The Zhitong Finance App learned that optical connectivity giant Credo Technology (CRDO.US) announced its latest results after the US stock market on Tuesday. Financial reports show that in the first fiscal quarter of the 2027 fiscal year ending August 1, 2026, the company's revenue increased 114.7% year over year to US$479 million, higher than analysts' average forecast of US$472 million; adjusted earnings per share increased 130.8% yoy to US$1.20, also higher than analysts' average expectation of US$1.17. Looking ahead, Credo expects revenue for the second fiscal quarter to be $525 million to $535 million, and the median forecast range of $530 million is higher than analysts' average forecast of $516.5 million.
However, Credo's gross GAAP margin for the first fiscal quarter fell 290 basis points year over year and 370 basis points month over month to 64.5%, while the median guidance value for the second fiscal quarter fell to 63.9%. Even though the company announced strong core performance indicators and revenue forecasts, relatively weak gross margin data caused the company's stock price to weaken after the market to a certain extent.
Credo's management's speech at the earnings conference call showed that the AI infrastructure is still in a period of rapid expansion. Credo is upgrading from a high-speed interconnect supplier with AEC as the core to a “system-level connectivity platform” covering copper interconnect, optical DSP, silicon photonics, complete optical transceivers, and next-generation NPO/scale-up connectivity solutions. The optical business will become the most important growth engine in the future.
Here are the details of Credo's first fiscal quarter earnings call.
I. Management speech
William Brennan — President, CEO and Chairman
The first fiscal quarter was another strong quarter for Credo. Revenue reached a record $479 million, up 10% month-on-month and more than doubled year-on-year. Non-GAAP gross margin was 68%, and non-GAAP net profit exceeded US$236 million, up 140% year over year. Credo is growing at a rate that few semiconductor companies can achieve, and we are continuing to expand our profitability.
At the core of this growth is our seven consecutive quarters of three-digit year-over-year revenue growth. We continued to see extremely significant growth in the 2027 fiscal year, with the optical business growing the fastest. Investments in AI infrastructure are still growing rapidly. Clusters are growing in size, data rates are increasing, and connectivity requirements are becoming more complex. As these systems continue to expand, connectivity is about far more than just bandwidth. Reliability, energy efficiency, signal integrity, telemetry, and maintainability are all important.
We also believe AI infrastructure will become increasingly heterogeneous. The future will not have one architecture, one protocol, or one physical medium that will be used for all connections. Future AI systems will use a combination of optical interconnect and copper interconnect according to different transmission distances, protocols, and topologies, and customers will choose the right technology to optimize their own architectures.
That's where Credo excels. While network reliability has always been a key metric for Credo, our ability to innovate, execute, certify, and deploy to a wide range of customer needs is our core differentiator.
Our focus is to help customers launch clusters faster, maximize processor utilization, and maintain reliable operations at scale. Today, our product portfolio covers connectivity needs from mm to km, and includes optical and copper connectivity solutions.
Let me dwell on these areas in more detail.
The first is AEC (active copper cable). AEC is still our biggest business and continues to grow. Today, we have established deep relationships with 5 hyperscale cloud service providers, and our cooperation with NeoCloud (New Cloud) customers continues to expand.
Among existing customers, we continue to see an increase in AEC penetration as deployments scale up. Higher data rates are also providing another growth driver, as the transition to 200Gbps, 1.6T ports per channel is underway.
AEC has always been a system-level product of Credo. We provide complete solutions and co-optimize chips, firmware, manufacturing testing, and system certification. This system-level approach has been our core differentiator since we created this product category.
As AI clusters grow larger, their value proposition remains clear: for short distance connections, high reliability and low power consumption, or both, are becoming increasingly important in large-scale deployments. We continue to see AEC maintain a healthy growth trajectory, mainly driven by deeper penetration among existing and new customers, as well as increasing bandwidth in next-generation clusters.
Now it's time to switch to the optical business. Our optical business is progressing very well, including optical DSPs, silicon photonic PICs, and ZeroFLAP optical transceivers. Our optical DSP business achieved record revenue in the first quarter. Revenue includes deployment of 50 Gbps and 100 Gbps per channel solutions.
Even as we begin the transition to a 1.6T solution, we still think the 800G port has a long life cycle. At 200Gbps per channel, customer engagement with our 1.6T DSP was high, covering fully retimed (fully retimed) and LRO solutions. Our first 1.6T DSP revenue is still on schedule to be realized later this fiscal year.
This quarter, following the completion of the acquisition of DustPhotonics, we also confirmed our first silicon photonic PIC revenue. Our initial order was won from 800G and 1.6T optical transceivers, and we expect these products to continue to be released throughout the year.
