CALIX, INC(CALX)Q2 2026 法說會逐字稿

Greetings everyone, welcome to the Calix Q2 2026 earnings conference call. At this time, all participants are in a listen-only mode. A question-and-answer session will follow the brief prepared remarks. If anyone should require operator assistance during the conference, please press star zero on your telephone keypad. As a reminder, this conference is being recorded. It is now my pleasure to introduce your host, Nancy Fazioli, Vice President and Investor Relations. Nancy, please go ahead.

Thank you, Darryl. Good morning everyone. Thank you for joining our Q2 2026 earnings call. Today on the call, we have President and Chief Executive Officer Michael Weening and Chief Financial Officer Cory Sindelar. As a reminder, today after the market closed, Calix issued a news release which was furnished on a Form 8-K and our stockholder letter was also posted in the investor relations section of the Calix website. Today's conference call will be available for webcast replay in the investor relations section of our website. Before I turn the call over to Michael for his opening remarks, I want to remind everyone that on this call we will refer to forward-looking statements, including all statements the company will make about its future financial and operating performance, growth strategy, and market outlook. Actual results may differ materially from those contemplated by these forward-looking statements. Factors that could cause actual results and trends to differ materially are set forth in the Q2 2026 letter to stockholders and in the annual and quarterly reports filed with the SEC. Calix assumes no obligation to update any forward-looking statements which speak only as of their respective dates. Also on this conference call, we will discuss both GAAP and non-GAAP financial measures. A reconciliation of GAAP to non-GAAP measures is included in the Q2 2026 letter to stockholders that was posted yesterday. Unless otherwise stated, all financial information referenced in this call will be non-GAAP. With that, Michael, please go ahead.

Thank you, Nancy. Much like the dot-com revolution before it, the AI revolution is changing how society functions and will transform business models in every industry. The key difference between the dot-com era and the AI era is the rate of change. The AI rate of change is unprecedented; those who move fast will lead the industries they serve. The Q2 was the beginning of Calix, the AI leader, as we began realizing the value of our 15-year investment through the first full quarter of our AI-native Calix One platform being live. Calix One has access to data, insights, and the ability to autonomously or through team member augmentation improve operations, marketing, support, and subscriber experiences for our service provider customers regardless of size. These expanding Calix One capabilities enable our customers to address the threat of broadband commoditization through differentiated experiences, resulting in winning new subscribers, growing revenue, higher retention across consumer, business, MDU, and municipal segments they serve, while improving operating costs with a predictable implementation of AI capabilities. Customer interest in Calix One exploded in Q2, driving record RPOs and record software and services revenue. While the strength of the platform was evidenced by the significant jump in software and service margin as we had all customers live, we also tripled the number of customers that signed up for Calix Agent Workforce Cloud. More important is the makeup of those customers. They were not only the innovators who partner with Calix early in the product life cycle and are committed to the value that can be realized in their business model by being first to market. The customers who signed up in Q2 span the entire adoption life cycle from early adopter, which is expected, to late majority, which signals a significant Calix Agent Workforce Cloud-enabled shift across our base. The shift is that every business leader knows they must have an AI strategy or they will be at risk. The late majority adoption proves that Calix One is a secure, trusted, and predictable approach for every customer to adopt AI. It is secure and trusted as we have invested 15 years into the platform and domain knowledge to meet our customers' needs. Since November 2023, we have evolved the platform to be AI-native while ensuring that our processes, culture, security, and governance expanded to meet the needs of the AI era for our customers. It is predictable as our AI-native platform architecture allows us to use any AI model, and it is our belief that hardened open-source models meet our workflow, use case, and industry needs. That means Calix has solved for the largest issue that is raging across all industries: how to use AI predictably, as one cannot calculate an ROI on a cost that can quickly run out of control. In our architecture, we have issued tokens through hardened open-source AI models and the acquisition of pure compute. Calix One customers are adopting a platform that delivers AI capabilities that can be trusted, are secure, and are offered at a predictable cost, which will yield an ROI that will grow every month with their subscription as we acquire new subscribers, grow revenue, reduce churn, and lower operating costs through the power of Calix Agent Workforce Cloud. At Investor Day, I shared a slide that showed our annual feature rate peaked at 181 in 2018. Our second-generation platform feature rate peaked at 918 in 2024. With the third generation of our platform launched, our customers can expect that our past innovation velocity will be quickly surpassed with Calix Agent Workforce Cloud. Which brings me to the Calix team. This team continues to win awards as one of the best cultures in any industry. Having outlined a subset of awards in the investor letter, including Fortune recognizing Calix in the 100 Best Companies to Work For list, the strength of this type of culture has never been more important. While AI is a powerful technology, it requires teams to get the most of it. Our customer success organization will help our customers leverage Agent Workforce Cloud to transform, and inside Calix, our leaders are taking a human-centric approach to AI to ensure that Calix transforms internally by understanding how we work so that we can agentify those portions of our business that will benefit from it. We are using AI to gain operating leverage across Calix. In summary, we are committed to transforming our customers and ourselves through the power of AI. With that, I'll turn it to Cory to cover the details of an amazing quarter.

