James Moylan: Thanks, Tal.
Operator: The next question comes from Meta Marshall with Morgan Stanley. Please go ahead.
Karan Juvekar: Hi. Thank you. Yeah, this is Karan Juvekar on for Morgan Stanley. So I guess the first question, I guess, how should we think about the telco versus non-telco customer mix as we go across next year? Would it be fair to say generally flat sequentially from fourth quarter levels just as service provider remains a little constrained? And then I guess as that service provider spend does come back towards mid-year or towards end of the year, I guess how should we think about a more normalized run-rate contribution from here?
Gary Smith: You know, I think you will see service provider, you know the mix I think will from an order point of view will skew more towards the service providers as they come back into the second half. Still expect very strong order flow from cloud. But if I look at it overall, I do think the service provider — it will come back to a little more of the normalized balance, but I do think that cloud is now a larger part of our business sort of direct. And I think it might not continue at the levels it at out right now in absolute percentage share terms with service providers. But I think the service providers will you know come back stronger in the second half. You know so it will be blended this year is another way of saying that.
Karan Juvekar: Yeah, okay, that makes lot of sense. And then just one more, one follow-up. So I know inventories ended up a little bit higher than you were expecting. I was just — and you mentioned just shift of product mix to customers. Is there any more detail you can give? And maybe on that $250 million to $300 million number you gave for next year on inventories coming down, the linearity of that across next year would be helpful.
Gary Smith: Yeah. On the question of the inventory level, really it reflects the different customers, make changes in their request to deliveries. As we move through the quarter, it’s very common they do it very often. And frankly, if it goes, if our mix shifts from service providers to GCNs during the quarter that results in higher levels of fixed — of finished goods inventory. We saw that as we came through 2023. We believed that as we move through fiscal year ’24 that a lot of that movement inside of their delivery dates is going to settle down. And so, we think that our finished goods inventory in particular will come down during the year to a level that much more closely is in the level of what we’ve done in the past. Now I said, we’re going to reduce our inventory about $250 million to $300 million a year that will be generally pro-rata across the year.
Karan Juvekar: Okay, helpful. Thank you.
James Moylan: Thank you.
Gary Smith: Thank you.
Operator: The next question comes from Ruben Roy with Stifel. Please go ahead.
Ruben Roy: Yes, thank you. Gary, I had a question, in the prepared remarks, you made a comment about cloud customers beginning to provision networks for AI traffic. I’m wondering, as you get closer to your cloud customers and you start thinking about AI traffic relative to historical bandwidth growth, would you — would you say at this point that AI traffic would be incremental to sort of that 30% plus growth that we’ve seen historically. The reason I ask is just in the context of the 6% to 8% longer-term, you know growth return to you know sort of the normalized growth, it doesn’t seem like AI traffic would be incremental, but I’m just wondering how you’re thinking of that.
James Moylan: Now listen, that’s a great question. I mean I think — let me give you a sort of perspective on what we’re seeing right now. We are beginning to have the sort of first conversations around with the cloud providers, where they’re placing orders on us, but you know in advance of anticipation for some of the cloud traffic. I would say that, you know they’re working on the applications outside of sort of the ChatGPT versions that are out there now, they’re working obviously on all of these various applications. And I don’t think anybody particularly has any sort of insight into what the, you know Ruben, what the model for that will be from a traffic flow point of view. But I think you know we’re all of the view that at some point all of that’s got to come out of the data center to get monetized.
And it’s no one really has the ability to sort of model that out. But we know it’s big. None of that — when we talk about sort of 6% to 8%, none of that is really taken that into account because we can’t — we can’t kind of model that. But I think it’s going to be — you know, my personal opinion, it’s going to take traffic growth up, you know, which has classically been in the sort of 30%. It’s got to — it’s got to take that up. And so, I think, when will that be, the timing of it, and exactly the model of it, don’t know. But might be, you know my personal guess would be end of ’24, coming into ’25 when you start to see these additional applications come out of the datacenter. But we really haven’t modelled being able to model that in our projections.
Gary Smith: One of the industry analysts did come out recently with a view of a higher growth rate by, you know, 5% to 6% in terms of bandwidth demand as a result of AI.
James Moylan: So close to like 40%, going up from 30% to 40%.