More importantly, silicon photonic PIC adds another important technology to our optical platform. Today, we can co-optimize DSP and PIC and combine them with firmware, telemetry, and PILOT software. This level of integration creates opportunities to improve reliability, power consumption, signal integrity, and diagnostic capabilities.
As scale-up architectures evolve towards near-package optics (NPO), this also puts us in an advantageous position. We are seeing increasing customer NPO activity around the scale-up network, and confirmed design wins are expected to start growing in FY2028. As part of the Open CPX MSA Alliance, we're bringing many of the benefits of the current pluggable ecosystem, including telemetry, interoperability, and maintainability.
Our opportunities include both optical components and complete system-level solutions. This is an important evolution for Credo. In the past, we have been very successful in solving connectivity issues at the semiconductor and AEC system levels. Today, we're applying the same approach to the field of optics.
Our ZeroFlap Optics business continues to advance. ZeroFlap Optics combines optimized optical hardware, PILOT software, and switch-level SDK integration to continuously monitor link health and identify and mitigate possible link instability. The goal is to reduce cluster startup time and improve long-term network availability, both of which can provide significant financial benefits and end customer outcomes.
Production and shipment have already begun. We expect that during the 2027 fiscal year, 800G and 1.6T products will welcome more customers, including both hyperscale cloud service providers and NeoCloud. With DSP, PIC, and ZeroFlap Optics, we can now cover a larger portion of the optical link.
This changed the opportunities Credo faced. We sell components where our customers prefer; but more importantly, we also integrate these components with hardware, firmware, and software to provide a complete optical transceiver with unprecedented system-level reliability. Taken together, the momentum of the DSP, PIC, and ZeroFlap Optics businesses continues to push us firmly towards the goal of achieving more than $600 million in optical business revenue in fiscal 2027.
Now turn to Retimer. Our Retimer business also achieved record revenue in the first quarter. Growth was mainly driven by the use of Screaming Eagle Retimer and scale-up deployment of 100Gbps per channel, while our Blue Heron Retimer also began contributing revenue to the 200Gbps per channel market.
As PCIe Gen 6 adoption increases, we continue to see opportunities for Toucan Retimer; at the same time, Screaming Eagle and Blue Heron also have opportunities in the Ethernet and UALink fields. Scale-up architectures are rapidly evolving, and customers are making different choices in terms of protocols, topologies, and connectivity methods. Our ability to support multiple protocols allows us to successfully participate in these diverse architectures.
Now I'm going to discuss two important areas of emerging growth. We are also continuing to make progress in active LED cable (ALC) and OmniConnect gearbox solutions. Using miniature light-emitting devices, our ALC solutions combine the many reliability and power benefits of copper connections with transmission distances of up to 30 meters. Customer engagement continues to increase, and we plan to present ALC solutions at the OCP conference in October. We are still targeting initial revenue for the 2028 fiscal year.
At the same time, we've also seen strong customer engagement in OmniConnect's innovation. Our OmniConnect SerDes and Weaver gearbox solutions are designed to address fan-out issues in next-generation AI architectures as memory bandwidth and capacity requirements increase. This is particularly important for reasoning, as memory capacity, bandwidth, packaging, and cost are increasingly becoming important constraints at the architecture level. We believe that each GPU in the OmniConnect solution can represent thousands of dollars in Credo product value, and that revenue will begin in FY2028.
All in all, the first quarter was another strong quarter for Credo, and customer engagement across the company's business areas was still very strong. AEC continues to grow as we expand existing customer partnerships, add new customers, and advance to higher data rates. Our Retimer and optical DSP businesses also achieved record revenue. At the same time, the scope of our optical business is expanding.
We believe that as AI networks evolve to 1.6T and 3.2T solutions, and scale-up architectures drive increased use of near-package optics, a system-level approach will become increasingly important. AEC helped Credo achieve the scale it is today, and we continue to see that the business has room to grow in the future.
What's different today is that we're building optics as another major growth engine with a larger business foundation and as a mature player in the industry. Our current optical opportunities have expanded from DSP and silicon photonic PIC to complete ZeroFlap Optics and NPO solutions. As we solve a wider range of problems for our customers, our product value opportunities expand significantly.
The AEC continues to grow. The optical business is growing faster. Based on ongoing customer collaboration and volume across the entire product portfolio, we are still confident that we will achieve the very substantial growth we expect in fiscal year 2027.
