AWEVFAWEVF
LOADING
|||
Graham Herring: [Abrupt Start] We'll take Q&A at the end of the call and we will moderate the Q&A. You may either use the ask questions by submitting your raise a hand function and I'll unmute you and introduce you. Or you can submit questions using the chat functionality of this Zoom webinar. With that, let me hand over to John, Tony and Rahul.
John Holt: Thank you. Good morning, everyone, and welcome to the call today to discuss Alphawave's half year 2024 results. Thanks for joining us. I'm John Lofton Holt, Co-Founder and Executive Chairman of the company. It's my pleasure to be with you here today. I'm joined today by my Co-Founder, our President and CEO, Tony Pialis; and our Chief Financial Officer, Rahul Mathur. First of all, a bit of housekeeping. I must remind everyone that today's briefing and some of the answers to your questions may contain forward-looking statements. These statements reflect management's current views, and there are risks associated with them. You can find a full explanation of these risks on Slide 2 of the results presentation. The full report and slides as well as a recording of this call will also be available on our website shortly after the call concludes today. I'd like to kick off today's call with a review of our long-term performance since founding the company and since our IPO. Then I will discuss progress that we've made in 2023 and 2024, delivering on our vision for the company that we articulated at the IPO. Specifically, I will discuss the integration of the businesses and consolidation of our M&A transactions from 2023 and how this has turned us into a vertically integrated world leader in communication semiconductors for AI and the data center. After that, I'll be pleased to turn the call over to my Co-Founder, our President and CEO, Tony Pialis, to discuss the business highlights and strong performance of our vertically integrated semiconductor business and end markets that we serve. Tony will discuss our technology leadership, our execution and the competitive landscape for our technology. Finally, we'll have Rahul Mathur, our CFO, walk through a detailed review of our half year 2024 results and our mid-year and long-term forecast. This is the agenda for today's call and who will be taking through each segment of the presentation. When we founded Alphawave in 2017, our vision was to become a global leader in connectivity semiconductors. From 2017 to 2021, we delivered on this vision and became the most advanced semiconductor connectivity IP company, focused on delivering the most advanced connectivity solutions for the most demanding applications. During this time, we held technology leadership as we do today, delivering to the most demanding customers on the planet. In 2021, when we took the company public, we raised capital to accelerate our vision, and we deployed this capital to leverage our IP leadership to transform Alphawave into a virtually integrated connectivity leader. This capital was deployed quickly and efficiently. And through a series of acquisitions alongside our organic growth, we spent 2022, 2023 and the first half of 2024 building what is now Alphawave Semi, a vertically integrated global leader in connectivity, delivering IP, custom silicon, chiplets and silicon products to the world's biggest hyperscalers and semiconductor companies. The basis for our success, past, present and future, continues to be our undeniable leadership in technology, excellence and execution. We've delivered consistent and strong growth since founding the company in 2017, and I'd like to take a minute or two to examine some highlights of our key KPIs over the years. I'm pleased with the organization and their steady and predictable delivery over the last seven years. We've grown our customer base over 100 customers, including most of the top 20 semiconductor companies in nearly all the world's global hyperscalers. Since founding the company, we've delivered nearly $750 million of revenue while delivering cumulative sales as measured by bookings of well over $1 billion as of the end of the first half of 2024. We've done all this while delivering on the vision we had when we founded the company in building a first vertically integrated semiconductor company focused on connectivity. We now have the capability to deliver the most advanced technology to the world's most advanced hyperscalers and semiconductor companies in any form factor they need: IP, custom silicon, chiplets and silicon products. As you will hear later from Tony and Rahul, this underpins our short-term, midterm and long-term forecast and expectations for the company. In addition to investing in our technology and product portfolio, we've also been investing into our leadership team, business infrastructure and governance. Over the last 12 months, we've added seasoned executives like Rahul Mathur, our Chief Financial Officer; and Charlie Roach, our Chief Revenue Officer. Both Rahul and Charlie have led and have been executives in successful global publicly traded semiconductor companies. Their pedigree has allowed them to hit the ground running and immediately contribute to our business. In parallel with these enhancements, we have also invested in future growth through new business and information technology infrastructure. A key part of this is the implementation of a single global instance ERP system, which will increase efficiency across our vertically integrated business and reduce our cost over time. Through 2023 and so far in 2024, we have also enhanced our corporate governance to better focus our leadership on our core markets of AI and the data center. You will hear some announcements shortly about several new Board members joining our organization to add skill sets in these areas and to further expand our diverse corporate governance. On the balance sheet, we've also worked hard in 2024 to continue to expand and strengthen the balance sheet, and we will continue to do so in 2025. Earlier this year, as we announced, we simplified and restructured our debt, which provides us increased flexibility in the future and aligns our debt covenants to operational metrics for how we actually run our business. Finally, we've also been focused on our WiseWave equity stake. As we have previously said, we are in the process of liquidating that equity stake but in a systematic and disciplined way that provides shareholders with a very positive outcome given the difficult macroeconomic environment in China. It's been a pleasure spending a few minutes with you this morning, and I'll be happy to take questions at the end of the call during the Q&A. It's now my pleasure to hand the call over to my Co-Founder and our President and CEO, Tony Pialis. Take it away, Tony.
