Credo Technology Group Holding (CRDO) Q1 2027 earnings call transcript

I would now like to turn the conference over to Dan O'Neill, Treasurer and VP of Investor Relations. Please go ahead, sir. Good afternoon. Thank you all for joining our first quarter of fiscal 2027 earnings call. Today I am joined by Bill Brennan, Credo's Chief Executive Officer in Dan Fleming, Credo's Chief Financial Officer. During this call, we will make certain forward-looking statements. These forward-looking statements are subject to risks and uncertainty discussed in detail in our documents filed at the SEC. These documents can be found in the Investor Relations portion 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 his 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 statement.

Given these risks uncertainties and assumptions, the forward-looking events discussed during this call may not occur. In actual results could differ materially and adversely from those anticipated, implied, or inferred. 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 changes in the company's expectations, order actual results, except as required by law. Also, during this call, we will refer to certain non-GAP financial measures, which we consider to be important measures of the company's performance.

These non-GAP financial measures are provided in addition to, and not as the substitute for, or superior to, financial performance prepared in accordance with the US GAP. A discussion of why we use non-GAP financial measures and reconciliations between our GAP and non-GAP financial measures is available in the earnings release we see today, which can be accessed using the investor relations portion of the website.

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

Thanks, Dan. And thank you everyone for joining our first quarter fiscal 27 earnings call. The first quarter was another strong quarter for Credo. Revenue reached a record $479 million, increasing 10% sequentially and more than doubling year over year. Non-GAP gross margin was 68%, and non-GAP net income exceeded $236 million, up 140% year over year. Credo has been growing at a pace that very few semiconductor companies have achieved, all while expanding profitability. At the heart of this growth is seven consecutive quarters of triple digit year over year growth. We continue to see outsized growth in fiscal 27, with our optical business growing at the fastest pace. AI infrastructure investment continues to grow rapidly. Cluster sizes are increasing, data rates are moving higher, and connectivity requirements are becoming more challenging.

As the systems scale, connectivity is about much more than bandwidth, reliability, power efficiency, signal integrity, telemetry, and serviceability all matter. We also believe AI infrastructure will become increasingly heterogeneous. There will not be one architecture, one protocol, or one physical medium that is chosen for every connection. Future AI systems will combine optical and copper interconnects across different reaches, protocols, and topologies, with customers choosing the right technology to optimize their architecture. This is where Credo shines. While network reliability remains Credo's north star, our ability to innovate, execute, qualify, and deploy across the wide range of customers' needs is core to our differentiation. Our focus is on helping customers bring clusters up faster, maximize processor utilization, and maintain reliable operation at scale.

Our portfolio now spans connectivity from millimeters to kilometers with solutions across optics and copper. Let me walk through each of these areas in more detail. Starting with AECs, AECs remain our largest business and continues to grow. We now have deep relationships with five hyperscalers, and our engagement with neo-cloud customers continues to expand. Within our existing customers, we continue to see increased AEC penetration as deployment scale. Higher data rates provide another growth vector with a transition to 200 gig per lane, 1.6 T ports ahead. AECs have always been a system-level product for Credo. We provide the complete solution and optimize the silicon, firmware, manufacturing, test, and system qualification together. Our system-level approach has been fundamental to our differentiation since we created the category.

As AI clusters get larger, the value proposition remains straightforward, high reliability, and low power for short reach connectivity, or both become increasingly important at scale. We continue to see a healthy growth trajectory for AECs, driven by deeper penetration with existing and new customers, and with increasing bandwidth in next generation clusters.

Now, turning to optics. Our optical business is progressing very well and includes optical DSPs, silicon photonics picks, and zero flap optical transceivers. Our optical DSP business delivered record revenue in Q1. Revenue included deployments across our 50 gig and 100 gig per lane solutions. We see a long tail for 800 gig ports, even as we begin the transition to 1.6 T solutions.

At 200 gig per lane, customer engagement with our 1.6 T DSP is strong across both fully re-timed and LRO solutions. Our first 1.6 T DSP revenue remains on track for later this fiscal year. During the quarter, we also recognized our first silicon photonics pick revenue following the dust photonics acquisition. Our initial wins are an 800 gig and 1.6 T optical transceivers.