The common theme behind all of these products remains reliability. As AI infrastructure continues to expand, our mission is to provide connectivity solutions that operate reliably, consume less power, provide network visibility, and keep expensive processors highly utilized. That's the focus of our work, and we're excited about what's ahead.
Daniel Fleming — Chief Financial Officer
I'll first review our first quarter results and then discuss our outlook for the second quarter of fiscal year 2027.
In the first quarter, we reported revenue of US$479 million, up 10% month-on-month, above the upper limit of our guideline range; on a year-on-year basis, revenue increased 115%. The first quarter once again set a record for revenue, thanks to the significant year-over-year increase in revenue from 4 US domestic customers, and also marked our 7th consecutive quarter of three-digit year-on-year revenue growth.
In the fourth quarter, our top four end customers all accounted for 10% or more of revenue. As a reminder, the customer structure will vary from quarter to quarter. We expect that 3 to 4 customers will still contribute more than 10% of revenue in the next few quarters and throughout the fiscal year. At the same time, we are also continuing to diversify our revenue base and expand our business among hyperscale cloud service providers, NeoCloud, and other customers.
Our team achieved a non-GAAP gross margin of 68% in the first quarter, at the midpoint of our guidance range. Total non-GAAP operating expenses for the first quarter were US$95.2 million, above the upper limit of our guidance range due to continued strong R&D investment; operating expenses increased 16% month-on-month. Non-GAAP operating profit for the first quarter was $230.6 million, while non-GAAP operating profit for the fourth quarter was $216.7 million. The non-GAAP operating margin for the quarter was 48.2%.
Even as we continue to invest heavily in R&D, the profit side of the company once again shows the significant operating leverage we are achieving. Non-GAAP net profit for the quarter was US$236.3 million, a record high, up 4% from non-GAAP net profit of US$226.7 million in the fourth quarter. Non-GAAP net profit in the first quarter more than doubled year-on-year, clearly reflecting the huge effects of rapid revenue growth, strong gross margin, and strict control of operating expenses. Non-GAAP net margin for the quarter was 49.3%.
Cash flow from operating activities in the first quarter was US$90.2 million, a decrease of US$92 million over the previous quarter, mainly due to changes in working capital. Capital expenditure for the quarter was $7.3 million, and free cash flow was $82.9 million. By the end of the quarter, we had cash and cash equivalents of $764.3 million, a decrease of $679 million from the fourth quarter, mainly due to cash expenses resulting from the acquisition of DustPhotonics. We still have sufficient capital to continue to invest in growth opportunities while maintaining large cash reserves. Inventory at the end of the first quarter was US$313.1 million, an increase of US$62.2 million over the previous quarter.
Now on to our performance guidelines. We currently expect revenue for the second quarter of fiscal year 2027 to be between $525 million and $535 million. We expect non-GAAP gross margin of between 67% and 69% for the second quarter. We expect non-GAAP operating expenses for the second quarter to be between $100 million and $105 million. We expect the diluted weighted average number of shares in the second quarter to be around 200 million shares. These expectations are based on the current tariff system, which is still constantly changing.
As the 2027 fiscal year continues to advance, we still expect an inflection point in the second half of the year, mainly driven by more than 600 million US dollars in optical business revenue. Among them, ZeroFlap Optics, Silicon Photonic PIC, and Optical DSP will all contribute more than 100 million US dollars, ultimately driving total annual revenue growth of more than 85% year over year.
We expect non-GAAP gross margin for FY2027 to be broadly in line with FY2026 levels. We expect non-GAAP operating expenses to increase by about 55% year over year, which is significantly lower than the revenue growth rate, as we continue to invest in R&D to support new product development and seize the huge growth opportunities ahead. As a result, we expect the non-GAAP net interest rate to be close to 50%.
II. Q&A session
1. Tore Svanberg - Stifel Research Division:
Congratulations to the company for setting a quarterly record. I'd like you to elaborate a little bit on the company's current position in the field of optics. You've once again confirmed your goal of over $600 million, but now you're also talking about NPO, and maybe even starting a system-level NPO. So when we consider this $600 million, how should we anticipate the business structure between the different components, whether in fiscal year 2027 or fiscal year 2028?
William Brennan:
I think we are very pleased with the ever-expanding product portfolio that is being brought to market. As we have indicated, our optical DSP business is doing quite well at the component level. The team from DustPhotonics brought a lot of business momentum to Credo.
In fact, after just a few months, this momentum has been further strengthened, and we're happy to announce that we've received design wins from two major customers for the next generation of products that will be rolled out in FY2028, and may even begin later this fiscal year. On the ZeroFlap Optics side, we continue to make progress. We're working with a number of customers, including hyperscale cloud service providers and NeoCloud. We are pleased with this year's developments.
