Youssef Squali: And then maybe one more. The 3D opportunity, can you talk a little bit about that and the partnership with NVIDIA? I think you talked about it being a 2024 event. Would you expect it to be early in 2024 or is this kind of by end of year? And how should we be thinking about the revenue opportunity for this one?
Jarrod Yahes : So Youssef, as you know, this is something we’ve been working on for some time. This has been an area of investment. Our training of 3D generative model is a different and likely a more challenging endeavor than an image generation model. And so there’s been a lot of work taking place behind the scenes. We’re excited to bring this to market. We already have been working with alpha customers on the large customer side, who have an interest in this technology. There is the potential to significantly lower cost of content creation across a range of use cases and opportunities from gaming to film development. And so there is a lot of interest and there are not products in market that really are trained with the level of clean data that our product will be trained with.
We do anticipate having this out in the early part of the year. This is not a second half of the year event. This is a first half and maybe even a first quarter type of rollout and we’re already in extensive testing with alpha customers. So we are quite excited about this opportunity. This is not necessarily going to be a retail opportunity at first. It’s going to be an API offering for some of the most sophisticated large customers in the world, but we are quite excited about this opportunity here.
Operator: Our next question comes from Andrew Boone with JMP Securities.
Andrew Boone: I wanted to ask about Giphy and where that sits today as well as how this factors into your new kind of 2017 framework and what we should be contemplating there? And then there are press reports that Reddit just signed a $60 million a year deal for their data. Can you just step back and talk big picture about pricing and how you guys are thinking about pricing your deals? And with respect to that $60 million a year figure that’s out there. How do you guys think about your data sales?
Paul Hennessy: I’ll take Giphy and Jarrod can take data. On Giphy, Giphy plays a large role in the data distribution and services model going forward. We’ve been since acquisition in the middle of last year, we’ve been dusting off the ad platform. We’ve been going to market both with ad sales and working with our API partners for value exchange. And as you heard in my prepared remarks, given the amount of traffic and what I would call modest CPM rates, we believe this business is in the hundreds of millions of dollars and therefore plays an important role in the growth of our new area called data distribution services. So Giphy is critical to that element and we’re super excited with the momentum and the performance thus far.
Jarrod Yahes: I would just add on to that by saying that Giphy is already growing. Giphy is already acquiring clients. We’re very excited about the potential here and Giphy also plays into data, as Paul mentioned previously. Hearing that Reddit is looking at a $60 million deal annually for its data is not entirely surprising. I think there is a broad realization that training generative models on data that is scraped, that is not paid for, where content creators are not remunerated for their works is not a sustainable long-term business model. There is a case pending with The New York Times that I think people are eagerly awaiting the outcome of. And I think while it is possible to scrape data and use it in a model, ultimately if enterprise customers are going to want to use the works of that model, they are going to want to know what ingredients are used in the training of that model.
And so that is going to benefit our business and that is benefiting our business. There are companies that are taking shortcuts today and they are able to train our models, but I think what they’re going to find is, if you’re going to want to actually commercialize that model, you are going to need to convince your end customers that the training data set that was used was rightfully acquired. And so we believe that that’s a significant tailwind for our business. As Paul spoke about, we think this is a $9 billion TAM with very significant 20% type of growth potential and we are just in the early stages of gearing up this business for growth, bringing it to the cloud ecosystems for distribution, adding to our business development team in order to get our data out there and today working with many of the hyperscalers, but also newly extending our reach into smaller companies in the generative ecosystem, many of which have received billions of dollars of venture capital backed financing.
So we’re excited to sell and expand into the data ecosystem.
Operator: Our next question comes from Nitin Bansal with Bank of America.
Nitin Bansal: You mentioned that you were expanding your delivery model by leveraging cloud marketplace partners, which allows you to go from like wholesale provider to like retail providers. Can you throw some light on the pricing structure of this data for retail consumers? And secondly, your competitors also have similar set of data. So what kind of competition are you seeing in the data market? And what are you seeing as the expectations for that in the long basis growth?
Paul Hennessy: As you would expect the way we price our data is effectively depending on the volume of data consumed. So there is a volume based pricing where the more you purchase, the lower the unit price of that data. There is differential pricing for images as compared to videos, music and 3D. And ultimately, these deals are fairly individually negotiated depending on the use cases of the customer. Some customers would like access to this data for generative model training for a number of years, other customers are looking for a shorter period of time. So, I think that impacts the pricing as well. Ultimately, as we think about this, we think that there are tremendous opportunities here in order to grow and these cloud ecosystems are going to be the place where that distribution takes place.