Tony Pialis: Thank you, John. As a vertically integrated semiconductor company, let me walk you through our product offerings. Since founding the company in 2017, we've built a truly integrated business that can deliver technology in numerous ways through IP, custom silicon, chiplets and silicon products. We deliver world-leading connectivity solutions as silicon IP, which customers integrate into their own silicon. We also deliver our IP into custom silicon that we develop and sell to the end customer. Our connectivity products target the world's fastest optoelectronics used in our data centers. Finally, while the industry has been talking about chiplets, we are the first to deliver them as a portfolio of compute and connectivity chiplets to accelerate AI and networking. Next, let's take a look at our financial and business highlights for the first half. We closed the first half of this year with a revenue of $91 million, which was impacted by the merger of two of our South Korean customers. We closed the first half with an adjusted EBITDA loss of $12 million, highlighting our continued investment into high-margin, high-volume connectivity products targeting AI and data centers for hyperscalers. We continue to invest in maintaining our leadership in the IP optoelectronics and chiplet space for AI and data centers. The AI revolution is just beginning, and we strategically make these investments to ensure we will continue to be a major supplier to the hyperscalers over the next decade. We secured 23 new design wins, four of which were custom silicon wins that will also bring hundreds of millions of dollars of higher-margin, incremental revenue over the next several years. Bookings are the best leading indicator of the growth and forecasted revenue for the business. We closed the first half with $225 million of new bookings. $204 million of these bookings were in high-quality licenses and NRE targeting leading hyperscalers, systems OEMs and leading semiconductor companies. This is a growth of 95% year-over-year and is the best leading indicator demonstrating the success of our transformation to a vertically integrated semiconductor company, delivering licenses and silicon to the world's largest AI and data center operators. Next, let's review our remarkable business growth over the last several years. Our bookings trajectory has been remarkable, exhibiting exponential-like growth due to the various products we delivered to the industry, all of them integrating our leading connectivity IPs. We've had three consecutive quarters with over $100 million in new bookings, with more than 90% of our bookings being derived from high-quality licenses and NRE. Over 92% of our IP and NRE bookings in the first half were on advanced semiconductor processes. And all of the custom silicon deals we won in the first half were chiplet based. As I review our funnel, I am amazed at how every custom silicon opportunity we are engaged in are all on advanced 4-nanometer, 3-nanometer and now 2-nanometer process nodes, all requiring our leading-edge connectivity IPs and all are chiplet based. The chiplet revolution is now truly underway. Next, let me review how we win in the industry. Connectivity is the largest bottleneck for accelerating development of AI. Today, over 40% of the time data sits idle within a hyperscaler data center, being transported around the network, being shuttled across buildings and campuses. Rather than buying more GPUs, hyperscalers understand they need to accelerate connectivity. We win by deploying the world's fastest connectivity to break this bottleneck. We are the only provider in the world that delivers this form of high-performance connectivity as both IP and silicon products. This combination of offerings has made us critical to the largest players in the world and has our competitor