Thank you, Michael. We saw continued strong and broad-based demand in Q2, delivering record revenue of $293 million, a 5% sequential increase and 21% year-over-year growth, exceeding our guidance range. Importantly this quarter, we experienced a return to strong software growth. Software and service revenue grew to a record $50 million, up 7% sequentially and 16% year-over-year, as customers began to realize value from agentic workflows on Calix One. Last quarter, with the completion of our platform migration, we said our focus would pivot to agentifying our BSP customers and RPOs would accelerate—

Sorry, Cory, can I please pause you? Darryl, Operator, I understand from some of the people on the call that the call has been dropping. Could you please come on and let us know if we should restart the call?

I am here, and I have heard everything thus far.

Okay. Apparently, those who are participating via webcast experienced portions of the call that dropped, and I'm wondering if we should restart the call once again.

I can confirm as well the webcast has been stable; I have been spot-checking and I have heard the call as well through the webcast.

Okay. You can confirm that the webcast is acceptable and there are no issues for the listeners?

As far as I can tell, there are no issues.

Okay. Apologies, Cory. You can go ahead with where you were continuing.

This quarter, we experienced a return to strong software growth. Software and service revenue grew to a record $50 million, up 7% sequentially and 16% year-over-year, as customers began to realize value from agentic workflows on Calix One. Last quarter, with the completion of our platform migration, we said our focus would pivot to agentifying our BSP customers and RPOs would accelerate in the second half of 2026. Our customers moved faster than anticipated as we saw record RPOs of $386 million, up 3% sequentially and 11% year-over-year. Current RPOs were $162 million, up 3% sequentially and 21% year-over-year. Based on the strong momentum exiting Q2, we continue to expect RPO growth to accelerate in the second half of the year as we deliver additional agentic workflows and demonstrate the value of Calix One. As we discussed last quarter, we would be running on a single cloud infrastructure this quarter, and as such, having garnered the corresponding cost reduction when combined with the agentic platform-driven demand, we yielded an 810-basis point sequential improvement in non-GAAP software and service gross margin. To give you a sense of the continued leverage in the model, I will break with precedent and say that we expect software and service gross margin to set a new record in Q3. As the AI leader in the broadband space, we are also focused on being the leader in human-centric AI deployment inside our own company. We demonstrated meaningful operating leverage in Q2 of 2026. Non-GAAP operating expenses were approximately $122 million or 42% of revenue, down from 45% in the prior quarter, reflecting both leverage in our growth model and early productivity gains from our human-centric AI investments, as well as lower incentive compensation and timing of certain expenses. Turning to appliances, appliance revenue was a record $243 million, a 4% increase sequentially and a 23% increase year-over-year. Non-GAAP appliance gross margin was 52.9%, representing a decrease of 460 basis points sequentially and 170 basis points year-over-year due to higher memory costs, which was partially offset by memory surcharges. As a result of the above, non-GAAP net income was $31 million, or $0.47 per diluted share, above our guidance range. We generated free cash flow of approximately $12 million. Our balance sheet remains strong. We ended the quarter with cash and investments totaling $194 million after deploying $69 million to repurchase 1.6 million shares. DSO was 42 days and inventory turns were 2.7, reflecting deliberate investments in inventory to secure supply and meet continued strong demand. Turning to guidance: for Q3 of 2026, we expect revenue between $301 million and $307 million, up 4% at the