CloudX - Impact to AppLovin?

[brittlerock]

Today (10:00 AM EST, 2/4) it was reported that startup CloudX has developed an LLM driven capability that automates much of the work performed by engineers operations teams. There is a perception (my word, as I don’t know if anyone really understands what the impact of this is) that this will disrupt AppLovin’s near lock on the mobile ad vertical. APP is down 16% as I write and in apparent free fall. CloudX went GA this morning.

The company is not publicly traded, so it’s hard to get a read on exactly what threat is posed to AppLovin by this announcement. I have browsed cloudx.com to determine if they overtly have any statements regarding how they might compete, let alone disrupt, AppLovin. Maybe I’m just not knowledgeable enough to see it, but I don’t see it. They do claim to have the capability to automate much of the effort involved with game development. If I am correct in that observation I would think they pose a major threat to Unity, but so far as I can determine, more mobile games would function more as a catalyst for AppLovin than a threat.

So, is this truly a threat to AppLovin or a knee-jerk market reaction to news which may enhance AppLovin’s ability to generate revenue and a stellar buying opportunity? I wish I had more confidence in my assessment. In fact, I simply don’t know.

So far, there has been no comment from AppLovin to staunch the hemorrhaging of the stock price.

[jonathan1]

I’ve no idea - but APP has been in free fall for a while now. We are down nearly 50% from the highs in APP just a few short weeks ago. I did make some comments on the short reports and on Project Genie in my end of month portfolio review a few days ago. The Cloud X threat only came out today. I hope others can chime in too. I myself am going to keep holding my large allocation in APP until management confirm what is happening on the ER and in the call next week. Not long to wait till then. It may of course be the buying opportunity of the year! I hope that APP are taking advantage of this drop to buy back still more shares themselves.
I guess this sell off is very overdone. APP have their own AI engine called Axon. Surely that can rival and better whatever Cloud X might be?

Jonathan

[brittlerock]

Slightly more information is available in the linked article from AdExchanger. I read the article and remain pretty much in the dark as to how exactly this will have a negative impact on AppLovin - but I’m the first to admit that I find the modern advertsing world pretty bewildering. I have been pretty much taking it on trust that AppLovin was pretty much in the driver’s seat with their AXON AI engine.

What I do see, is what might be a potential impact to the AppLovin MAX application which is the instantaneous auction platform used to place the vast majority of mobile ads with content providers. I might be wrong, but so far as I know there are only two ways for a game maker to make money from a free-to-play game: 1) in-app sales, offering the game player various tokens for a fee that enhance the game experience, or 2) selling access to the game for an ad to run during the game. AppLovin doesn’t realize any revenue from in-app sales (so far as I know). All their revenue as garnered from a MAX commision for mediating the transaction between the publisher (game maker) and the advertiser. And a fee for the services of AXON ad manager. MAX commisssions are small, but the transactions number in the billions/day. I believe the commission is paid primarily by the content provider. The AXON fee, paid by the advertiser is significant. From what I gathered from the AdExchanger article, CloudX provides an alternative mediation service that would likely impact MAX revenues. If there’s a potential impact to AXON revenues, I don’t see it. But, I hasten to add, I’m operating on very scant information.

As for my prior observation that I saw a potential impact to Unity - well, I guess that was off base in that the article states that Unity and CloudX have some kind of partnership arrangement. So does Meta along with Magnite, Liftoff (I’ve never heard of this company before) and others I imagine.

CloudX Hits GA With Plans To Rewire The Mobile Ad Stack Using AI Agents |...

[GDavenport]

At this point, there is little reason to sell. APP is on track to deliver results (and confirm with data) how they have been building the business. They will also have a chance to answer analyst questions about the above issues.

Also, from a macro perspective the CNN F&G index dropped 30 points today from 65 (greed) last night to 34 at the moment.

There are a bunch of rotations going on which are magnifying the story here.

[Smorgasbord1]

Note that APP reports in a week, so I expect that explains much of the quietness from the company.

[monkeydluffy]

Here’s my understanding:

First of all, we need to know the difference between an ad mediation system and an ad network / ad bidder, and how they help each other. An ad bidder will take many user signals as input and output an ad with a bid. And a mediation system can request multiple ad bidder at the same time and choose an ad with the highest bid, so it can maximize app developer’s return.

MAX is AppLovin’s ad mediation system while AppDiscovery is its ad bidder. AppDiscovery uses AXON 2.0 as the recommender engine. App developers who use MAX can configure which ad bidder (ad network) to enable. So MAX will send ad request to many different ad bidders. If an ad from AppDiscovery wins the auction, AppLovin earns the revenue. If ad from a different ad network (e.g. Google or Meta’s), AppLovin charges nothing, except that for some DSP ad providers, AppLovin does charge a small portion of fee.

It seems that AppLovin is providing MAX “for free”, but the real benefit is that, by providing this mediation SDK, AppLovin can collect more useful training data to make their AXON engine more powerful. It can collect raw user signals as well as the bid from other ad network. With the raw user signal data, AppDiscovery will also be able to provide more accurate recommendation, compared to external ad networks whose data access is relatively limited. This is like the “home advantage” in football games and, with MAX, AppLovin is both player and referee.

Now, let’s consider the impact from a new LLM based mediation system.

An app usually only uses one ad mediation SDK which could be MAX or the existing competitors (e.g. AdMob from Google). If a new mediation SDK comes to take the market share, it’s definitely not good for AppLovin. Although AppLovin does not earn money directly from mediation, it will lose the “home advantage“ if app developer switches to other SDK and the AXON engine could degrade if AppLovin lose access to some user data.