When I was considering the overall optical opportunity, you mentioned CPX, and I think you've probably seen our previous press release, and we've joined this alliance. We plan to seek solutions for the growing scale-up market, and basically cover the entire market. In my opinion, CPX is an important development for the industry. This is a way to solve the scale-up density problem, and can achieve a density 10 times higher than scale-out. We're going to advance this market in the same way. When customers require it, we will sell components, and we will also promote system-level solutions and gradually introduce related content in the future.
However, you made an important point. I think the 2027 fiscal year is just a stepping stone on the path to the future development of our optics business. If we look at data from market forecasters, especially those focusing on the optical pluggable market, this segment of the market alone is expected to grow from 60 million units in 2026 to 175 million units in 2030. This is an amazing growth in this segment of the industry, and in addition, scaling up will bring additional growth on top of this. Therefore, we believe that with a broad portfolio of solutions, we will continue to achieve tremendous growth until 2030. Growth is driven not only by our optical product portfolio, but also by AEC and other copper connectivity solutions.
2. Quinn Bolton - Needham Research Division:
What I would really like to know more about is the importance of Credo joining the Open CPX Alliance and what kind of solutions you might be providing. Will more be available in kit form? Will you supply a complete optical engine? I think these should all be laser-based, not microLED, but can you elaborate on what kind of opportunities this new NPO scale-up field opens up for you?
William Brennan:
When we all think about scaling up opportunities, we consider changes in external dimensions. When we look at front-end networks and scale-outs, there is no real catalyst driving the market to change from a pluggable form factor. However, for scale-up networks, there is a fundamental requirement, which is the adoption of higher density form factors.
We've been talking, and the industry continues to discuss things like XPO, CPX, and how these technologies can be combined with NPOs, and ultimately to CPO. All of these solutions are designed to address the need for a 10x increase in density. And the way we enter the market is to some extent not limited by specific technology. Basically, we act according to customer requirements and the direction the customer wants to develop. As a result, I think we've been very technologically neutral when bringing our solutions to market. We will continue to do that.
So, of course, for all NPO solutions, including CPX, we will launch silicon photonic PIC first, but as we continue to add more features to our product portfolio, we will also consider providing a complete optical engine in the long term.
3. Tom O'Malley - Barclays Research Division:
When you look at predictions for the next year and the year ahead, much of the business volume in revenue is driven by ZF Optics' predictions. And this is unique. As you can imagine, you would use contract manufacturers to produce these products.
Meanwhile, we are hearing about supply issues and price increases in the foundry sector. Can you talk about the current challenges in scaling up production capacity? What makes you confident that you can meet the metrics you gave us during the last earnings call? What makes you believe that such huge revenue figures can be achieved in the short term? Also, please tell us about your current communication with customers, especially those customers whose media reports may be increasing expenses and ZF Optics' predictions have also been raised?
William Brennan:
Yes, you pointed out one very interesting thing. We're in a very dynamic market right now. You're right, from the supply side, the important point is that we've been investing a lot of time between 18 and 24 months ago because we know this is the direction we're headed.
As Dan's update just showed, our working capital is increasing. From a supply chain perspective, we're investing more. I am very confident in my ability to supply a growing number of products in the second half of the year and over the next few fiscal years.
The other half is driving demand. Over the past year, I've seen our progress in marketing and really building partnerships with our customers. Therefore, all of these factors must be combined.
But I want to emphasize once again that what we are really adopting is a long-term strategy. Of course, the market should and did pay a lot of attention to the very rapid volume we have set expectations for. But I think more importantly, from an overall perspective, this opportunity will take us to a completely different company size. So this is clearly a very important part of our growth strategy and company expansion.
4. Sean O'Loughlin - TD Cowen Research Division
Congratulations to the company for its strong results and current momentum. Maybe we can sort out some basic things. Dan, you mentioned that in the future, 3 to 4 customers will contribute more than 10% of revenue, and the company continues to diversify its customers. I'd like to ask specifically, can you tell me the specific data for these 3 to 4 customers this quarter? However, when we consider customer concentration, there is also the issue of platform concentration. Have you also diversified your product portfolio within these customers?
Daniel Fleming:
Let me start with the share of these customers who account for 10% or more of revenue. Our largest customer contributed one-third of our revenue, or 33%; our second-largest customer accounted for 28%; followed by 13% and 10%, respectively. The top three customers were basically the same as the previous quarter, or the same three customers, but the ranking order was slightly different. One thing to note is that our fourth customer, which accounts for 10% of revenue, is different from our fourth 10% customer in the previous quarter. This company also used to be 10% of our customers. As for the diversification of these customers' internal product lines, of course, we are constantly strengthening and expanding overall. It's not just AEC that these hyperscale cloud service providers are buying in large quantities.