midpoint over the prior quarter. This reflects continued strong broad-based demand, even as customers are more tightly managing their own inventory in response to higher memory costs. For 2026, we expect annual revenue to grow at the higher end of the 15%-20% growth range provided last quarter. Our Q3 non-GAAP gross margin guidance is 52% at the midpoint, reflecting higher memory cost impact. As we navigate this industry-wide exogenous event, we remain focused on supply to ensure our customers can continue to meet strong subscriber demand while we maintain a footprint-aggressive stance. As you are well aware, there are many inflationary cost pressures across all industries. In our space, memory cost represents the most extreme of these pressures. As we partner with our customers on surcharges, they value certainty: certainty of costs, and most importantly, certainty of supply. Our surcharge program is structured to deliver exactly that, with the goal to recover the incremental memory costs without adding profit. This means over the long run, memory surcharges will be gross profit neutral while remaining a headwind to gross margin. Given the implementation of our memory surcharge program, we expect appliance gross margin will bottom in Q3 of 2026. The Q3 non-GAAP operating expense guidance is $124.5 million at the midpoint, a sequential increase driven primarily by the timing of expense and higher incentive compensation, partly offset by continued productivity gains from our human-centric AI investments. Michael, back to you.

Thanks, Cory. In the last year, there's been much debate about AI, agents, and the effect it will have on customers and the markets they serve. We believe that Calix is uniquely positioned to take advantage of the AI market opportunity in the broadband market based on our 15-year investment in the Calix platform and our work since November 2023 to evolve the platform and our processes for the opportunity ahead. Our customers all know they need AI, and we are best placed to deliver subscriber growth, revenue growth, churn reduction, and cost improvements with Agent Workforce Cloud in a trusted, secure, and predictable way. Predictability of AI costs is the biggest issue gaining AI adoption, and our architecture has solved that problem for our customers. Trusted, secure, and predictable costs are now possible for all customers, regardless of size, and we began to see the results of Calix One and Agent Workforce Cloud in Q2. Tripling Calix One contracts while delivering record RPOs and record software and service revenue is just the start. This is the beginning of Calix, the AI leader. Now that the platform is live and running, we are excited for the opportunity for our customers to grow in the markets they serve and for our Calix team members as they leverage AI to transform how we do business and deliver operating leverage for our shareholders. Nancy, let's open the call for questions.

Darryl, you can open the call for questions. Thank you.

Thank you. We'll now be conducting a question-and-answer session. If you would like to ask a question, please press star one on your telephone keypad. A confirmation tone will indicate your line is in the question queue. You may press star two to remove yourself from the queue. For participants using speaker equipment, it may be necessary to pick up your handset before pressing the star key. One moment, please, while we poll for your questions. Our first question has come from the line of Joseph Cardoso with JPMorgan. Please proceed with your questions.

Hey, good morning. Thanks for taking the questions. For my first, I wanted to touch on your expectations for gross margins to bottom here in Q3. I appreciate it's a difficult operating environment. What's driving the confidence here relative to calling the floor? Is it largely related to cycling past the grandfathered backlog relative to the surcharges, or are there other levers you're seeing relative to inventory, product redesign, etc., that's kind of driving your confidence in calling a floor in Q3? Then I have a follow-up. Thank you.