And we expect these products to ramp throughout the year. More importantly, silicon photonics picks at another important technology to our optical platform. We now optimize the DSP and pick together and combine them with our firmware, telemetry, and pilot software. This level of integration creates opportunities to improve reliability, power, signal integrity, and diagnostics. It also positions us well as scale up architectures move toward near-package optics. We're seeing increasing customer activity around NPO for scale up networks with confirmed design wins expected to begin ramping in our fiscal 28.

As part of the OpenCPX MSA consortium, we'll bring many of the advantages of today's pluggable ecosystem, including telemetry, interoperability, and serviceability. Our opportunity here includes both optical components and complete system level solutions. This is an important evolution for Credo. We've historically been very successful solving connectivity problems at the semiconductor and the AEC system level. We're now applying that same approach to optics. Our ZeroFlap optics business continues to progress.

ZeroFlap optics combines optimized optical hardware, pilot software, and switch level SDK integration to continuously monitor link health and identify and mitigate when link instabilities become likely. The objective is to improve cluster bring up time and long-term network availability, both of which delivers significant financial advantages and customer outcomes. Production shipments are underway, and we expect additional customer ramps during fiscal 27 across both 800 gig and 1.6T with both hyperscalers and neoplads. With DSPs, picks, and ZeroFlap optics, we now address much more of the optical link. That changes the opportunity for Credo. We sell components where that's preferred by customers, but more importantly, we also integrate those components with hardware, firmware, and software to deliver a complete optical transceiver with unprecedented system level reliability.

Taken together, the momentum across DSPs, picks, and ZeroFlap optics keeps us firmly on track to deliver more than $600 million of optical revenue in fiscal 27. Now, turning to retimers. Our retimer business also delivered record revenue in Q1. Growth was primarily driven by scale up deployments using our screaming eagle retimer at 100 gig per lane, and with our blue heron retimer beginning to contribute at 200 gig per lane.

We continue to see opportunities for the two can retimer as PCIe Gen 6 adoption increases, and for screaming eagle and blue heron across ethernet and UA link. Scale up architectures are developing quickly with customers making different choices around protocols, topology, and connectivity. Our ability to support multiple protocols allows us to successfully participate across these architectures.

Now, I'll discuss two important emerging growth areas. We also continue to make progress with both active LED cables or ALCs and our omniconect gearbox solutions. Our ALC solutions use micro emitters to combine many of the reliability and power advantages of copper with reach of up to 30 meters. Customer engagement continues to increase, and we plan to demonstrate ALC solutions at OCP in October. We remain on target for initial revenue in fiscal 28. We're also seeing strong engagement around our omniconect innovation.

Our omniconect 30s and weaver gearboxes address the fan out issues that come with increasing memory bandwidth and capacity requirements next generation AI architectures. This is especially relevant for inference for memory capacity bandwidth, packaging, and cost are becoming increasingly important architectural constraints. We believe omniconect solutions can represent thousands of dollars of credo content per GPU with revenue beginning in fiscal 28.

In conclusion, Q1 was another strong quarter for credo, and customer engagement across the business remains very strong. AECs continue to grow as we expand with existing customers, add new customers, and move to higher data rates. Our retimer and optical DSP businesses also deliver record revenue. At the same time, the scope of our optical business is expanding. We believe that the system level approach will become increasingly important as AI networks move to 1.6 T and 3.2 T solutions and scale up architectures drive greater use of their package optics. AECs help take credo to the scale we've achieved today, and we continue to see growth ahead for that business. What's different today is that we're adding optics as another major growth engine from a much larger base and as an established player in the industry.

Our optical opportunity now extends from DSPs and silicon photonic picks to complete zero flap optics and NPO solutions. Our content opportunity expands significantly as we saw the broader set of challenges for our customers. AECs continue to grow. Optics is growing faster, and based on the customer engagements and ramps underway across the portfolio, we remain confident in the outsized growth we expect to deliver in fiscal 27. The common threat across all these products remains reliability. As AI infrastructure scales, our job is to provide connectivity that works reliably, uses less power, provides visibility into the network, and keeps expensive processors operating at high utilization. That's what we're focused on, and we're very excited about lies ahead.

With that, I'll turn the call over to Dan.