5. Blayne Curtis - Jefferies Research Division:
Regarding the AEC business, I would like to ask about the timeline for 1.6T. At the same time, I would like to know how the design of the 800T is progressing compared to the 1.6G. You guys talk about NeoCloud all the time. I think some of your existing customers are moving to higher speeds. Can you explain how these designs are laid out and what the time points are?
William Brennan:
We expect the AEC product portfolio to begin expanding in roughly a similar time frame to the overall market and our ZF Optics products. I'd like to take a step back and introduce how we view the AEC market from an overall perspective.
There has been a lot of discussion in the industry about copper and optics and the trade-offs between the two. Our view of the AEC market is that it is part of the pluggable transceiver market. Therefore, when we talk about the optical transceiver market growing from 60 million to 175 million in the next 4 years, there is currently no separate division of the AEC market.
But the AEC market actually represents the 1 meter to 7 meter portion of it. So it's only natural that this segment will grow as well. We believe this will be a long-term opportunity. In the 800G field, we've already seen copper as an alternative to optics very clearly. Copper replaced laser-based optical transceivers due to the first stage of the network connection from the GPU to the first switch, the need for higher reliability, and secondarily the need for lower power consumption. So, at the end of the day, what we're seeing is a whole market for pluggable transceivers, from copper to laser-based optical connections.
Finally, we'll also talk about ALC as another option in the field of pluggable transceivers, and its transmission distance can reach 30 meters. So when it comes to what I think the 1.6T market pattern will look like, I'm very optimistic about the AEC opportunity.
As we showed at OFC 6 months ago, the industry has been discussing many of the next generation deployment options, and we've shown a very elegant solution, all with AEC connectivity. In this category, we will provide connections up to 6.5 meters long. As a result, we believe this market will begin to show some contributions in the second half of this fiscal year 2027, and then release on a larger scale in FY2028.
6. Joseph Cardoso - J.P. Morgan Chase:
I'd like to ask about the overall opportunity. Until now, most opportunities have been focused on training, but now we are also seeing more and more developments in the field of reasoning. So I wanted to know how is Credo positioned itself in dealing with inference opportunities? In particular, which part of your product portfolio stands out the most? Also, how big are the overall chances compared to training?
William Brennan:
I'd like to talk about the OmniConnect solution we're currently bringing to market, and related products we expect to contribute revenue in the 2028 fiscal year. Specifically in the field of reasoning, as these solutions continue to gain momentum, a common problem is memory fan-out. Memory fan-outs are limiting bandwidth and total memory deployment capacity, both of which are critical for high-performance inference. Therefore, for our OmniConnect solution, this product has two main parts.
In the first part, we authorized a highly optimized high-speed SerDes with a very small form factor, very low power consumption, and transmission distances of up to 10 inches. Therefore, it actually solves the fan-out problem in the XPU front-end, and also solves the distance problem that can be achieved between the GPU and the memory.
The second part is the gearbox we are developing. The first GearBox we're developing is a solution we call Weaver. It is a gearbox connected to an XPU embedded serdes. It has a transmission distance of up to 10 inches and is connected to the exact same SerDes. Subsequently, it will become an LPDDR interface. It should be noted that our first product will use LPDDR5, and the second LPDDR6 product will also be launched.
So, in a sense, this architecture is future-proof, because as the memory market moves from 5th generation to 6th generation, our XPU partners don't need to tape-out again, or redesign, just replace the gearbox.
Our first customer, Positron, is currently doing a fantastic job. As we look at the memory limits and the opportunity to increase the memory capacity to 2TB — this is their first announced product configuration — for cutting-edge models, if all data can be put into memory, performance will explode. As a result, they are redefining the competitive landscape in this field. Even from a bandwidth perspective, we have a roadmap to achieve the same bandwidth as HBM5 in the future.
At the same time, we can also expand memory capacity and eliminate reliability issues caused by packaging XPU and HBM in the same package. Specifically in the field of reasoning, we're really excited because it's going to change the rules of the game for many cutting-edge model applications.
7. Vivek Arya - Bank of America:
I want to go back to the AEC growth issue in the second half of the year and beyond. If I were to use your 85% growth forecast for the full year, that would mean an incremental increase of around $500 million in the second half of the year. But if the optical business will account for the majority, then the $600 million figure seems to mean that you are making relatively conservative assumptions about AEC growth. I believe I may have misunderstood the share of the optical business in the first half of the year. Therefore, I would like you to further explain the segmentation of the optics and AEC business in the first half of the year and the second half of the year, and what this means for AEC growth in the second half of the year. Also, how should we model AEC growth in the long term?