Yeah, Joseph, it is exactly as you outlined. We have grandfathered a certain portion of the backlog. As we go through the next few quarters, that backlog as a percentage of the total will shrink. New orders are being assessed surcharges at our current cost structure, and we're now adjusting those on a monthly basis instead of quarterly. Over time, we expect that gap to improve toward gross profit neutrality. Our goal for the whole surcharge program is to get to gross profit neutral. As we partnered with our customers this quarter to address these higher memory costs, it became clear that they value certainty, certainty of cost and, more importantly, certainty of supply. We modified our program to address this. Consequently, we did not raise prices on backlog for a second time. Meanwhile, for new orders, we will adjust to reflect our actual costs and adjust it monthly. As more revenue comes in from those new orders, the closer we come to being gross profit neutral. That's why I believe in Q3 we will bottom in appliance gross margin.

Got it, Cory. Thank you. For my follow-up, you raised the full-year guide to the higher end of the range, which, at the high end, implies a solid sequential ramp into Q4. First, am I thinking about that correctly relative to how you're framing the exit rate for the year? If so, could you help us think through the drivers, given all the moving pieces? You guys have discussed the new platform momentum. I believe BEAD should start flowing through in Q4, and I'd assume some benefit from the surcharge adjustments. Maybe help us think through the acceleration into Q3 and Q4 and what is implied in the guidance. Thank you.

Yes. If we take a look at what we're seeing from our customers, the demand environment remains strong. Understanding that we can see all their activity through our clouds, even though we're seeing higher surcharges and some changing ordering patterns, the underlying demand environment remains very strong. That has not changed. The quickest return on investment is adding a new subscriber to an existing network. It doesn't really matter what those higher memory costs are when it comes to connecting a new subscriber to their network; they are going to move and buy the premises equipment. Number one is the underlying demand environment is strong. Two, you saw a return to our software growth. Exiting Q2, the momentum is very high. We reiterate the fact that we expect our software to continue to re-accelerate into the back half of the year. You are correct that we expect to have some BEAD revenue pick up in Q3 and Q4 as we exit the year. There is no real change in that environment. We are looking at a back half that is strong across the board. To contextualize with customer interactions, I've been on the road quite a bit over the last quarter talking to customers. As organizations move through their build cycles, when they come to the end of building fiber, the entire organization pivots away from being a construction company to becoming a sales and marketing organization that can win subscribers. I continue to hear that theme in many places. At a CEO conference, a panelist said he could see the end of the big builds for his company, and everything he's thinking about is how to add subscribers. For Calix, adding subscribers means we install a new subscriber, provide new services, and have a strong revenue opportunity. Our customer success organization and the AI tools we are putting in place are great at helping on the marketing side. The other side is that we have cracked the code on how to make AI predictable. Customers know they need AI. The biggest challenge is how to do it in a predictable way. Predictability cannot be understated; Calix will become the easy button because we offer a predictable cost model. We had 25 to 30 CEOs in our office a couple of weeks ago; we provided a predictable roadmap over the coming quarters showing workflow by workflow how we're going to agentify their business and where we see payoff and ROI. Those CEOs left feeling enthusiastic about the opportunity to transform their teams and how AI will clearly deliver a strong ROI. That will snowball as other companies see this implemented. You have to adopt it; it's about the fastest path with the most predictability, and that's what we've cracked the code on.

Thank you, Michael.

Thank you.

Very much appreciate that color.

Thank you.

Thank you. Our next question has come from the line of Scott Searle with Roth Capital Partners. Please proceed with your questions.

Hey, good morning. Thanks for taking the questions. Cory, maybe to dive in quickly on the impact of the memory charges in Q3: can you quantify both the sales and EPS impact? It looks like from a cursory glance that EPS would have been higher if you had fully implemented memory surcharges across the board in Q3. Second, looking to prior guidance from Analyst Day for 2027 and 2028, could you give early thoughts on the revenue outlook and particularly on the OpEx front? There was much lower OpEx in Q2. It sounds like you're leveraging internal AI capabilities and efficiencies. Does the model start to change or accelerate when we should start to see operating leverage?