Thank you, Bill, and good afternoon. I will first review our Q1 results and then discuss our outlook for Q2 of fiscal year 27. In Q1, we reported revenue of $479 million, up 10% sequentially, and above the high end of our guidance range. Year over year, revenue grew 115%. Q1 marks another revenue record, driven by substantial year over year growth across four domestic customers, and marks our seventh consecutive quarter of triple digit revenue growth year over year.

Our top four end customers each came in at or greater than 10% of revenue in Q4. As a reminder, customer mix will vary from quarter to quarter. We continue to expect that three to four customers will be greater than 10% of revenue in the coming quarters and fiscal year, and we continue to make progress in diversifying our revenue base across hyperscalers, neoclouds, and other customers.

Our team delivered Q1 non-gap gross margin of 68% at the midpoint of our guidance range. Total non-gap operating expenses in the first quarter were $95.2 million, above the high end of our guidance range due to our strong R&D investment, and up 16% sequentially. Our non-gap operating income was $230.6 million in Q1, compared to non-gap operating income of $216.7 million in Q4.

Our non-gap operating margin was 48.2% in the quarter. Our bottom line once again demonstrated the substantial leverage we were delivering in the business, even with our continued heavy investment in R&D. Our non-gap net income was $236.3 million in the quarter, a record high, and a 4% sequential increase compared to non-gap net income of $226.7 million in Q4.

Our Q1 non-gap net income more than doubled year over year, clearly demonstrating the magnitude of our top line growth, strong gross margins, and disciplined approach to managing operating expenses. Our non-gap net margin was 49.3% in the quarter. Cash flow from operations in the first quarter was $90.2 million, down $92.0 million sequentially due primarily to changes in working capital.

CapEx was $7.3 million in the quarter, and free cash flow was $82.9 million. We ended the quarter with cash and equivalence of $764.3 million, a decrease of $679 million from the fourth quarter due primarily to the cash outlay for our acquisition of dust photonics. We remain well-capitalized to continue investing in our growth opportunities while maintaining a substantial cash buffer. Our Q1 ending inventory was $313.1 million, up $62.2 million sequentially.

Now, turning to our guidance, we currently expect revenue in Q2 of fiscal 27 to be between $525 million and $535 million. We expect Q2 non-gap gross margin to be within a range of 67% to 69%. We expect Q2 non-gap operating expenses to be between $100 million and $105 million.

And we expect Q2 diluted weighted average share count to be approximately 200 million shares. These expectations are based on the current tariff regime, which remains fluid. As we move forward through fiscal year 27, we continue to expect an inflection into second half driven by more than $600 million in optical revenue with zero flap optics, silicon photonics, picks, and optical DSPs, each contributing more than $100 million, resulting in more than 85% year over year total revenue growth for the full year. We expect non-gap gross margin in fiscal year 27 to be broadly consistent with fiscal year 26 levels. We expect non-gap operating expenses to increase approximately 55% year over year, well below our revenue growth rate as we continue to invest in R&D to support the new product development and address the significant growth opportunities ahead.

As a result, we expect our non-gap net margin to be in the vicinity of 50%. And with that, I will open it up for questions.

Sure, I think we feel great about the broadening portfolio that we're bringing to market. As we've indicated, you know, the optical DSP business trust is going quite well at a component level. The team that came in to Credo from dust brought a lot of momentum. And in fact, after, you know, just a few months, the momentum is picked up and we're happy to be able to say that we've got design wins with two major players for next generation ramps that will occur in fiscal 28 and maybe, you know, starting in in the late part of this fiscal year. As the optics, we continue to make progress. We're engaged with multiple customers, both hyperscalers and new clouds.

And we feel good about the way that the year is shaping up. I think as I think about the overall optical opportunity, and you did mention CPX, and I think you may have all seen the press release earlier that we have joined the consortium. And we plan on pursuing solutions, you know, for the scale up market that's developing.

Really, across the board, CPX is, I think, an important development in the industry. That's one way of solving the challenge of going to 10 times more density than what you're seeing in scale out. And we'll pursue this market the same way. We'll pursue component sales where that makes sense with customers and we'll also pursue system level solutions that we'll talk about over time.

But you bring it up an important point. Fiscal 27, I think, is just a stepping stone for where we're going with our optical business. And if we think about the market forecasters, specifically, you know, the ones that are focused on the optical transceiver pluggable market, just that piece alone is expected to grow from 60 million units in 26 to 175 million units by 2030.

