GM. This is Milk Road PRO, breaking down whether the market's favorite advertising machine is still a buy at 29x earnings.
Today's edition features our latest PRO report on AppLovin, the business quietly running the ads inside roughly 140,000 mobile games and throwing off serious cash while it does. The whole case now rests on one question: can its AI engine repeat that success outside of games, where rivals already sit on much deeper buyer data?
Here's what we've got for you today:
- ⚙️ How it makes money: AppLovin owns MAX, the exchange reaching over 1B daily users, and also bids inside it through AXON, an engine that fires about 2.5M ad requests per second.
- 💰 Why it stands out: an 85% margin, about $1.3B in free cash flow, and over $11B in yearly ad spend running through the platform, more than Pinterest, Snapchat, and Reddit combined.
- 🧠 The make-or-break question: can AXON predict who buys a mattress or an insurance policy as well as it reads gamers, on turf where Meta and Amazon own the richer data.
Securitize is the platform institutions trust to tokenize their assets onchain. See what nine years of regulated tokenization looks like.
Prices as of 2:00 p.m. ET. Powered by CoinGecko.
THIS STOCK GREW 59%, BUT PRICED AT 18% 🤔
AppLovin (APP) is one of the best businesses in public markets right now.
It's the company that runs the ad auction inside roughly 140,000 mobile games, meaning it’s basically how those games make money. It does this using AXON, a platform that decides in real-time which ad fills each slot and who pays for it.
Last quarter, AppLovin grew revenue 59%, ran an 85% adjusted EBITDA margin, and threw off about $1.3B of free cash flow.
This is an exceptional business. The only real debate is what it’s worth.
At roughly $527, APP trades around 29x forward earnings versus my probability-weighted fair value of about $660. That’s enough upside to keep it interesting, but not enough to ignore execution risk.
That risk boils down to one question: can AXON gain adoption outside gaming?
If AppLovin proves it can replicate its success in e-commerce and other verticals, today’s valuation will probably look cheap. If not, investors are paying a premium for an outstanding business whose biggest growth engine is slowing.
And if you want to see exactly what price I’m buying at, you’ll need Milk Road PRO, which is 33% off this week only! Otherwise, let’s dive in.
HOW APPLOVIN MAKES MONEY
The cleanest way to understand APP is to split the business into two parts that both used to hide under the word “gaming”: the supply side, where ads show up, and the demand side, who buys those ads.
Source: Milk Road PRO
Supply is still mostly gaming today: full-screen ad slots inside about 140,000 third-party mobile games.
But when we look at demand, that’s where we see real growth. In June 2026, AppLovin opened up a self-serve part of its platform to every advertiser, not just game studios, largely focused on attracting e-commerce (followed by insurance and fintech).
AppLovin owns two assets and plays two roles in its own market. Think of a stock exchange. AppLovin built and owns the exchange, and it is also one of the traders on it.
The exchange: MAX
MAX is AppLovin’s mediation platform. A developer installs the MAX kit, and MAX runs the real-time auction that decides which ad fills each slot. It reaches over 1B daily active users, making it one of the largest monetization layers in mobile.
Competitors bid inside it: Meta’s Audience Network, Google AdMob, and Unity’s ironSource. And AppLovin ensures neutrality in the process - so much so that they actually sold their own gaming studio back in 2025 to avoid bias.
The trader: AXON
AXON, now branded AppLovin Ads, is the buying side: a proprietary deep-learning system, not an LLM, that predicts each user’s conversion value and bids for advertisers. It processes roughly 2.5M ad requests per second.
Advertisers never touch a single auction. They set a goal, such as a target return on ad spend, a cost per purchaser, or a value per lead, and AXON turns that into millions of per-impression bids automatically.
Why the auction is elegant
When a user finishes a Solitaire level and an ad slot opens, MAX calls for bids.
AXON predicts which advertiser values that specific user the most and submits a single bid on its behalf.
The auction compares AppLovin’s bid against Meta’s, Google’s, and everyone else’s. The highest bid wins. If AppLovin wins, it serves the ad. If it loses, a rival does.
As CEO Adam Foroughi puts it, better prediction wins the auction. If AXON identifies high purchase intent more accurately than competitors, it consistently wins the most valuable impressions.
AppLovin doesn’t own the games. It owns the stream of ad impressions it wins across ~140,000 apps. The moat is the prediction engine and the closed-loop data, not the apps themselves.
Those impressions are valuable. Most are unskippable full-screen, interstitial, or rewarded ads that average more than 35 seconds of attention, far longer than a typical social feed scroll. More than $11B of annual advertiser spend already flows through the platform, more than Pinterest, Snapchat, and Reddit combined.
THE MOAT
The durable core is the exclusive ~1B-DAU audience on MAX plus the closed-loop auction.
