. We've talked about how just as the numbers Dan reviewed that, say, Amazon, as an example, that we saw surge in the first quarter something that wasn't linear and up for the second quarter.

And so you can't read too much into a given 90-day period with a given customer. The key for us is to make sure that we've got as much design activity across all of the hyperscalers. And that's really going to fuel our growth long term.

[Sujeeva De Silva, analyst]

Okay. And then I know on the customer diversification staying there, the 2 hyperscalers you're shipping to now. Can you confirm that you're shipping to all 4 of the large hyperscalers? Or is that -- are they in the pipeline? Or just to understand how you broadened out across the 7 large customers yes.

[William Brennan, executive]

Yes. So we've talked about really being focused on the U.S. hyperscalers. And we kind of classify companies within that category, including Oracle. And we've also talked about emerging hyperscalers, companies like xAI, companies like Omniva. And so we've -- what we've talked about is we've talked pretty openly about our relationship with Microsoft, our relationship with Amazon.

We haven't been too clear about exactly which is the hyperscaler that represents customer #3 out of the 5 U.S. hyperscalers. We did mention during our press release during the OCP conference that we've been doing a lot of good work with xAI specifically in the ZeroFlap category. And so I can say that we're doing well really with several customers that will be ramping production. With those other hyperscalers, we're making progress. I've been pretty consistent saying that with all of our products, starting with AEC that I expect that all of the U.S. hyperscalers will be consuming AECs in some form or fashion long term. And whether that's Credo AEC or one of our competitors, the category itself, there's no question in my mind that all the U.S. hyperscalers will be using AECs.

As it relates to optical DSPs, we're really approaching that the same way that we're trying to ultimately penetrate every single end customer. And so I think we're partially there. We're not fully there yet. And so I think that's -- that represents further growth opportunities. p

[Operator]

And our next question will be coming from the line of Vijay Rakesh of Mizuho.

[Vijay Rakesh, analyst]

Just a quick question on the AEC side. I was wondering what percent of revenues was going to the AI side, AI -- into AI servers. And if you would be looking to add new 10% customers in 2025 -- calendar '25, sorry. And I have a follow-up.

[William Brennan, executive]

Yes. So just to restate the question, I think you were asking about AECs in particular, how much is related to AI. Is that correct?

[Vijay Rakesh, analyst]

Yes. That's right. .

[William Brennan, executive]

Yes. I'd say the large majority of the AEC shipments that we're making right now are based on AI applications. That's very clear, especially with the solutions that are 100 gig per lane speeds. There's no question those deployments are all really 100% is based on AI.

[Vijay Rakesh, analyst]

Got it. So the second part in terms of new 10% customers in calendar '25, I guess?

[Daniel Fleming, executive]

Calendar '25, I would not expect additional 10% customers based on what we've laid out. There have, as we mentioned, this emerging hyperscaler tripped that threshold in Q2. So they could trip that threshold again in the future. So it will be kind of a repeat of we've had is our current expectation through calendar year FY '25 or calendar year '25, sorry.

[Vijay Rakesh, analyst]

Got it. And on the optical DSP side, just wondering, obviously, a nice ramp there. But how many customers do you have on the optical DSP side now?

[William Brennan, executive]

Yes. The way that -- you can look at it a couple of different ways. You can look at the optical module manufacturing partners that we're working with and the number of relationships there or you can look at the number of end customers being hyperscalers. And I think we're -- I'm not in a position to articulate the exact number of customers that we're engaged with. But we are seeing much broader design activity at the optical module level, and we're seeing increased activity from end customers for qualifications. .

[Operator]

And the next question will be coming from the line of Joshua Buchalter of TD Cowen.

[Joshua Buchalter, analyst]

Congrats on the results. I was hoping you could maybe spend a minute or 2 talking about your go-to-market and competitive position, specifically on optical DSPs. I mean -- you cited some pretty impressive power stats on your 3-nanometer optical DSP and also mentioned that 50 and 100 gig per lane was hit record numbers. Should we be thinking about the 200-gig product serving as sort of a step function change and how you feel about your competitive position against some pretty big incumbents and 50 and 100 are [ proven ] grounds. I'd be curious to hear you talk about your -- the catalyst for further inflection in the DSP market.