Let's talk about that. The revenue outlook is firming up. There are plenty of demand drivers as we look into 2027 and 2028. Software is re-accelerating. The power of the agentic workflows is resonating with customers and will continue to drive our software and software gross margins higher. We have the tailwind of BEAD, which will happen at some point in 2027. In the meantime, customers are continuing to add subscribers and we see no letup in the demand environment. We reiterate our 15% growth target for 2027 and 2028; they are on track. In terms of OpEx leverage, we are a leader in the broadband space and will continue to be an AI leader in the human-centric deployment of AI inside Calix. You are starting to see evidence of that work in the quarter. We expect more of that to come as we are committed to that strategy. It's too early for us to accelerate OpEx improvement materially; it's a keen focus inside the company and we'll continue to pursue it. I'm not changing guidance at this point; our plan is to drive OpEx at a lower rate than revenue growth next year, leading to operating leverage. Our goal is to get to growing OpEx at half the rate of revenue or better, but that is a goal, not a commitment.

Thanks, Cory. Just on the memory impact to sales and EPS in Q3, that would be helpful. Michael, quick question on the competitive landscape: there was talk in Q2 around Starlink impact. Can you update your thoughts on the terrestrial competitive landscape and the satellite impact on your customers?

On EPS, Scott, essentially it was a flip. We overperformed on EPS in Q2; the decision not to change surcharges on backlog had an impact across quarters. We picked up a nickel in Q2 and we lost a nickel in Q3. For the year, we expect to be EPS neutral. That was part of the decision we made partnering with customers to get through higher memory costs. On the competitive landscape, especially fiber competitor versus fiber competitor, that changes market to market, town by town. Competition exists and it depends on how many people are building in a market and their offerings. Our position remains: if you're in a market and you deploy the full Calix solution, you're uniquely advantaged. If you are the broadband provider who builds a dominant brand in a town, you'll beat competition because you are selling consumer services, winning businesses, MDUs, and providing roaming capabilities through SmartTown that let you partner with local stakeholders and become the dominant local brand. That brand dominance helps win subscribers. Regarding Starlink and satellite providers, they have a place. In super rural areas where it would cost hundreds of thousands to run fiber, satellite is a good option, and for boats it's a good option. Where it won't be an option is in towns with fiber; no logical customer will pick Starlink over fiber due to capacity and experience differences. If service providers implement smart accessories like outdoor Wi-Fi and high attach rates, customers become extremely sticky because they have Wi-Fi everywhere they need it and security protections. Satellite will have a place in the maybe 5% to 10% of the market, but beyond that it comes down to customer execution and service providers can crush satellite when they execute well.

Thanks, Scott.

Thank you.

Thank you. Our next question has come from the line of Christian Schwab with Craig-Hallum. Please proceed with your questions.

Hi, thanks. I want to be clear on the decision not to surcharge backlog again. It appears customer certainty on pricing and that decision was made sometime during Q2, as gross margins are implied differently than at Analyst Day on the 23rd. Also, we bought a substantial amount of stock at higher prices than where we're going to open up this morning. Is that fair, or was that the plan all along?

No, Christian, it was not the plan all along. Our plan from the start was cost recovery and to get to gross profit neutral. As we partnered with our customers, it was important for them to have certainty around price. We implemented the surcharge in Q2 for backlog. To say there are no surcharges on backlog is not accurate; there are. What we said is we are not going to adjust it again as we move forward for those customers, providing them certainty because customers were upset about changing price after the fact. We also moved to adjusting surcharge pricing on a monthly basis versus quarterly to increase the frequency to adjust new orders. That had an impact on Q3 gross margins, but our goal remains to maintain a gross profit neutral stance. Over time, we will narrow the gap.

Okay, that's clear. On a go-forward basis, reiterating 15% annual growth through 2027 and 2028, if memory prices continue to increase and we have surcharges, that 15% is an organic number, correct?

Yes, our growth is organic. Are you asking if it's inclusive of surcharges? All our growth assumptions are organic and account for the environment we expect.