This is amazing growth for this portion of the industry. And then add growth on top of that for what happens in scale up. And so we think that with our broad portfolio of solutions that we're going to experience continued outsized growth through the 2030 timeframe. And it's going to, you know, have contributions from not just our optical portfolio, but also AECs, as well as our other copper solutions. So I probably gave you a little more color than you asked in that question. So hopefully that gives you what you were looking for.

As, you know, we all think about the scale up opportunity. You know, we're thinking about changing the form factors. And when we look at front end and we look at scale out, there's no real catalyst to change from the pluggable form factor. But for scale up networks, there's, you know, a fundamental need to have more dense form factors. And we've talked, you know, there's been an ongoing industry conversation about things like XPO, CPX, how that fits in with NPO, and then ultimately CPO. All of these solutions address the need for a 10x density improvement. And the way that we're approaching the market is somewhat agnostic. We will basically look at what our customers are asking us and what our customers are driving towards.

And so we've, I think, done a very good job of being agnostic as it relates to the solutions that we're bringing to market. We're going to continue to do that. And so, of course, with all of the NPO solutions, including CPX, we'll lead with the Silicon photonics pick, but we'll also look at doing the complete optical engine long term as we add more functionality to our portfolio.

Yeah, you point out something that's lots of fun. We're in a very dynamic market right now, and you're right from the supply side. It has been important that we have spent so much time even going back 18 to 24 months ago, and we knew that this was a direction that we're heading. And so you can, you know, you can see from Dan's update that working capital is increasing, we're leaning in from a supply chain standpoint. I feel great about our ability to supply, to supply increasing volumes in our second half, and then throughout the next couple of fiscal years.

The other half of that is driving demand. And so I see that, you know, over the past year we've been very successful in marketing and really engaging with customers. And so all of this has got to come together. But again, I'll point out that we're really playing the long game here. And of course, there should be, and there is a lot of emphasis on the very fast ramp that, you know, we've set expectations on. But I think more importantly, looking at the big picture, this involves an opportunity that takes us to a different scale as a company. And so this is very, you know, very much an important part of our growth strategy and our scale as a company.

Sure. Let me start with just the percentages of those 10% plus customers. So our largest customer was a third of our revenue at 33%, second largest 28, and then followed by 13 and 10. Those top three customers were similar to or the same three as in the prior quarter in a slightly different order. One thing to note, our fourth 10% customer, right at 10%, was different from our 10%, you know, fourth customer last quarter.

They have been a 10% customer in the past. Now, in terms of diversification of product line within those for sure, there is a strengthening and broadening across the board. It isn't just AECs that these hyperscalers are consuming of others.

Sure. We expect that our AEC portfolio will ramp in a similar kind of time frame as the rest of the market. And for that matter, RZF optics products. I want to, you know, step back and give some perspective on the way we view the AEC market.

There's been a lot of conversation about copper and optical and just the trade off there. And the way that we view the AEC market, it is part of the pluggable transceiver market. So when we talk about forecasts for specifically optical transceivers going from 60 to 175 over the next four years, it's that there's not really a breakout for the AEC market, but the AEC market really represents the one meter to seven meter segment of that market. And so it just makes sense that that segment is going to grow as well. And we see that really being long-term. We see that in a big way, in the 800 gig category, what we saw was copper replacing optical.

And it was replacing optical laser-based optical transceivers because of the need for higher reliability and secondarily lower power as it relates to that first connection and the network from GPUs to that first switch. And so the bottom line is that we see the pluggable transceiver market for all of the pluggable transceivers from copper to laser-based optical. And ultimately, we'll talk about AECs being another pluggable transceiver option within the spectrum and setup looks like for 1.6T. We're quite bullish about the AEC opportunity. And as we showed going back six months ago at OFC, many of the next generation deployments that have been discussed in the industry, we showed very elegant solutions all connected with AECs.

And we'll deliver up to six and a half meters in that category. So we think it takes shape towards the end of, towards, you know, there'll be some contribution in our second half fiscal 27. And then fiscal 28, it comes in in a much bigger way.