That being said, I do think the current situation is an over reaction to this stock. An LLM does not necessarily make a recommender system better than those using traditional ML models. AppLovin’s data is its moat, especially in the gaming category. This is why AdMob is the most used SDK overall but AppLovin has a much bigger market share than AdMob for gaming apps. Even Google can not beat AppLovin in its domain. An LLM new comer won’t, either.

App developers will run AB tests to try different mediation SDKs, but, at the end of day, they will only choose the one with the highest eCPM.

Luffy

[brittlerock]

@monkeydluffy OK, it appears that you have a better grasp of how the pieces fit together than I, but the way I see it, the preference of the app developer (content provider) will be driven by where the advertisers wish to spend their advertising dollars. If AXON consistnetly provides the highest ROAS, that’s where the advertisers are going to spend their advertising budget.

I am not 100% confident about this, so correct me if I am wrong, but I don’t think that the service that brings the content provider and the advertiser together must be MAX in order for AppLovin to make money. If AXON is the vehicle for targeting the ad AppLovin earns a fee. Are AXON and MAX inseparable? Must the transaction be executed by MAX - I don’t think so. In fact, the new ecommerce vertical that AppLovin has just opened for their advertisers does not use MAX by default. That is true with mobile games, but I don’t believe that is so with ecommerce and most likely won’t be true with CTV either.

Separately, I don’t think MAX is a free service which benefits AppLovin only by virtue of the data capture. I am quite certain that every transaction executed on the MAX platform results in a small commission paid to AppLovin. While the commission is small, the volume is enormous so the commsion revenue is non-trivial.

[monkeydluffy]

@brittlerock

I don’t think that the service that brings the content provider and the advertiser together must be MAX in order for AppLovin to make money. If AXON is the vehicle for targeting the ad AppLovin earns a fee. Are AXON and MAX inseparable?

I believe you were right that advertisers will prefer the platforms that offer the best ROAS. But here’s the logic under the hood: In order to deliver high ROAS, AXON, the recommender engine needs to be able to predict ad conversion or ad clicks accurately. The more accurately it predicts, the higher ROAS it can deliver. MAX collects all data passed to it by the app, even including data that app developer would rather not share to AppLovin and just wants to pass to Facebook or Google. This data is extremely valuable training data that makes AXON smarter. If app developers switch to other mediation SDK, AppLovin could loss meaningful training data. The loss of training data will hurt AXON’s accuracy.

I am quite certain that every transaction executed on the MAX platform results in a small commission paid to AppLovin.

I asked Gemini and it says AppLovin does not charge any commission for SDKs integrated with adapter from MAX. These SDKs include the major demand side competitors like AdMob from Google and FAN from Meta. The app developer gets directly paid in this case and can check their payment on Google and Meta platforms. The commission you were referring to might be other DSPs that do not have an integrated SDK (e.g. The Trade Desk, etc) AppLovin does charge 5% commission in this case. Please correct me if you find more credible source since I got this information all from AI.

In addition, I did a little more research on CloudX and I do feel it’s something every AppLovin investor should pay attention to and dig deeper. It’s app developers who choose which SSP SDK to use to monetize their inventory. They do not care about ROAS, but they care about how much money they can earn from the inventory. Here’s how CloudX can compete with MAX and other SSPs.

1. A key pain point from app developers who use MAX is that they do not want to share certain data with MAX in fear of AppLovin using that data to benefit their competitors. For example, your App is App A. “user X just made in-app purchase of $100 in your app“ is extremely helpful signal for the ads models. If app developers share this with MAX, AppLovin will store this data in their user profile linked to a device ID. There’s another app App B who is direct competitor to your app and also wants to acquire users willing to pay. In another day, this same user opens App C, which happen to use MAX as well. The user may receive a accurately targeted ad to promote installing App B. You will be risking losing this user since App B is a direct alternative app. CloudX solves this pain point because it offers a trusted execution environment (TEE), which guarantee that CloudX does not collect any user data and just does pass-though. With CloudX, you can configure which DSP can receive information such as ““user X just made in-app purchase of $100 in your app“. You may feel comfortable sharing this info with Google who end up giving you higher return.
2. Traditionally, app developers allocate additional head counts for ad operation. CloudX uses Agentic AI to automate this approach and eliminate the need of human ad ops. The money saved from this might offset any potential loss on CPM.
3. As I said, with MAX and AppDiscovery, AppLovin is both a player and a referee. App developers worry that there could be “hidden fees“ caused by this information asymmetry. As a neutral SSP, CloudX’s more transparent and may only charge a very small tech fee.
4. The founder of CloudX, Jim Payne, was the founder of both MoPub and MAX. MoPub was acquired by Twitter then sold to AppLovin. MAX was acquired by AppLovin as we all know. So essentially, this founder knows everything about MAX and the market seems to believe that he has the right context to disrupt MAX.

Luffy

[monkeydluffy]

Sharing some more perspectives:

Morgan Stanley just reiterated their buy rating on $APP with $800 price target.

And an mobile ad tech expert I follow on X said CloudX won’t have big impact on AppLovin.

Luffy

[gmcnatt]

@monkeydluffy

The founder of CloudX, Jim Payne, was the founder of both MoPub and MAX. MoPub was acquired by Twitter then sold to AppLovin. MAX was acquired by AppLovin as we all know. So essentially, this founder knows everything about MAX and the market seems to believe that he has the right context to disrupt MAX.

so maybe Applovin would then just buy CloudX?

[Rudderspin]

I was thinking the same thing - if CloudX is a serious or existential threat, then APP can seek to buy the company. With $1.6B plus in cash on hand they have the resources and have done this twice before from this founder. At this point, it may be cheaper to just hire him to work for APP instead of buying him out every few years. He seems to know their business very well.

Bumped up my 7% position back up to 8% on the dip. The valuation now is very compelling. The stock price has dropped ~40% YTD, bringing the PEG down to around 1.0.