Francois Locoh-Donou: So, for us in AI, Samik, we’re, of course — I think like a lot of other companies that if we’ve got focused in three areas, one that I’d say is generics to other companies, which means we are looking to leverage the new capabilities to enhance our productivity. And we — as you know, we are focused on earnings growth. We said we want to deliver double-digit earnings growth, which we are now confident we will this year. But continuing to drive that, we want to drive more productivity over time. And these tools would help us through that in certain areas. The other two areas that are really specific to our business, Samik, is one, we have been using AI already, especially in our security products. Part of the rationale for the Shape Security acquisition was also to bring in AI capabilities and AI expertise in the company.
That has allowed us to profile application traffic at a very granular level, which is an incredible powerful capability. And we’re going to enhance these capabilities with new machine-learning models going forward. And I would expect that in security, in particular, you will see us move more and more to AI-enabled security, which is where increasingly what we think it will take to solve security problems. And we have a lot of data being in front of 40% of the world’s websites. We have — sorry borrowing over 300 million website and being is 40% of the world’s web application. We feel we have a lot of access to data that can fuel these machine-learning models. We will also see some opportunities in terms of new workloads. So that is still early days and probably harder to define in the short-term, but what we see, Samik happening is a lot of the new AI-related workloads we think have two attributes that will be interesting for F5.
Number one is these are kind of next-generation modern workloads built with an API first approach, which will create a lot more API connections and API calls and accelerate this explosion of APIs. Those APIs need to be connected and secured, and orchestrated and we are levered to that opportunity. In API and API security. And number two, these workloads or the data they have to access is quite distributed, which also will accelerate adoption of multi-cloud architectures, and we are levered to that multi-cloud opportunity. And so, we think those two attributes of AI-related workloads will play well to the opportunity for F5 down the road.
Samik Chatterjee: Got it. Great. Thank you. Thanks for taking my questions.
Operator: Thank you. Our next question is from Simon Leopold with Raymond James. Please proceed with your question.
Simon Leopold: Thanks for taking the question. I was looking at really where the systems revenue had been prior to the pandemic. And wondering how you would think about the idea that a normal revenue run rate might be in that sort of $170 million and $180 million a quarter. Clearly, a lot has changed. Gilbert, since I guess late-2019. Could you maybe help us think about sort of the puts and takes to even if we don’t know the timing to sort of get a better sense of what should be the sort of base run rate for hardware?
Francois Locoh-Donou: Yeah. Simon, I don’t think we’re — we can kind of direct you to what is or should be the base run rate for hardware. But I think what we can share with you is this. Our perspective is that the trajectory of the hardware in terms of the number of units that we have out there should be declining in the mid-single digit percentage. And it could be a little more than that, it could be a little less than that, but it’s in that zone. And when you look at the sort of overall normalized trend of the number of units we have under obligation, the number of hardware units that are out there and deployed, the — and you look at a longer trend in the last couple of years, you’d see that’s kind of the trend that we’re seeing.