[William Brennan, executive]

Yes. We think that our growth in the second half of this year as well as in fiscal '26 is really going to be driven by 50 gig per lane and 100-gig per lane designs. We are excited about the 200 gig per lane optical DSPs that were just taped out. We believe that market is going to play out on a little bit longer term than than fiscal '26. We really see that becoming a high-volume market probably in our fiscal '27. And so hopefully, that gives you color. The near-term growth is really going to be driven by the large opportunity that we're seeing for 50 gig and more and more so 100-gig per lane designs.

[Joshua Buchalter, analyst]

Got it. And to follow up, you saw a competitor made an announcement about AECs at one of your lead customers this morning. I know you guys have different architectures. But second sourcing and resilience has been increasingly important in the AI ecosystem. I'd be wonder, if you could maybe speak to how you think your customers think about second and third sourcing in the AEC market specifically? And -- are you guys on -- generally, are you going up against -- or are your customers using AECs from multiple vendors on the same projects? Or is it spread across different clusters. Just be curious to hear your thoughts there?

[William Brennan, executive]

Yes. We're definitely seeing our hyperscaler customers have the desire to have multiple sources. No question about that. And what we're also seeing is that if we can continue to play out our strategy of delivering for samples, delivering first qualification units, being the first to be qualified and the first to ramp. That's always a good position to compete from. And so I think we've seen that play out at several -- in fact, all of the customers that we've been working with. And I think we're uniquely organized to be able to deliver customized solutions for given customers. We're in a unique position in a sense that we've got more than 20 SKUs that, that we've developed just in the last several quarters for different customers with different flavors of hardware customization as well as firmware customization.

And so we're really approaching the competition as how can we do the best job at delivering for each customer. I'm very comfortable living with the expectation that all of our customers are going to ultimately second source. So competing from the high ground is really what we're focused on.

[Operator]

And our next question will be coming from the line of Karl Ackerman of BNP

[Karl Ackerman, analyst]

I have 2 as well. First question, I was -- is on your licensing business. I was hoping you could discuss the revenue contribution you anticipate in the January quarter, whether that's also below 10% of your implied outlook. And in regards to your licensing revenue, how do we -- how do you think about the visibility you have in that line item, given, I believe, more of your licensing revenue appears to be royalty based now rather than more of these onetime repayments. So if you could just highlight that, that would be helpful. And I have a follow-up.

[William Brennan, executive]

Well, I'll say that we've been very consistent over the last many, many quarters that we're fundamentally a product company. This theme has been very consistent. And I think that especially with the product revenue inflection point we're seeing right now, that IP licensing will become a smaller percentage over time. We will continue to treat the business strategically. We're really ROI driven as it relates to the different opportunities that we'll see. You asked specifically about hitting a 10% threshold on a much increased number as we talk about guidance.

I would say that my expectation is not to cross that 10% threshold in this current quarter. As we look over time, the -- I think we'll continue to see that this business is somewhat lumpy in the way that the revenue recognition rules are written. And we'll have different IP licenses that we pursue that will be a really good return for the company. But we can't be too specific beyond that.

[Karl Ackerman, analyst]

Sure. Of the 4 customers greater than 5% of revenue that you mentioned, are they primarily concentrated in AECs. Or is the customer [ breadth ] also occurring in your DSP offering as well as line card offering? Maybe you can just speak about the broad segmentation or buckets where these other 4 customers are in, that would be helpful.

[Daniel Fleming, executive]

Yes. It's definitely broad-based is what I would say. And in fact, if you talk about the 10% plus customers that we had, those are largely concentrated in AEC. But we did make comments. We had quarterly strengths, sequential growth among all of our top 3 product lines, AEC, Optical and our line card retimers. So if you kind of -- you could assume correctly that the additional 4 customers we talked about was broadly based, not just AEC related.

[Operator]

Thank you. And this does conclude the Q&A session for today. I would now like to turn the call back over to Bill Brennan for closing remarks. Please go ahead.

[William Brennan, executive]

So thank you very much for all the questions. Really appreciate the ongoing interest in Credo, and I look forward to the callbacks. So thank you very much.

[Operator]

Thank you for joining today's conference call. This does conclude today's meeting. You may all disconnect.