scrambling. Next, let me review how we penetrate and win with hyperscalers. Our entry points into semiconductor companies and hyperscalers are not all the same. Some semiconductor companies and hyperscalers start engaging with us on silicon IP for the AI CPU and GPU processors they develop today, or we help them develop their custom processors for their specific applications with our custom silicon offerings. Other hyperscalers initially engaged with us on connectivity products to drive their electrical and optical cabling to connect the front-end and back-end networks. Today, with AI scale-up and scale-out, for every one historical wire, data center operators require at least three new wires. Finally, we are winning today with the launch of the industry's first portfolio of chiplets delivering Arm's Neoverse three cores in conjunction with Alphawave's leading connectivity solutions to enable the next generation of customizable AI silicon via chiplets. Next, let's review what technologies hyperscalers are demanding. As I mentioned earlier, expanding the bandwidths of traditional front-end networks and scaling up AI processors and the new back-end networks requires a significant increase in optical modules and electrical cables. At present, AI is facing critical connectivity limitations. Industry leaders such as NVIDIA and AMD have announced that they are introducing new compute silicon every year. This rapid pace of innovation necessitates industry solutions to provide new ARM-based compute and connectivity IPs and chiplets in the world's leading silicon fabs. Hyperscalers are further demanding supply chain diversification, which requires leading providers like Alphawave to make its technology available in TSMC, Samsung and now Intel. Now let's review the full stack of technologies we offer to hyperscalers to address these challenges. In the field of IP, we offer a comprehensive range of connectivity solutions for AI. No other company has such a wide array of technology and expertise covering next-generation interfaces, such as 224-gig SerDes, PCI Express Gen 7, HBM4 and UCIe 64 gigabit. In custom silicon, we have successfully transitioned the pipeline to a high-margin business focused on AI and digital infrastructure. We are currently at the forefront of the industry by introducing our own range of customizable chiplets to power generative AI computing and networking. Lastly, our connectivity products provide the fastest speeds in the world to support the next generation of copper and optical networking used in our data centers. Now let's examine how all of these solutions will fuel our growth in our industry and in our business. The chart on the left indicates that our total addressable market has expanded to almost $40 billion since last year. What is fueling this growth? It's the rising demand for custom silicon for GenAI and the ongoing chiplet revolution in our industry. So how does this translate into revenue growth for Alphawave? In 2023, we began offering custom silicon solutions in addition our silicon IP and secured several significant design wins, particularly in advanced nodes. These wins are expected to generate approximately $0.5 billion in lifetime silicon revenue. Production of these designs will commence in 2025, marking the start of this revenue stream. By 2027, we will have three different generations of custom silicon wins in production. Additionally, we will ship our first generation of connectivity products in late '24, with production ramping up and continuing to scale through 2027. When we incorporate our customizable chiplet silicon solutions for AI, we are on a strong trajectory to meet and exceed our $1 billion run rate by 2027. Next, let me hand things over to our CFO, Rahul Mathur, for a detailed review of first half financials and an update on our financial outlook. Rahul?