To break that up between premises and access: on the premises side, I see no impact from higher memory costs. As customers bring on new subscribers, particularly in an existing built network, it's their highest ROI activity. On the access side, driven by CapEx budgets, higher costs could cause some demand destruction for builds, but the large majority of our revenue comes from the premises side and access is a smaller part of the business. So to say that all surcharges are additive to our revenue growth rate would not be accurate.

That makes it very clear. Thank you. No other questions.

Thanks, Christian.

Thank you. Our next question has come from the line of George Notter with Wolfe Research. Please proceed with your questions.

Hi, thanks. A question about Agent Workforce Cloud. I know you decided to bundle that with other cloud offerings and not charge customers à la carte, with the view that you would monetize via increased subscriber penetration. A few months into the Agent rollout, do you feel that decision is still the right one to drive subscriber penetration? Do you think that model will work well?

Great question, George. If anything, that's 100% the right approach. Our customers wanted simplicity. Previously everything was à la carte, and customers found it complex. You can't deploy AI effectively unless you're across the enterprise. Operations, marketing, and service need to work together in workflows to deliver outcomes. By bundling and simplifying, customers can adopt at the pace they need. Our greatest growth driver is when customers add subscribers. Incremental revenue per subscriber is small compared to adding a subscriber. Our intent is to roll up the market: the tripling of cloud contracts in Q2 shows strong demand. We have become the easy button for AI and for our path to winning more subscribers, increasing revenue per subscriber, and reducing churn.

Got it. As a quick follow-up, do you have metrics or perspective now on customers that are actually using the agentic features? What's the feedback look like?

It's good. We've just started rolling out the workflows and we're starting to see productivity numbers. We're now in the process with our success organization of doing clear measured ROI with customers. Everyone sees the value, but measuring ROI takes time because you need to compare current costs to costs after the workflow change, typically measured over 90 days. Through Q3 and ramping into Connections in Q4, you'll start seeing a lot of workflows and the ROI data. We will share customer results via press releases showing measured returns—whether it's 10% or 70%. Expect those to accelerate through the end of summer and into early fall.

Thanks, George. Next caller.

Thank you. Our next question has come from the line of Tim Savageaux with Northland Capital Markets. Please proceed with your question.

Hey, good morning. You mentioned a return to record gross margins on the software side, which I think prior were maybe 66% or 67%. Where can that gross margin go? We've discussed the potential for a seven in front of it at some point. What's your expectation for the trajectory of software and services margins farther out and whether there's a ceiling we should think about? Also on the appliance side, do you expect appliance gross margins to make their way back to the low to mid-50s by the end of 2027? What slope should we look at?

Thanks, Tim. We clearly see a pathway to a 7 on the front of software and services margins; we'll continue to march toward that. Where it asymptotes is uncertain and depends on approaches related to large customers, private clouds, and how quickly that business scales. In a private instance for a large customer, the margin is higher and that could accelerate margin expansion. It's hard to say exactly where it asymptotes out to, but 70% is in sight and there's a lot of headroom to expand software and services gross margin. Regarding appliance margins, it depends on memory costs. If you believe hyperscaler capital deployment and the impact on memory pricing, next year could be harder or easier depending on that trajectory. Our program is designed to get us to gross profit neutral over time, regardless of surcharges, and we'll see how memory costs evolve next year.

Great. Thanks.

Thanks, Tim.

Thank you. Our next question has come from the line of Michael Genovese with Rosenblatt Securities. Please proceed with your questions.

Great, thanks. Given where we are with Calix One's rollout and the visibility to driving an acceleration, is there a way to say how many quarters in a row you think RPO could accelerate from here? Should we think about RPO going up through the end of the calendar year? Is there reason to think it would also be accelerating in early 2027?