Appreciate that question. I'd like to maybe talk a little bit about our omni-connect solutions that we are bringing to market now and that we expect to contribute revenue in our fiscal 28. It's specifically related to inference. As these solutions continue to gain momentum, one universal issue is around memory fan out. The memory fan out issue is limiting bandwidth and also total memory deployment of memory capacity, both of which are critical to achieving high performance inference. And so with our omni-connect solutions, there's kind of a two-piece story to the offering. First thing is we license a highly optimized high speed 30s that's got a very small form factor, very low power, and that's got reach of up to 10 inches.

And so effectively addressing the fan out issue that exists on the XPU beachfront as well as the distance that you can get, the distance you could achieve between the GPU and memory. So the second piece is the gearboxes that we're developing. The first gearbox that we're developing is a solution that we call weaver. And it is a gearbox that interfaces between that embedded 30s on the XPU. So it's up to 10 inches reach and interfaces with that same exact 30s. And then it becomes an LPDDR interface. And it's important to note that our first product will be LPDDR5. And there will be a second product we do for LPDDR6. And so in a sense, this architecture is future enabled because when the memory market shifts from 5 to 6, there isn't going to be a need for our XPU partners to do another tape out, another design, just simply change the gearbox.

And so our first customer positron is doing really, really great things. When we look at memory limitations and the opportunity that if you can increase memory capacity up to two terabytes, which is their first announced product, just the performance with Frontier model is basically you can fit everything in memory and your performance just goes off the charts. And so they're redefining what is competitive in that space. And if we look even from a bandwidth standpoint, we've got a roadmap that will achieve very comparable bandwidth to even HBM5.

And that's while expanding on memory size and eliminating the reliability issues of packaging, XPUs, and HBM in the same package. But related specifically to inference, we're really excited about how this is really going to change the game for many Frontier model applications.

Sure. I think that as we look at the entire year and we look at the growth across each one of the products that we're bringing to market, we see growth across the board. Now, of course, with our optical solutions, this is really the first year that we're ramping and it makes a lot of sense based on what we're bringing to market that we'll be able to achieve a really fast-paced growth. I think AECs will continue to grow. You've got to consider where we've grown from. AECs have driven growth over the last two fiscal years more than doubling from 24 to 25 and they're more than tripling from 25 to 26. And so I think that as we look at our opportunities in AEC, we continue to see expanding opportunities.

But in a sense, I think we're looking at clearly a slower growth overall compared to the fact that we're growing into such a large market with optics. And so it makes sense that optics will grow faster because we're growing from a smaller base. But long term, I think you'll see AECs grow. And as we continue to scale as a company, I think you'll see balance, a really nice balance between copper solutions and optical.

Sure. The pilot software platform is a really important part of the total ZF optics offering. And when we look at what we've done, we've basically had to start with a custom DSP design, one that would enable telemetry to be lit up on every link between an XPU and a switch. And when I say every link, there's really six in total.

There's three in one direction and three in the other. And this is completely different than the telemetry that has been discussed up to this point in the industry. And so what we're able to sense on a real-time continuous basis is really rich telemetry data that looks on indicators of link stability, even going down to the 30s level, because that, of course, is core to our platform on everything we do. But we're looking at a real-time measurement of i-height and SNR post-FEC histograms, really rich telemetry data, even down to that level.

And so when we are sensing this continuously, we can sense when the link integrity is decreasing. And so you can think of it as right now in the industry what exists is a green light when you've got a connection and a red light when there's a failure. What we're adding is like a check engine, a yellow light that says, okay, you've identified something. And then the mitigation piece of it is about acting. And so there's different approaches that are enabled by pilot. You can make the decision on a transceiver by transceiver basis to take that transceiver that looks likely to have a link flap and in an orderly way take that GPU out of the cluster.

We've got other customers that are looking at a network level, more of a centralized approach. And the ultimate outcome is the same from a reliability standpoint. But in this case, both opportunities allow you to collect a tremendous amount of data because as you're taking a link down, you've got a snapshot of exactly what was happening within that transceiver and what's happening within the network. And so you're right that the data set that we're starting to gather and that we will gather over time will lead to better solutions. It will lead to next generation solutions being optimized even more so for the types of failures that we're detecting early. I will say we're doing other things that are really interesting.

And that's related to conversations with customers. We had a damage on a transceiver because that will become a latent defect. And so we figured out a way to sense even the slightest ESD damage, not the type of damage that would cause a transceiver to fail, but the type of damage that over time would result in a failure in that connection.