Rahul Mathur: Thank you, Tony. I'd like to walk everyone through our first half numbers as well as our guidance for 2024 and beyond before we take your questions. As we expected, first half revenue was substantially below what we saw in the first half of 2023. If you recall, last year, we saw a significant amount of lower-margin legacy China custom silicon revenue that came with our acquisition of OpenFive. Year-on-year, our royalty and silicon revenues are down almost $90 million. Over 2023, we also recorded approximately $60 million of revenue associated with the subscription license agreement with WiseWave for which all of our delivery obligations were completed last year. Our first half revenues were also impacted by the merger of two of our AI partners in Korea, which meant the cancellation of one program in late June. Without that cancellation, our first half revenue and adjusted EBITDA would have been very much in line with what we communicated at our Capital Markets Day. The good news is that we continue to work with these Korean partners and the new bookings we made in July gives us more confidence in our 2025 guidance. As Tony mentioned earlier, we ended the first half with $486 million of backlog. Over the past several quarters, we've systematically improved our internal processes and forecasting ability. And the guidance I'll discuss later reflects the work we've done to understand the trends between bookings, revenue and backlog. We report both licensing and NRE revenue as well as silicon and royalty revenue. What you see on these charts is the spectacular momentum we have as a company. Our bookings trajectory continues to outpace our industry. What you also see are the trends from an ending backlog perspective. As I mentioned, through OpenFive, we acquired a substantial amount of lower-margin custom silicon backlog that we shipped in the first half of 2023. This is why our first half silicon and royalty revenues are almost $90 million down below what we saw last year as we expected at the Capital Markets Day. As revenue is a trailing indicator of our business performance, I think it's important to look at bookings, backlog and revenue together. On the bottom half of this chart is more detail how -- on how accounting works over the course of the product life cycle. When we recognize bookings and receive cash can be very different from when we recognize revenue, and this can also vary depending on the type of business. We typically do a design start after we receive an order from our customer. In many cases, we receive cash upfront associated with that order, which is one of the reasons we generated significant cash from operations in the first half. Over the next 12 to 18 months, we recognize revenue over the development period. When we have a design that combines our IP and ASIC capabilities, we recognize revenue on a percentage of completion basis. In these cases, the cost of the mask tends to be substantially higher than the cost of the development work. Because of this, the majority of revenue recognized is linked to the timing of tape-out, even though much of our development work has already concluded. A change in tape-out timing for month-to-month or year-to-year can have a substantial impact on our revenue forecast. The updated guidance we provided for 2024 reflects our latest view of timing of tape-outs in 2024 for these key customer projects. Another thing to note is that the $486 million of backlog at the end of first half does not include a substantial portion of production revenue expected in the future. Our silicon revenue is recorded as revenue on a transaction basis like typical semiconductor product revenue. I'd now like to discuss our P&L for the first half. Our reported revenue reflects the trends I discussed earlier. We expect our gross margins to stay in the 50% range through 2025. In general, our silicon IP business is almost 100% gross margin whereas our custom silicon and chiplets product should generate 30% to 40% gross margin. We expect our connectivity products to generate 50% to 60% gross margins once they begin to scale. This implies approximately 50% gross margins through 2025 growing to the high 50s or even 60% at a $1 billion run rate. Our financials also demonstrate continued investment in R&D. It is critical for our company to make these R&D investments to hit our financial targets into the future. Adjusted EBITDA margin of 15% in 2024 reflects the ramp of our business as we continue to invest in standard products, which aren't yet revenue for us in 2024. Without the late June cancellation discussed earlier, our first half 2024 revenue and adjusted EBITDA would have been very much in line with our expectations. On this next slide is more information about our R&D investments and our employee footprint. We have design centers in Canada as well as Israel, India and in California. We continue to invest in an industry-leading design team that's focused on continuing to extend our technology leadership in the critical areas of connectivity and compute. We capitalized $33.8 million of development costs again for future products in the first half, and I continue to expect $50 million to $60 million of cash R&D expenses to be capitalized each year for the coming years. We are on track to grow our headcount by approximately 10% this year, a modest increase as anticipated. On the next slide, I'll discuss our balance sheet and cash flows. As the slide indicates, this is a key priority for us as a company and certainly as a finance organization. We're focused on ensuring that we have the capital available to fund the investment in the products that are going to change our lives going into the future. From a cash perspective, we ended the first half with $76 million of cash, in line with our expectations