Advertising startup CloudX is launching today with a goal of unlocking the AI opportunity in mobile. If experience speaks to the potential for success, the startup should be in good shape.
Led by key founding members of the MoPub and MAX teams including CEO Jim Payne and Dan Sack, CloudX has set its sights on automation and agents with a new mobile supply-side platform (SSP).
In preparation for the launch, the company closed a $30M Series A round of financing from lead investor Addition (run by Lee Fixel, formerly of private equity firm Tiger Global), and participation from DST Global, Terrain, and seed investors ENIAC, Javelin, and Breakpoint Capital.
Central to the company’s initial product development is what CloudX calls a Trusted Execution Environment (TEE). According to a press release, TEE “underpins every auction and transaction, delivering verifiable transparency and data integrity, while providing the secure framework needed for AI agents to transact and optimize monetization in real time. This new foundation enables intelligent optimization for mobile publishers today while paving the way for the next wave of generative environments, where ad infrastructure must be reimagined to unlock deeper engagement for users and long-term value for publishers.”
Beginning today, the company’s SDK is available to any mobile publisher. Partners Meta, Liftoff, and Magnite are already aboard. Broader access is planned for early 2026.
Mr. Payne spoke to tipsheet about CloudX and the road ahead including:
- The transition back to the startup world.
- Comparing CloudX to past startups, MoPub and MAX.
- Go-to-market: automation and agents.
- How publishers will work with CloudX initially.
- Details on demand-side partners including Meta.
- How AI is a part of CloudX strategy.
- On mobile agentic buying and selling, protocols such as AdCP.
- The creative opportunity in mobile today.
- CloudX and ads for AI-enabled chatbots.
Scroll down for the interview which has been lightly edited for clarity.
TIPSHEET: Can you discuss your transition back into the startup world?
JIM PAYNE: For me, the AI wave is too hard to resist jumping back into it. There’s a couple of reasons why that’s the case.
One important reason is that technology enables a much better product to be created for the publisher. And so I’ve been very motivated to unlock that, because it’s not just about what AI can do to make workflows more convenient. We actually need the platform to be connected to the broader ecosystem.
So, when Dan Sack and I sat down in the Spring to talk about what we could actually deliver to the publisher, a big piece of that came from additional liquidity that we could bring to the marketplace. We essentially created this platform to do both of those things: bring true automation to the mobile publisher, and connect them to a broader ecosystem with whom they can actually deploy this product.
How is it different this time than with MoPub or MAX?
(Background: MoPub was acquired by Twitter in 2013 and later re-sold to AppLovin by Twitter in early 2022. MAX was acquired by AppLovin in 2018. Mr. Payne and Mr. Sack were a part of the leadership team at both startups. Mr. Sack was also VP of Platform at AppLovin for 4 years.)
MoPub was the very earliest stage of mobile advertising. Back then, it was just a handful of ad networks and the mediation category wasn’t really a category, because you’re really just moving between three different choices in the waterfall – it was kind of a simple problem.
Similarly, the publisher community was fairly young – these are individual, two- or three-person gaming studios that had launched something that became really large -and also the user acquisition and app install business, which was very nascent.
With MoPub, we were trying to bring a bit of the programmatic approach to mobile advertising for the first time. And in so doing, we built the sell-side platform (SSP) tool which was connected to an exchange. That really worked, but it was very early, and we weren’t really sure where the mobile advertising industry was going and how important it would end up being. Obviously, we got off the train a little bit early.
And then MAX was oriented around making things more convenient and liquid for the publisher – that was in 2017 we came up with the idea. And that was basically: things were stuck in mediation and didn’t quite get all the way over to a unified option. We were basically working with Facebook at the time and trying to move a lot of that liquidity over to something more unified.
I think both of those were incremental opportunities in my mind, whereas the the opportunity in front of us now with automation —and especially vis-a-vis the size of the mobile advertising ecosystem today — to me, that’s where the step-change function comes in to play. What we can do is finally, for the first time, deliver to the publisher a lot of the capabilities that they’ve wanted to do but it’s been too expensive, cumbersome or difficult to do…
That would be things like: modeling price floors for themselves; setting up true programmatic, line item infrastructure; doing deals with buyers at certain price floors; and carving out inventory and doing segmentation. You could do all that stuff with MoPub, but it was very hard to do, it was error prone, there were a lot of people involved, the technology didn’t scale and the buyer ecosystem wasn’t there to take advantage of it. And I think a lot of those things have now changed.
What the CloudX platform is about — and what we have been working on for the better part of the year — is applying an agent-oriented approach to what a publisher’s workflow might be in its ideal state. And that is a sophisticated surface area interacting with buyers. And there’s a lot of liquidity that’s coming through our secure auction that we’ve developed in conjunction with Meta – and that unlocks unique liquidity from them.
Basically, the agents can start doing a lot of the work that a traditional ad monetization person would have to do, manually: setting up line items, targeting, making changes to price floors, seeing what’s working, anomaly detection, and so on. And then over time, the system itself can get smart and build intelligence those features and improve over time automatically.