It's going to accelerate for sure. If you think about the year we've had, Q1 was hampered because we were converting 1,200 customers from the previous platform to the new platform. That conversion was disruptive to sales because sales teams were working with customers to get through migration issues. We crossed that at the end of March, which meant Q2 we started to focus on value and customers asked, 'What's in it for me?' The impact was a tripling of contracts. An important point is the types of customers who bought Agent Workforce Cloud contracts were very different. In the past, Q2 would have innovators and early adopters. The middle majority and late majority would sit on the sidelines, waiting for ROIs. This cycle was different. We had a customer I personally had tried to close on SmartHome for almost a decade, and that late-majority customer signed up for Calix Cloud and asked us to extend the contract beyond our normal three-year term. Why? Because AI is non-negotiable. That customer behavior suggests earlier renewals and stronger RPOs. With Calix One, we expect customers of all sizes to renew early to move onto the platform, and I expect total RPOs to grow.

A longer way of saying this: our customers really know they must adopt AI, and our predictability—trusted, secure, and predictable costs and workflows—is a major differentiator. It is going to be a huge inhibitor for companies that cannot provide predictable outcomes and predictable token costs. Inside Calix, we have been careful and methodical. We've sat down with our 1,200 customers and we know the workflows they run. We know what can be agentified easily and where quick ROI will occur. We can deploy AI in a highly predictable way because we control the architecture—we can use hardened open-source models and avoid unpredictable token costs. We are now the easy button for service providers who want to implement AI and see measurable ROI. That will snowball as customers see peer implementation. We're starting to measure these ROIs with customers and will publish customer stories demonstrating measurable returns. Heck yes, let's go make money for our customers and for our investors.

Great. Awesome. Thanks for all that. Last question: the Q3 guide sequentially is a bit below historical. Specifically on BEAD, was there any change from the second half of the year into the first half of next year, or any other reason holding back Q3 guide from being a bit higher?

No, the Q3 guide is in line with what we outlined for the year, and we're moving to the higher end of our guidance range provided last quarter. Underlying, everything is tracking according to plan.

Great. Thanks very much.

Thanks, Michael. Operator, we'll take the last question.

Thank you. Our last question will come from the line of Ryan Koontz with Needham & Company. Please proceed with your question.

Hey, thanks. Some housekeeping: I know you're not reporting detailed customer tiers, but could you give any color on what you saw across different segments, and the source of the 12% customer concentration in the quarter? Also, an update on any new Tier 1 engagements looking at private cloud options?

Ryan, we're not going to provide a breakdown of customer mix. The 12% customer in the quarter is not something we're at liberty to disclose. The expectation is they were a 10% customer in the quarter but they likely will not be a 10% customer for the year—it was a blip. As for customer engagement, we are broadly engaged across all sizes. The contracts closed in the quarter were mostly smaller customers, which is normal. We are deeply engaged across larger customers too; those deals have longer cycles, typically 12 to 24 months, and we're now deep into these cycles and can actually demo and show the platform rather than just present slides.

Makes sense. On RPOs, we saw a divergence in Q2 between current picking up and long-term not growing as much. Can you expand on long-term RPO expectations going forward? Do you think that will normalize with renewals?

Ryan, RPO is a function of the tail and customers coming back up for renewal. My expectation is with Calix One contracts we'll see an early renewal cycle. Strengthening current RPO is the metric to focus on, and with Calix One I expect customers of all sizes to renew early to move onto the platform. I expect total RPOs to grow.

Makes sense. Maybe lastly on token costs and open source: you've migrated to GCP and are using hardened open-source models. How do you compare competitively to using frontier or off-the-shelf models—what cost savings or advantages are you seeing?

When a company builds AI in a bespoke manner, they enter a complex software lifecycle. The difference between frontier models and hardened open-source models is small for the workflows we're executing. Our use cases are well-defined workflows and context from our knowledge layer, executed by agents within clearly defined boundaries to drive outcomes. We are not asking models to do broad creative tasks; our workflows lend themselves to hardened open-source approaches. That makes our approach highly applicable and cost-effective.

Thanks so much.

Thanks, Ryan. Thank you, Darryl.

Thank you, Ryan.

We can close the call.

Thank you. This does now conclude today's teleconference. We appreciate your participation. You may disconnect your lines at this time. Enjoy the rest of your day.