as we continue to invest in mask sets for our silicon products businesses. I expect to end 2024 with approximately flat or down slightly cash due to the investments we're making in our products. As you saw from our results, we generated $50 million of cash from operations before tax in the first half. Our sustained focus on working capital improvements allows us to continue making investments and bridge through this transition period before we see growth in standard products. As we discussed at our Capital Markets Day, we had initiated a strategic discussion regarding our debt facility in the second quarter. As mentioned in our trading update, we modified our debt facility to align covenants with our operational metrics, provide flexibility in case we wanted to do additional financing with equity or equity-like instruments and added additional capacity. We were targeting completion of this modification in June but, in late June, were notified of the program cancellation Tony mentioned earlier, impacting first half revenue. Without that cancellation, again, our first half results would have been very much in line with what we expected at Capital Markets Day, and we would have maintained compliance with our previous covenants for the second quarter. Because of the late notice of the cancellation, we were not able to formally modify our debt covenants until July. This technically represented a breach as of June 30 and resulted in the debt being presented as current as at that balance sheet date. However, the July amendment gave us cover for any such breach, and we expect to maintain covenant compliance from now onwards. We expect to be able to manage through our current investment cycle with the cash on hand, and our July amendment gives us additional flexibility if needed. This next slide provides more detail about our cash movements over the course of the first half. Ending cash of $76 million was very much in line with our expectations and the decline from last year and reflects the $25 million of capital expenditures we made to support future growth. Net of taxes, we generated $47 million of cash from operations, and we remain focused on working capital. I'd now like to discuss our guidance for 2024 and 2025. Our expectations for revenue are driven by the timing of what we expect to see in customer tape-outs. Our revised guidance for 2024 revenue reflects our latest expectations on timing for these programs and incorporates the 10% risk that I described at our Capital Markets Day. We recorded $91 million of revenue in the first half, and our updated guidance implies $230 million of revenue in the second half. Our updated guidance also implies approximately $60 million of adjusted EBITDA for the second half across $230 million of revenue, so approximately 40%. As I mentioned previously, our ending first half backlog of $486 million and the implied guidance of $230 million of revenue and $60 million of adjusted EBITDA in the second half of this year improves our confidence in our ability to meet our revenue and adjusted EBITDA targets for 2024 and 2025. We expect first revenue for our connectivity products in the second half of this year and expect revenue from our new standard products, including chiplets and the connectivity products, to make up approximately 10% of our 2025 forecast. I spoke about our gross margin trends earlier, and OpEx of approximately 30% demonstrates our continued investment in R&D. This is critical for our company to maintain. In 2025, we also expect to see adjusted EBITDA margins grow to 20% to 25%, and we expect these to scale to 35% to 40% at $1 billion run rate as standard products provide leverage to our model. We continue to expect to see CapEx of roughly 10% and expect to capitalize $50 million to $60 million of R&D each year. Our revenue trajectory is based on the bottoms-up analysis of our bookings and backlog trends and demonstrates the phenomenal growth profile of the company. As I mentioned at our Capital Markets Day, we're growing as fast or faster than the exciting markets Tony described and doing so while making critical investments required to fuel future profitable growth. We're also maintaining our discipline operationally so that we have the cash and liquidity we need to manage our company. With that, I'd like to begin our Q&A session. Could we please have our first question? Thank you.
A - Graham Herring: [Operator Instructions] Janardan, you have the first question, Janardan.
Janardan Menon: Yes, hi, thanks for taking the question. My question is first on the standard products. So your backlog is strong, and that gives you quite a lot of visibility on the NRE and IP outlook through the second half of this year and into 2025. But you are counting on quite a lot of revenue to come through in the next two or three years, all the way to 2027 on the standard products, both chiplets and connectivity. Can you give us an update on -- I mean what kind of confidence can you give us that those kind of revenues are likely to come through, I mean, in terms of your engagements with more than one hyperscaler or customer on the -- and how those -- how the outlook is expanding in terms of interactions on the connectivity products and as well as on the chiplet side? That's my first question. And how can we monitor that -- since that's going to be an increasingly important part of your revenue outlook beyond the near term, how can we monitor that? And what kind of data points can you give us on an ongoing basis? Secondly, I couldn't quite understand what was the covenant that you breached on -- at the point of the end of the quarter in June. Can you just elaborate a little bit more on that? And any clarity on how the new covenants are in shape, which sort of take away that breach, would also be useful.