And so now when we've got customers lighting up racks, they can determine immediately if a transceiver was mishandled and needs to be replaced. Other things that we're sensing is dust on the fiber plant. This is really, really important because even the smallest speck of dust can have light bouncing back in the other direction causing a multi-bath interference situation. And so what we're doing is far beyond. Now, pilot allows us to integrate within the network our customers. And that really just changes the game on giving our customers tools to be active in identifying and mitigating. And look, the goal here is twofold. It's really to deliver the fastest time to revenue. So bringing a cluster up in five or six days versus six to eight weeks.

I think we've seen the contracts that have been reported in the market. You can measure a month on the order of hundreds of millions or even a billion dollars of advantage when you've got that really expensive gear that could be generating revenue. But it's sitting idle because you're trying to bring a cluster up on the order of weeks versus days. The other big one is uptime after deployment. And driving to a 99% uptime or even higher is the objective here. And that just delivers a better fabric, better product for the end customers. And so it's a huge differentiator. I think at a cluster level. But yes, pilot is critical to enabling that. It's a combination of the custom hardware.

But the interface within the network is really the critical piece.

Carl, I apologize. The first part of your question we didn't hear. So I just want to make sure I've got the right perspective.

So in looking at our optical components business, this is a really important part of our business, short term and long term. And ultimately, it's the path where we're going to pursue that pluggable optical transceiver market that is based on mainstream standards.

And the combination of having an optical DSP and a PIC and being able to offer that system level, even within a component offering, that will help our customers deliver what we see as the most competitive combination of system performance, power and yield. Many times, our module customers are working directly with hyperscalers. And the hyperscalers are active in basically pointing to the components that they want to be put together within the modules that our module customers offer.

And so the hyperscalers play a big role in both the component sale part of our business to our module customers, as well as our ZF optics modules that we're building ourselves.

Long term, I think that that is going to be a balance that we see that we'll continue to where we're going to see that, in fact, our component sales and our module sales will be complementary in the sense that the broad part of the market will be addressed by components. And a very specific part of the market that really, really is focused on reliability will be kind of a new product that's offered in that transceiver space.

I appreciate that. The conversation about fiscal 28 and 29 and 30 is something very active conversation within our leadership team at Credo. Let me first touch on on ALC as a part of our portfolio. I think this is, you know, the ALC product that we're bringing to market first is using micro LED technology.

And the promise of this technology is really to deliver the same reliability and the same power efficiencies at a core technology level as AECs.

Key difference there is we'll extend the length to 30 meters. And so it's kind of ALCs will represent our third differentiated pluggable transceiver solution. So, you know, different things we first did, you know, kind of created the product category with AECs, followed by ZF optics, again, creating a new product category. ALCs will be the third leg of that stool. The bottom line is I look at that pluggable transceiver market and I think that's just step one for ALCs and the micro emitter technology. A next natural step forward on that is to apply that to what comes with the scale up opportunity. Because again, in scale up, this is another technology alternative, but the promise there is that at a core technology level, we'd be addressing, you know, some of the problems that have prohibited that market from taking off with, you know, with solutions specifically related to reliability, availability, and serviceability.

And so we view ALC as a big multi-billion dollar opportunity, followed by, you know, as big of an opportunity with scale up and as highly complementary to the suite of technologies that we've brought to market. So the way that I think about the future, more specifically to answer your question is that we're trying to put ourselves in several multi-billion dollar town opportunities.

You can just analyze our portfolio across the pluggable space and you can see that it's really tens of billions of dollars of opportunity that we're now going to be addressing in our fiscal 28 timeline given the fact that ALCs be part of the portfolio. And so we're trying to put ourselves in position to address a very large market. Our growth as a company will follow based on our success and executing with every customer. I think we're quite bullish on the opportunity and that's without even talking about the massive opportunity that the entire industry has in front of us with scale up. Even think about omni-connect, we've articulated in the past that that's a multi-billion dollar opportunity annually as well.

So I think that as we think about fiscal 28, you're right. We think about outsized growth again for another year, but we think that will continue for the years that follow as well.