William Brennan:
When we look at growth throughout the fiscal year and every product we're bringing to market, we're seeing growth across the board. Of course, for our optical solutions, this is the first year we actually started scaling. Based on the products we're bringing to market, it's quite reasonable that we'll be able to grow very fast. I think AEC will continue to grow.
But you have to consider what kind of foundation we grew from. AEC has been driving growth for the past two fiscal years, more than doubling from fiscal year 2024 to fiscal year 2025, and more than tripling from fiscal year 2025 to fiscal year 2026. So when we look at AEC opportunities, we still see expanding opportunities.
But in a sense, our overall growth rate will clearly be slower compared to the fact that we are entering such a huge market through our optics business. So it makes sense that the optical business is growing faster because we started growing from a smaller base.
Over the long term, I think you'll see AEC continue to grow. As the company continues to expand, I think you'll see a very good balance between copper and optical solutions.
8. Sebastien Naji - William Blair Research Department:
I'd like to ask about PILOT and some telemetry data your solution is collecting. As the number of products deployed continues to grow, are you accumulating enough link-level telemetry data that it itself begins to be a competitive advantage? For example, can this data help you identify failure patterns or optimize future DSP designs? Is this further deepening your moat? Can you share your views?
William Brennan:
The PILOT software platform is a very important part of the overall ZF Optics' product portfolio. As we look at what we're doing, we basically have to start with a custom DSP design. This DSP design enables telemetry on every link between the XPU and the switch. When I say each link, there are actually 6 in total. There are 3 lines in one direction and 3 in the other direction. This is completely different from telemetry currently discussed in the industry before.
As a result, we can continuously sense very rich telemetry data in real time. This data can reflect link stability indicators, and can even go down to the SerDes level, because SerDes is of course the core of everything in our platform. However, we also measure eye height (eye height) and SNR post-histograms (SNR post-histograms) in real time, and we can even obtain very rich telemetry data at this level. As a result, as we continue to sense, we can see that link integrity is declining.
You can think of it as a green light mechanism that currently exists in the industry — a green light is displayed when the connection is normal; if a fault occurs, a red light is displayed. What we have added is a function similar to a car's “check engine” indicator, that is, a yellow light. It says: OK, you've identified some issues.
The next step is to take mitigation measures. PILOT can support different processing methods. You can make decisions on a transceiver by transceiver basis: for transceivers that seem likely to cause link jitter, the corresponding GPUs can be removed from the cluster in an orderly manner. We have other customers that are considering a more centralized approach at the network level.
From a reliability perspective, the end result is the same. But in this case, both opportunities allow you to gather large amounts of data. Because when you shut down a link, you can record exactly what happened to that transceiver and what happened to the entire network at that time. So you're right, the data sets we've begun collecting and will continue to accumulate in the future will drive us to develop better solutions. It will also drive the next generation of solutions to be further optimized to better handle the types of faults we detect in advance.
I'd also like to say that we're doing some other very interesting things related to our communication with our customers. A customer asked us if you could detect even the slightest ESD damage on the transceiver, as this damage could be a potential hazard. As a result, we found a way to detect even the slightest ESD damage. This damage is not the kind of damage that immediately causes the transceiver to fail, but rather the kind of damage that eventually causes the connection to fail over time. So now when customers start lighting up the rack, they can immediately determine if a transceiver has been damaged due to improper operation and determine if it needs to be replaced.
Other issues we've detected include dust on fiber infrastructure. This is very important because even the smallest speck of dust can cause light to reflect in the other direction, causing multipath interference. So what we're doing now is far more than that.
PILOT allows us to iterate within the customer network. This is a real game changer because it gives customers the tools to proactively identify issues and take mitigation actions. Also, there are actually two goals here. The first goal is to achieve the fastest time to revenue. In other words, let the cluster launch in 5 to 6 days instead of 6 to 8 weeks.
I think we've seen some contracts announced on the market. You can calculate that when those very expensive devices could generate revenue, but because the cluster takes weeks instead of days, and the equipment is idle, every month of delay could mean an opportunity cost of hundreds of millions of dollars or even 1 billion dollars.
Another important goal is uptime after deployment, and aim for 99% uptime or more. This can provide end customers with better network architectures and better products. So I think this is a huge differentiating advantage at the cluster level. But yes, PILOT is critical to achieving this goal. It's a combination of custom hardware, but what really matters is the interface between it and the network.