Tony Pialis: Thanks. Hi Janardan, Tony here. Let me handle your first question. So on products and chiplets, right now, in terms of products where we are delivering and expect purchase orders in the next quarter or two as well as sockets that we expect to close in the next quarter or 2, we probably have more than a dozen opportunities. We can meet our '25 and '27 guidance using our current derated model, which only captures a fraction of the potential of these opportunities. So when I look at the diversification for our Gen 1 and Gen 2 products and I look at what small percentage of these sockets and awards that we need to meet our current guidance, that's what gives me optimism that we can meet. And I expect to also be able to beat, okay? And our pipeline consists of dozens of opportunities, but we have a large amount well progressed beyond that early hyperscaler engagement that we announced back in late 2022. On chiplets, I did mention in my overview that every custom silicon opportunity that we have in the pipeline today is chiplet based. And so here, I am personally taken aback by how quickly in the last six to 12 months, the industry has moved to chiplets for virtually all advanced node designs targeting data center and AI. And so what is helping us win in major top-tier semiconductor companies today is our chiplet strategy, all right? This partnership with Arm has only expanded our leadership beyond connectivity into compute as well. And so as I mentioned earlier, we are in the right place at exactly the right time in the industry with the right set of products. Now how do we measure this and keep analysts and investors apprised of our success? Certainly, our trading updates. We will continue to include not only numbers but also descriptions and highlights of key wins over the quarters. And so as I do in every quarter, we describe how many custom silicon wins we have for that quarter. We will also continue to announce similar types of wins on the connectivity product space as well as the chiplet space. And at our Capital Markets Day, we also did announce projected silicon backlogs. Obviously, these backlogs are not committed, but they're based on our internal derated models. We will continue to publish this projected backlog number on a regular basis. So I think between the trading updates and the projected backlog beyond our committed backlog that we announce regularly, I think you should have a really good snapshot into the momentum of the business. On the second question, Rahul, would you like to cover it on the covenant side?
Rahul Mathur: Yes, absolutely. And Janardan, it's always a pleasure to chat with you. So the specific covenant was our net leverage ratio covenant in Q2. And as I mentioned earlier, this is something we were well aware of going through the quarter and worked very closely with our lending partners. As we went through the end of the quarter, we got the news about the program cancellation and then signed the updated covenants in July. I think that when we look at the math, had we not seen that cancellation, again, our first half revenue and adjusted EBITDA numbers would have been very much in line with our expectations, and we would not have been in breach. To answer one of the questions I saw in the chat, there weren't any extra costs associated with any technical breach or waiver. And our interest rate for our debt facility is slightly higher, but I think that really reflects the additional flexibility that we talk to our debt partners about if we need to do something more from a capital perspective. Janardan, does that answer your questions?
Janardan Menon: Yes. So is the current agreement -- it has a higher leverage ratio. Is that what I should understand from that? And then on -- to Tony, just one follow-up on the chiplet side. You had also mentioned -- you mentioned I/O and ARM-based chiplets, but you also mentioned memory chiplets. Can you just expand a bit on what you're doing on memory?
Rahul Mathur: I'll answer the first part. The net leverage ratio covenant has been amended to meet essentially exactly where we were in Q2. So it really is just a 1-quarter blip. And as we execute over the course of the back half and next half, we get more and more room across each of our covenants. So we're confident that we've resolved this correctly.
Tony Pialis: Understood. And Janardan, on the chiplet side, connectivity is throttling AI. Similarly so is memory, both memory bandwidth as well as memory capacity. And so training these days typically uses high-bandwidth memory. Some models need access to larger pools of memory. And so they leverage LPDDR or DDR. Look, when you're building an AI chip on 3- or 2-nanometer and a mass that costs you north of $15 million, customers struggle to choose only one of these memory interfaces. And so what we have done is we have deployed a UCIe-based chiplet or family of chiplets that allow customers to integrate either high-bandwidth memory or DDR or LPDDR. And rather than integrate it on a leading 2-nanometer node, it's a predesigned, prebuilt chiplet that they can instantiate into their package. And so look, they can have one line of products for certain customers that need memory bandwidth and deploy HBM. They could have other product lines that deploy an alternate chiplet that provide access to larger pools of memory via LPDDR. And so this flexibility in conjunction with I/O flexibility and compute flexibility is how we're going to drive the industry forward and win with this chiplet revolution.
Janardan Menon: Understood. Thanks.
Graham Herring: The next question we have from Rob Sanders. It's been typed in question. Can I say it? What is the typical deal size LTV-wise for AI inference custom silicon deals? And what is your -- I'm sorry, just the next question is, what is your likely win share in AI custom silicon for inference?