How do you think about the market for this today -specifically around agents? Is it there? Or are you building for something that you expect the advertising world in mobile is going to grow into?
The idea is that the market is so big that the platform can support a more sophisticated approach.
I think before, it was fairly simplistic, and that was fine. And now, publishers are running large businesses and they’ve got different segments of users that have different requirements – you want to show certain users higher quality stuff or target and segment them with price floors differently, etc. That wasn’t much of a problem when the market was relatively small and you were struggling to create a business overall.
As these businesses are now larger, you have to be more sophisticated to optimize what’s going, and think about long term value of the user population you have. So that’s why we think the time is right for this kind of approach.
And of course, if it’s not automated, it’s also not going to work either.
The twin forces of extreme automation and ease of use of complicated infrastructure that ordinarily wouldn’t really be worth the effort – now, because the market’s there, and the business and the liquidity are there, you actually make it possible to do in an easy to use way.
How will publishers work with you initially?
So what we’re launching [today] is basically a mobile advertising monetization platform. It’s got an SDK that will integrate into a publisher app across Android, iOS, Flutter and React Native apps. You can then carve out some segment of traffic and send it to our back-end system. And that back end system will host an auction for you. The auction, ultimately, is by the publisher’s design: who they want to participate; what traffic they want to allocate to that auction; and the price at which that auction will be conducted at.
And then they can decide if they want to actually show that ad in the client or not. All those inputs in the decision will be able to be coordinated by a line item set up and agents that can automate that stuff. Then the other side of it is to see what actually happened. We’ve got real time analytics. It all goes through Clickhouse. People have full access to logs. They can make the decisions based on all that transparent information.
Can you want talk a bit about the demand side that’s involved initially?
So our launch partners are Meta, Liftoff and Magnite… we’ve got a tranche coming later.
And the reason those folks have signed on in these early days is because we’ve created this secure auction that I talked about a little bit. It’s a cryptographically secure auction with audited source code in it.
When we were talking to Meta earlier this year, one of the things that they wanted to make sure is true is that their bids get expressed in the auction fairly, and that there’s no signal loss through the auction process – in other words, there are no models being built on bid information that they send through. And not to say that that is happening today, but obviously you don’t want to allow that to happen.
So we created a secure auction framework, and that framework allows for additional bid liquidity to come through. And Meta is an important large launch partner for us in that effort.
Those are our launch partners. I think that there will be more people to sign on to this particular notion of a “glass box” approach — where publishers have the levers to build their business in the way that they want to. And, it’s very attractive to folks who are trying to get additional access to liquidity and to supply and carve out stuff that’s going to be unique to them.
How is AI a part of CloudX’s platform and strategy?
We’ve basically built the platform to be operable by AI tools. So, if you think about the way that people write software today, it’s Claude code, you’re basically monitoring a set of agents that are shipping lines of software. And then you’re testing it on the other side —probably with some agents and automation.
So if you think about the approach that we’ve taken and why it’s interesting, we have created a platform such that all the monetization logic is done in a set of files. It’s infrastructure code like Kubernetes might be configured as files rather than UI where you’re clicking on things.
Because it’s files, the agents can go in and directly edit your line item set-up, your block lists, your domains… And then on the other side of what happened in the auction, you can examine the logs and then tie that back and say, “Are there issues? Did we make any mistakes here? Are there things that aren’t being bid on and we think they should be?” – all that can be done in an automated fashion because we’ve designed it as a piece of infrastructure that can be used by agents.
And I think that’s the key to the software product – it’s designed to be used by AIs.
And then the second piece is once we have all the data flowing through this platform, then we’ve got the ability to deploy the same type of deep learning algorithms that are used by ad networks for bidding and start to help publishers be smart about how to allocate their inventory and price it properly. Once we have all the logs together, we’ve got network effect — and a lot of publishers using the platform — then we’ll have those insights that we can distill down into something that can be used by publishers to easily set those prices in real-time.
The flywheel starts to really spin at that point, and again, those insights feedback into something that an agent can plug in to the platform and say, “Here’s a line item for this, here are the bidders I think should be allocated to this particular set of traffic… here are the bidders who are going to be best for your users that you want to retain…”
So those are the abilities that we’re trying to give the publisher through the platform. In the past, if you put that out in a platform that’s not automated, then it just degrades to a unified auction for each piece of traffic that comes through. We’re trying to make it a more sophisticated world that’s easy to use.
Regrading CloudX’s ability to address agents and agentic buying and selling – it’s still early today. There are efforts going on such as Ad Context Protocol (AdCP) which I’m sure you’re familiar with. Do you see working with something like that? Or, in fact, maybe you’re trying to create your own standard with the platform, and everybody uses this platform for agentic buying and selling on the mobile front?
So, I’m a huge fan of the AdCP effort. Obviously Brian O’Kelley is breaking some important new ground there. I would expect that we will have a version of AdCP that’s attuned to the mobile experience and the particular inventory requirements and signal and so on.
Right now, we’re thinking about how does a publisher go from the existing world where they’re very constrained to a world where they have a little bit more flexibility around the auction.