9. Karl Ackerman - BNP Paribas:
As we consider the independent DSP and PIC business in your $600 million optical business and the growth of this business over time, is the customer structure increasingly shifting to direct sales to hyperscale cloud service providers? These hyperscale cloud service providers are customizing their own optical transceivers. So how can you gain better customer visibility and greater customer stickiness as this happens?
William Brennan:
As we look at the optical components business, this is a very important part of our short and long term business. Looking at the ultimate direction, this will be our path to the market for pluggable optical transceivers based on mainstream standards. Having optical DSP and PIC, and being able to provide system-level solutions within component products will help our customers achieve what we believe is the most competitive combination of system performance, power consumption, and yield.
Many times, our module customers work directly with hyperscale cloud service providers. Hyperscale cloud service providers will actually actively specify the specific elements they want to be integrated into the module provided by the module customer. Therefore, whether in our business of selling components to module customers or in the ZF Optics module business that we manufacture ourselves, hyperscale cloud service providers play an important role.
In the long run, I think this balance will continue to exist. In other words, our component sales and module sales will complement each other. A broad segment of the market will be satisfied through components, and the part of the market that is very specific and really focused on reliability will form a new transceiver product.
10. Vijay Rakesh - Mizuho Securities:
You mentioned — from a broader perspective — that fiscal year 2027 was a stepping stone, and you've already achieved around 85% year-on-year growth. Looking ahead to the 2028 fiscal year, can you give us some thoughts? Obviously, the AEC is likely to grow by around 50% this year. You have ZF Optics amplifiers and SiPho. You mentioned that ZF Optics and SiPho will reach around $600 million in fiscal year 2027, which is equivalent to an annualized operating rate of over $1 billion. You also mentioned Active LED Cables. Can you give us some thoughts on FY2028 as all four business areas begin to expand?
William Brennan:
Discussions about 2028, 2029, and FY2030 are very active topics within the Credo leadership team.
I'll start by talking about ALC's positioning within our product portfolio. I think our first ALC product to market uses microLED technology. The promise of this technology is to achieve the same reliability and the same energy efficiency as AEC at the core technical level. The key difference is that we will extend the transmission distance to 30 meters. As a result, ALC will be our third differentiated pluggable transceiver solution. We first created this product category through AEC and then created a new product category again with ZF Optics. ALC will be the third leg of this “three-legged stool.” At the end of the day, when I look at the pluggable transceiver market, I think this is just the first step in ALC and miniature light-emitting device technology.
The next natural direction of development is to apply it to scale-up opportunities. Because in the field of scale-up, this is another technical alternative. But its core promise is that we can resolve some of the issues that previously prevented this market from taking off, particularly those relating to reliability, availability, and maintainability.
Therefore, we see ALC as a massive multi-billion dollar opportunity, and the subsequent scale-up is an equally huge opportunity, and it is highly complementary to the technology portfolio we have already brought to market.
So, to answer your question more specifically, my thoughts on the future are: We're trying to get ourselves into multiple multi-billion dollar TAM (Total Addressable Market) opportunities. You just need to analyze our product portfolio in the pluggable sector. As you'll see, the scale of opportunities we're entering today actually amounts to tens of billions of dollars, and as ALC becomes part of the product portfolio, these opportunities will be further unlocked on the 2028 fiscal year timeline. So we're trying to position ourselves in a position to serve a very large market.
The growth of our company will depend on our ability to successfully execute and implement with every customer. I think we're very optimistic about this opportunity. Also, this doesn't count the huge opportunities the entire industry has in front of scaling up.
Even OmniConnect, we've made it clear in the past that it's also a multi-billion dollar opportunity every year. So when we think about fiscal year 2028, you're right, we expect to see another huge increase, and we think this growth will continue for many years to come.
11. Mark Lipacis—Evercore ISI:
I think for you, the optical business is growing faster, and it's a newer market. Can you compare how the way you are developing the optical market now is different from the way you first developed the AEC market? I think there should be some difference because you actually created the AEC market. Also, as the scale of the optical business continues to expand, what impact does this have on the business model? Where can you achieve collaboration between the two businesses in terms of development, supply chain, and customer collaboration? And in what areas do new capabilities need to be built?
William Brennan:
There is indeed a very interesting contrast between the way we initially promoted the AEC business and some of the leverage that can now be achieved with ZF Optics.
On the AEC side, we initially conceived this product as an extension of copper connections, basically to solve some of the problems customers face when switching to higher speeds. To our surprise, customers took the initiative to reach out to us to discuss some very interesting innovations, such as telemetry, and other system-related rack-level innovations.