There is a pretty interesting contrast between the efforts that we pursued with AECs and some of the leverage and how it looks different this time in the ZF optics. With AECs, we imagined this product as being an extension to copper, basically addressing some of the issues that our customers were facing as they were going to faster speeds. And they were having issues with DAX. We were surprised when we had customers pursuing us talking about really interesting innovations at a feature level, things like telemetry, things like other system related rack level innovations. And we really opened that door to the customer base on feature set innovation. Our first customer Microsoft, the reason that they converted the AECs was really the functionality that we offered.

We developed a solution that was smart enough to sense when a tourport was failing or about to fail and then switching the data to a redundant tour in a hitless manner. Really, really smart solution in a cable format. Over time, the momentum built, and this is over over years, momentum built because as speeds increased, it was clear that DAX weren't going to cut it from a signal integrity, but also from a four-factor standpoint with the copper wires needed to become much, much thicker. So from a four-factor standpoint and a signal integrity standpoint, many other customers started looking. And then again, I mean, if we look at our solutions today, many of them have really innovated feature set solutions.

And so the growth happened over a product category creation, happened over several years. As part of developing our solution and having it be a complete solution and taking ownership of it, we developed a really differentiated way of coming to market from a qualification standpoint.

The rigor that we put our solutions through is sometimes far beyond what our customer calls look like, and that includes having our customers switch our customers next, running at speed as we're hammering the link, the entire link, not just our AEC, but the entire link from switch. And, you know, the objective is to harden the solution, finding link weaknesses, and then hardening that through firmware modifications. So that was definitely something that we're leveraging now as we bring ZF optics to market. But the difference with ZF optics was that AI happened, and AI networks were fundamentally different than, say, front-end networks, where there was kind of a built-in redundancy within the different tiers of the network.

And so we've had link flaps for a long time, but they just haven't surfaced as a major concern because there wasn't any massive hit to the network. But now that we've got AI clusters, you've got tens of thousands, if not hundreds of thousands of links that are all interdependent, that if you start getting flaps on some of those links, it can affect the entire cluster to the point where, you know, customers have talked about, you know, losing GPU utilization on the order of greater than 10% or even approaching 20%. So there was a, you know, with this new application, it was a perfect fit for following the playbook on AECs and doing interesting things, very innovative things in an area where innovation had not happened for many years, working closely with customers and basically charting the path to at a system level be able to improve reliability of the network.

That was really the difference. And it took some time to, you know, to develop the hardware and the software solution. But we're seeing that the market is taking off a lot faster than AECs because we're addressing a pain point that exists, clearly exists already. So this is, you know, this is welcomed every time we have a conversation with the technical networking teams within the customer base.

Sure. I think it's, I think there is a case to be made here about our go-to-market strategy and the fact that we are owning the entire stack. And I think when we look at the opportunity, just at a transceiver level starting with the 30s and then looking at it from a DSP and now a pick perspective, being vertically integrated gives huge advantage on your ability to deliver the best possible system solution, but it also gives you an advantage on Cox.

Then you think about going to market with a solution that's differentiated and the challenge is what are those features worth. And so I think we expect an advantage at a Cox level, but we also expect an advantage on an ESP level as we compare our solution to more standards-based solutions in the market. I do think that long-term we are now the pace setter on innovation in the optical space.

The market needs it and we're being pretty open with working with the standards groups on standardizing around some of the things that we're doing. And the challenge there is how can we innovate faster than the rest of the competition, the rest of the market. And I think we feel comfortable given the fact that we own the entire stack, that that's going to lead to the same kind of success that we've had with the AECs.

Appreciate the question. The world we live in at a component level is quite competitive. And so just by proper protocol, we're not able to talk too specifically about exactly which of the major players that we're engaged with. I will say that these first two major design wins that we're talking about do not include the DSP. So there's upside potential with that as we look at really co-marketing the DSP in the picks long term.

Right now, it's the dust team was really doing a great job with engaging deeply with major players in the industry. Amazing that a small team like that had such great traction. And I think it's evidence of the fact that their technical solution is absolutely leading edge in the market. Long term, I think there's going to be a great opportunity for us to look at these two components and really bring a lot of benefits to the customer base.

So I look at this as a great development in our overall portfolio and going to be really promising long term for us.

So I guess with that, we'll wrap up the call. So I really appreciate you all attending. And thanks for the thoughtful questions. We look forward to the follow up. Thanks so much.