Owning the marketplace and trading inside it gives APP three things a pure bidder cannot copy: a toll on every slot it wins, visibility into the auction, which feeds the data edge, and distribution to a billion users a rival would otherwise have to rent.
Source: Milk Road PRO
The flywheel is simple: more advertisers bring more demand density, AXON wins a larger share of slots, AppLovin pays publishers more, more publishers join MAX, that brings more inventory and data, prediction improves, and AXON wins even more slots.
But gaming ad spend tops out around $100B, and AppLovin already captures $11B of it. The flywheel can't grow much faster without new demand from outside gaming, which is exactly what the push into e-commerce, fintech, and insurance is for.
THE RAILS BEHIND BLACKROCK'S ONCHAIN FUND
Assets still run on decades-old infrastructure:
Slow settlement, walled access, paperwork everywhere.
Tokenization fixes that but who’s actually doing it at an institutional scale?
With nearly nine years in the business, Securitize has become the platform institutions trust to bring assets onchain.
Here's why it stands out:
- BlackRock, Morgan Stanley & ARK Invest have all invested in Securitize
- NYSE, VanEck & BNY have chosen Securitize to tokenize their stocks.
- The only public pure-play in tokenization infrastructure
This isn't a crypto project waiting for TradFi to catch up. It's the regulated bridge that the largest asset managers in the world already build on.
See what nine years of regulated tokenization looks like.
THE GROWTH QUESTION
The supply side is capped. There are only so many mobile games and only so much screen time. A company already running an ~$11B ad business cannot grow 59% forever inside a ~$100B market.
The only way to keep the flywheel spinning is to bring new demand to the same inventory. That’s why AppLovin is expanding into e-commerce, fintech, insurance, lead-gen, and connected TV.
The move is attractive because publishers, not advertisers, control ad load. New advertisers don’t create more ads for players. They simply add more bidders to a fixed-supply auction, pushing clearing prices higher. That lets AppLovin generate more revenue from inventory it already owns.
Source: Milk Road PRO
The marketplace side is easy to believe. Whatever an advertiser sells, AppLovin still owns the door to a billion users.
The harder question is the data. AXON became exceptional by predicting behavior inside games. Predicting who buys a mattress, insurance policy, or checking account is a different problem, and competitors like Meta and Amazon own much richer purchase-intent data.
So the entire thesis comes down to one question: can AXON generalize beyond gaming? If it can, today’s valuation is likely too low. If it can’t, AppLovin remains an exceptional gaming-ad business - but not necessarily the next great advertising platform.
History offers reasons for caution. Companies like eBay, Pinterest, Snap, Yelp, Groupon, and Twitter all tried to expand beyond their original strengths and struggled. AppLovin has earned the benefit of the doubt through strong execution, repeatedly building on its existing assets. But whether its data is good enough for e-commerce won’t be answered by a narrative. It will show up in the ROAS (Return on Ad Spend) numbers over the next few quarters.
THE NUMBERS
Q1 showed that this one has legs: 59% revenue growth with the 85% margin intact. But can it sustain?
Source: Milk Road PRO
Q2 guidance is $1.915-1.945B in revenue and $1.615-1.645B in adjusted EBITDA, an 84-85% margin. Full-year consensus sits near $7.9B in revenue and ~$14.75 EPS.
On the balance sheet, cash is ~$2.76B against ~$3.51B long-term debt, so net debt is ~$750M, under 0.5x EBITDA.
The buyback matters more than the leverage.
APP retired ~2.2M shares for ~$1B in Q1 and is running about $1B a quarter, roughly a 2% annual yield on a $190B market cap. That shrinks the share count ~2% a year, puts a soft floor under the stock, and quietly compounds ownership. Time is on your side if the fundamentals hold.
THE REAL AI EDGE
The price of intelligence is collapsing. Model output that cost $100 a year ago now costs a fraction of that.
Most people conclude this simply opens the long tail for AppLovin. That’s true, but not for the reason they think.
APP already runs at ~85% EBITDA, so model costs were never the bottleneck. The real constraints were creating ad creatives and onboarding advertisers. Cheap AI removes both.
Source: Milk Road PRO
On the predictive side, cheaper models don’t widen the moat. Every competitor gets them. The advantage still comes from AppLovin’s proprietary data. Better predictions improve ROAS, which attracts more advertisers, generates more data, and strengthens the flywheel. The models become commodities. The data does not.
The bigger opportunity is generative AI.
For advertisers, AI slashes the cost of creating videos, copy, landing pages, and campaigns. A brand that once needed a creative team can now launch ads in minutes, and AppLovin is building those tools directly into the platform.
For AppLovin, AI makes the long tail profitable. Self-serve was historically uneconomic because onboarding and supporting small advertisers required too much human effort. AI reduces that cost close to zero, making it viable to serve millions of small advertisers instead of thousands of enterprise customers.