Tony Pialis: Sounds good. So look, typical deal sizes for us on advanced nodes, I'd say anywhere between, these days, $40 million to $80-plus million. Again, it depends on the size. It depends on the complexity of the design. So look, they're very beefy, meaty upfront committed amounts. And customers only spend this type of dollar on NRE if they expect huge returns on the back end, and those returns on the order of hundreds of millions plus. So they are very, very large programs. And obviously, that's why if one of them delays across the reporting boundary, it currently has an impact on our numbers. In terms of share and what we win, yes, look, we're -- I don't have a single number. But look, our conversion rate and our pipeline is well north of 50%. So we are winning, and we're winning because of our leading connectivity technology that we bring to the table via IP, via prebuilt chiplets and via the momentum that we're generating in the market on the custom silicon side.
Graham Herring: Thank you very much. A question from Alyx Wood.
Alyx Wood: Just a bit more clarification on the debt facility. Obviously, it's been amended four times and was in breach in '23. What's the interest rate now on the RCF and term loan?
Rahul Mathur: The interest rate is essentially a function of SOFR plus another few hundred basis points. There wasn't a substantial change in the overall interest rate. And I saw your other question, there weren't any extra costs for anything that we walked through from a Q2 perspective.
Alyx Wood: So what's the number right now?
Rahul Mathur: So I don't think we specifically talked about the exact number, but what I'd say it's very competitive given the timing of when the debt was negotiated.
Alyx Wood: And -- okay. That's a bit strange, all right. Thank you.
Graham Herring: And just the last question that we have at the moment is, given this cancellation caused you to breach the covenant, did you receive any compensation from the customer?
Tony Pialis: So let me handle this one. Look, what we do on our debt facilities versus our customers are unrelated. Now what I will say is typically, our deals do not have termination clauses, or they're defined in terms of what are the repercussions of triggering a termination. And that -- look, that typically comes with additional fees. Rahul did mention in his section that as part of this merger and redefinition, there was additional fees that we brought in. But in addition to those fees, look, I'm more excited about the new design wins that were also awarded that will lead to a significant revenue stream, not only upfront with NRE but also on the back end with expected silicon that we will ship in upcoming years.
Graham Herring: Thank you. We have a question from Sandeep.
Sandeep Deshpande: Yes, hi. Thanks for letting me on. My question is on the cash on the balance sheet. We've seen over the past year on a year-on-year basis, your cash is down slightly less than $50 million. That is the cash which is currently available on your balance sheet. That's not the net cash as such. But the question I have is, how do you see this progressing to the second half of the year? What do you think -- how are you modeling your operating cash flow and the cash on the balance sheet into the second half of the year and into the first half of next year when these current challenges of the product ramp will not have necessarily all happened? Given -- will the cash burn remain through the second half of this year and into the first half of next year? Or do you think that you're going to produce cash into the second half of this year and the first half of next year?
Rahul Mathur: Sandeep, thank you very much for the question. What I'll tell you is that the cash drop from the end of last year to the first half of about $25 million is almost exactly what you saw our investment in capital expenditures, and that's primarily related to the mask sets associated with our standard products. I think if I look forward to the end of this year, I'd expect cash to be roughly flat or maybe down maybe $10 million for what we posted in the first half. And then over the course of next year, cash to be roughly flat or maybe increasing slightly as our business executes. So this is something that we are monitoring and managing very closely. And Sandeep, you would have seen it with the improvement in working capital. We generated post-tax, $47 million of cash from operations. And that really goes back to the timing of cash receipts and payments, which is not very closely related to revenue recognition. Revenue tends to be a trailing indicator for us. And so that's why you see the strong bookings momentum and our ability to maintain the liquidity we need to invest in our business.
Graham Herring: Well, thank you all. Can I ask if there are questions? Well, John, Tony, Rahul, thank you very much for a very thorough presentation. If anyone has any questions after this, please do not hesitate to contact ir@awavesemi.com. We'll hasten to get back to you with any queries whatsoever. But following that, thank you all, and thank you all for your attendance.