I think the second phase of that will be opening it up to buyers who can buy agentically if they’re operating an agent. I don’t think mobile is quite ready for that just yet. But it’s not hard to imagine that it would be ready because, ultimately, the way that our agents work and do integrations, line item setup, create options and set floors, is through an MCP (Model Context Protocol) – but it’s just an MCP “that only I could use.”
So, it’s not a stretch to say, “Let’s open that up to other people” or maybe have people on our platform say to each other, “I’m pricing my own traffic. What if I were to price your traffic and buy directly through you through an agent that we both agree on?” That’s not a huge leap. I think it’s coming.
Right now, our phase one is just about making the existing world a little bit more automated for people, but as you hopefully can see, we’re laying a foundation for a tremendous amount more automation. I think that’s probably inevitable in some capacity.
What are you thinking about creative in mobile today in general? There’s always been a feeling of limitation with creative in mobile. Will CloudX be addressing creative in any way?
I’m glad you asked the question because the answer is not really what publishers want.
The publishers we talked to, they want to be able to have control over what the creative experience looks like. They understand there’s a trade-off between aggressive creative and monetization. But at the same time, there’s also a trade off between creative and churn. So it’s a sensitive calculation.
There are some users that you can aggressively monetize and will never go anywhere -and there are other users who will move on very quickly. And so that sensitivity is something that’s top of the mind of most of the customers that we talked to, and people who are in our platform. They want to have more levers around that.
Today, obviously, mobile gets a bad rap on creative —it’s not always great. Some of it’s malware and spam and includes bad practices like “can’t find the ‘close’ button”, which we all agree should be rooted out.
But how do you actually root it out because it’s hard to find the stuff?
I think that’s a real area where we want to move the industry forward. There are ways for a publisher to create an auction that’s — let’s say just for Meta. Meta has a certain set of creative guidelines and templates that they use. Maybe you want to create an auction for your highest quality users – the ones who’ve been with you for the longest time, and just give a first look to Meta at an auction at a certain price floor for those users… -that’s something that is hard for people to do today. But our platform makes it easy to do that.
And then the secondary consideration there is: how do you segment and know which users should get that special option right? So that’s also a form of intelligence and modeling, and building some real machine learning around exactly what that trade off is and the economics are.
But I’m glad you asked the question for a second reason. There’s a new category of publisher that’s doing AI-first, generative experiences. Some of this stuff is entertainment content that is generative in nature about creating a character and a custom experience for me, for example. A lot of its text and there’s some signal in the text. And you need a native ad unit that actually matches that experience.
A 45-second unskippable video doesn’t work in that format. And it’s even more costly because the users are hard to acquire. They’re expensive to serve, and if you churn them out, the economics don’t really work. So this is an area where we want to move the ball forward a bit on actually creating new, AI-oriented, generative, creative experiences, too.
It’s not too hard to imagine a world in which text creative is being generated on the device using all that signal and the conversation that’s happening without it actually going through the programmatic pipes.
That’s an interesting problem that I think we’re going to want to attempt to solve down the line.
Is there a fit here for CloudX and advertising within an AI-enabled mobile chatbot?
Sure. Like I was saying before, there are entertainment experiences that are generative in nature and text-based. You could also imagine — and these don’t exist yet — a more commercial-oriented conversation with a chatbot. There’s a bunch of signal in that conversation, and you don’t want to send the entire conversation to a programmatic auction and expect someone on the other side to really understand what that. That should probably live on the device.
How do we actually extract that commercial signal and “vibe target”? – for lack of a better word. How do you package that up and make something that can be sold in a way that ad inventory is sold today?
And then how do you actually render that — whatever the winning bid is — in a forum that makes sense in that context that feels like it’s part of the content and, again, should probably happen on the device.
So I think the answer is, certainly there’s going to be an advertising format that is going to be compatible with chatbot commercial activity and commercial-oriented queries, harvesting intent, and showing people something that is very highly targeted to them. You know… helpful advertising like the original search ads were.
How do you implement that in a world of distributed intelligence where a lot of these chat conversations are happening across the entire ecosystem? I think CloudX is going to be positioned to figure that out.
The goal of the company is to create a platform upon which that solution is enabled. Ultimately today, without connecting everybody together, you can’t really do it, or it would just be one company that’s doing it just for them.
So, my approach over my various companies has always been to try to build a broad-based platform upon which transactions can happen, and try to enable as many transactions to happen as possible, and do it in a thoughtful way for the publisher, content creator and the user -as well as allow people flexibility to pick and choose among various approaches to modernization.
A year from now, what does a successful CloudX customer look like?
I think you give them the tools to monetize in a way that resonates best with them. So ideally, they’ve got a bigger business. It’s driving more revenue, and it’s also doing it in a way that feels very comfortable for the type of content that they’re creating. So, it’s a good match.
It’s a high quality monetization experience that also drives a lot of revenue for them, and most importantly, brings the revenue back that they’re collecting in the marketplace. If they add more signal and targeting, they carve out a better audience, and they match that audience to a better buyer for them that actually drives their bottom line so they can create more and better content over time. That’s where we will find success over the next year and beyond.