That said, cheap intelligence benefits everyone. Meta’s tools improve too, and barriers fall across the industry. AppLovin captures the upside because the parts AI cannot commoditize are the ones it already owns: exclusive inventory through MAX and years of proprietary behavioral data. AI expands the number of advertisers that can cross the moat. The moat itself remains inventory and data.
WHAT IT'S WORTH
Here’s what today’s ~$527 price is really assuming.
A stock price is a bet on future profit growth. Run the math backward, and today’s price implies APP can grow earnings around 17-18% annually over the next five years.
That’s a meaningful slowdown from today. APP is growing profits 50%+, and even consensus expects 25-30% annual growth through 2028. So the market is assuming growth eventually decelerates sharply.
Whether that’s too conservative comes down to one question: does AXON successfully expand beyond gaming? If it does, today’s price could prove cheap. If it doesn’t, the current valuation is probably fair.
Now compare it to businesses the market already respects:
Source: Milk Road PRO
APP carries 3-6x faster growth and higher margins than many peers, yet trades at a similar or lower multiple. The market is giving it very little credit for sustaining that growth.
My scenario fair value:
Source: Milk Road PRO
At ~$527, APP trades below my ~$660 probability-weighted fair value. That’s a real cushion, but not a huge one. Weighting my bull (30%), base (45%), and bear (25%) cases produces roughly 25% expected upside over the next ~15 months.
Ultimately, everything comes down to one variable: how fast AppLovin can compound earnings over the next several years. My scenarios assume 20%, 27%, and 35% CAGR, all of which depend on one thing: whether AXON proves it can generalize beyond gaming. The next few quarters of e-commerce ROAS data will tell us which path the business is actually on.
BULL AND BEAR
What I think goes right:
- AXON is a real AI moat. An 85% EBITDA margin on a scaled ad platform is driven by measurable ROAS, not marketing.
- TAM expansion. Self-serve AXON Ads opened to all global advertisers in June 2026, with e-commerce modeled at ~$1.45B for 2026. If it works beyond gaming, the TAM moves from ~$100B mobile-gaming ads toward $600B+ in digital advertising.
- Growth at scale. 59% YoY on a $7B+ run rate is rare. Most companies this size decelerate hard.
- Capital return. ~$1B a quarter in buybacks is a floor and compounds ownership.
- A scarce combination: marketplace plus data plus a track record of reusing both.
What could go wrong:
- Data-turf risk (the key falsifier). AXON's edge was built on gaming-engagement data. On e-commerce it fights Meta and Amazon's deterministic purchase-intent data. The engine may not win on someone else's turf.
- E-commerce disappointment risks a 20%+ compression toward the low-$400s.
- Privacy and regulation. There is open SEC scrutiny, and further Apple or Google tracking limits could erode the behavioral-data advantage.
- Insider selling. The CEO sold ~$51M in June via discretionary sales, and total insider liquidations topped ~$195M over 90 days.
- Self-serve quality dilution. Opening globally to unvetted advertisers risks fraud, brand-unsafe creative, and degradation of the exact audience that is the asset.
One question worth answering directly: does e-commerce put the core at risk?
Operationally, mostly no. As covered in “The growth question,” it adds bidders to a fixed-supply auction without raising ad load or cannibalizing game-studio budgets, and non-converting ads stop spending on their own.
The catch: “accretive if it works, harmless if it fails” applies to the business, not the stock. If e-commerce visibly flops, gaming is unharmed, but the generalization premium in the multiple deflates. The business has a floor. The stock, at this price, might not.
MY POSITION
APP is a rare business at a reasonable price.
And if you want to see whether I’m ready to make a move, you’ll need to go Milk Road PRO. It’s 33% off until Monday, so don’t wait to get in. It’s gonna be a big year for the next phase of AI, and I plan to get in on the action.
BITE-SIZED COOKIES FOR THE ROAD 🍪
Best crypto loan platforms in 2026. Milk Road pick: Ledn.*
Tom Lee: "The broad economy is going to benefit from all these IPOs because it's a massive wealth unlock. The stock market will do well until the unlocks happen."
Lyn Alden: "My guess is that we're unlikely to reach new all-time highs this year. This year, the base case that I would hope to see is a lack of new bottoms in place."
Joe Lubin: "There is no more high-powered money in the world than Ether." Demand is coming from three directions right now: TradFi, AI, and the native degen crowd.
Which crypto exchanges publish proof of reserves? Bitget has been reporting monthly since December 2022.**
*this is sponsored content. **this is partner content.
See what nine years of regulated tokenization looks like.
MILKY MEMES 🤣
ROADIE REVIEW OF THE DAY 🥛
VITALIK PIC OF THE DAY