As a result, we really opened the door to functional innovation to our customer base. Our first customer, Microsoft, turned to AEC because of the features we provided. We have developed a solution that is smart enough to detect when the ToR port is failing or is about to fail, and then switch the data to the backup ToR in a non-destructive manner.
This is a very, very smart solution in the form of a cable. Over time, this momentum has gradually developed. And this has been going on for years. As speed continues to increase, it is clear that DACs cannot meet the requirements in terms of signal integrity, and in terms of external dimensions, copper wires also need to become thicker and thicker.
As a result, many customers are looking for alternatives in terms of form factor and signal integrity. Once again, when we look at today's solutions, many of them have really innovative features.
As a result, the creation and growth of this product category took several years. In developing solutions, providing complete solutions, and assuming overall responsibility, we have also established a highly differentiated approach to market entry, particularly in terms of certification. Our solutions are rigorously tested, sometimes far beyond what our customers usually require. This includes having the customer's switch, the customer's NIC run at actual speed, and stress-testing the entire link. Not just testing our AEC, but the entire link from link to switch. The goal is to strengthen the entire solution, find weak points in the link, and then further enhance it through firmware modifications.
So this is clearly the set of experiences we are now using to bring ZF Optics to market. What is different about ZF Optics, however, is that AI has appeared, and AI networks are fundamentally different from front-end networks. In front-end networks, there is often built-in redundancy between different network tiers.
As a result, we have had link flapping issues for a long time. But they didn't become a major issue because the overall impact on the network was minimal. However, now we have AI clusters with tens or even hundreds of thousands of interdependent links. If some of these links start to flicker, it could affect the entire cluster. Customers have even mentioned that GPU utilization could drop by more than 10% or even close to 20% as a result.
Therefore, with the advent of this new application, we have a perfect opportunity to follow the AEC style of play and do very interesting and innovative work in a field where there hasn't been much innovation in many years.
We work closely with our customers to plan a path for improving network reliability at a system level. That's the real difference. Of course, this also takes time to develop hardware and software solutions. But we see that this market is growing much faster than AEC. Because what we're addressing is a pain point that already exists and clearly already exists. So every time we talk to the customer's internal technical network team, they welcome it.
12. Suji DeSilva - OTH Capital:
Congratulations to the company for making progress and achieving strong profit margins. You have provided a great deal of information about the optical business, and also explained in depth how you entered this market. I'd like to take this opportunity to revisit the competitive landscape. How do you think competitors will enter the markets you are entering? Are your methods different? Can they adopt a feature-driven approach like yours? Or what other key factors have enabled you to bridge the gap with your peers? Because a lot of people are talking about the same opportunities right now.
William Brennan:
I think there's one case worth discussing here, which is our market entry strategy and the fact that we have the entire technology stack. When we look at opportunities at the transceiver level only, we started with SerDes, then extended to DSP, and now further to PIC. This vertical integration capability is a huge advantage in providing the best system solutions, and also gives us an advantage in terms of COGS (cost of sale). Then, you also need to consider how to enter the market with differentiated solutions.
The challenge is, how much are these features really worth? Therefore, in my opinion, compared to standardized solutions on the market, we expect to have an advantage at the COGS level and also at the ASP (average sales price) level.
I really think that in the long run, we are now a leader in innovation in the field of optics. The market needs this kind of innovation. We are also very open to working with standards organizations to promote standardization of some of the technologies we are developing. The challenge here is how we can innovate faster than other competitors and the market as a whole. I think we have confidence in that. Because we have the entire technology stack, this will lead to the same results as our past successes in the AEC field.
13. Christopher Rolland - Susquehanna:
I'd like to focus on the DSP and PIC integration you've mentioned many times. First, can you tell us more about your two Dust design wins? Do they include DSP integration? Also, can you remind us how economically beneficial this segment of the market is? What does that mean to you?
William Brennan:
We are in a highly competitive component market. As a result, according to proper agreements, we can't specifically talk about who the specific large customers we are working with are. What I can say is that the first two big design wins we're talking about don't include DSPs. Therefore, when we actually promote joint marketing between DSP and PIC from a long-term perspective, there is still room for improvement in this part.
Currently, the Dust team has done an excellent job in deep collaboration with large customers in the industry. It's amazing how such a small team can achieve so much customer collaboration momentum. I think this is proof that their technology solutions are truly at the leading level in the market.
In the long run, I think we'll have a great opportunity to look at these two components and really bring a lot of advantages to our customer base. So I think this is a very good development for our overall product portfolio and is very promising for us in the long run.