What are the stupid projects making a lot of money now?

Are stupid internet projects really making serious money now?

Yes. Stupid-looking internet projects are currently making anywhere from a few thousand dollars to hundreds of thousands, while equally simple consumer apps can reach tens of millions a year.

By “stupid,” we mean projects whose pitch sounds almost embarrassingly basic when said out loud. Pay money to move your website higher on a leaderboard. Put your logo on somebody’s toilet. Buy a few pixels on a webpage. Photograph your lunch and get a calorie estimate. Ask AI what a tarot card means.

The gap between how small these ideas sound and how much money they generate is what makes them interesting.

Weird Revenue’s live tracking currently puts outbid.lol at roughly $253,000. The Million Dollar Toilet has reached $127,500. A modern Million Dollar Homepage clone has passed $30,000. Brand My Mac reached €7,018 within days by auctioning sticker positions on a laptop the creator had not even bought yet.

Recurring apps go much further. MyFitnessPal said Cal AI had already crossed $30 million in annual revenue when it acquired the company earlier this year. TrustMRR currently verifies an AI fortune-telling app at more than $80,000 over its latest 30 days and a Hyrox training app at more than $30,000.

So yes, there is real money here. The more interesting question is why one stupid idea makes $250,000 while 400 people copying it make almost nothing.

Project	What people pay for	Recent revenue
outbid.lol	A higher position on a leaderboard	~$253K
Million Dollar Toilet	Advertising on a real toilet	$127.5K
1 Million Pixels	Pixels on a webpage	$30.8K
Brand My Mac	Logo space on a laptop	€7K
Pay for No Reason	Literally paying for no product	$5K

Is outbid.lol the stupidest project making big money right now?

Probably. outbid.lol turned “pay more than the person above you” into roughly a quarter-million dollars after being built in about three hours.

Jonathan Wilke launched a webpage where companies bid for position. If another company pays more, it moves above you. The payments themselves create the ranking.

There was barely anything else to explain.

The latest clone census puts the original at $252,757. The top position climbed as high as $17,000, while more than 2,000 listings appeared on the original board during its first wave of attention.

The clever part was the behavior created around the leaderboard. Every company pushed down the page suddenly had a reason to talk about outbid.lol. Winners showed off their ranking. Losers complained about being overtaken. New advertisers saw competitors spending thousands and joined the fight.

One advertiser, CrowdReply, publicly broke down what happened after spending $12,700 for the top position. It reported more than 6,500 clicks, roughly 1,800 signups and over 50 booked demos.

That made the ridiculous bid easier for the next advertiser to justify.

The webpage itself was tiny. The bidding fight generated the content, distribution and price escalation.

Why did almost every outbid.lol clone make nothing?

Because copying outbid.lol took hours, while copying the crowd around outbid.lol was practically impossible.

We now have an unusually good dataset for testing this. The latest outoutbid.lol census has found roughly 470 verified pay-to-rank boards. Revenue can be read from 325 of them, and those boards claim $330,485 combined.

As seen above, the original dominates the format. It accounts for 76.5% of all readable revenue.

The remaining 324 measurable clones share only $77,728.

The median clone has earned $20.44.

More than a third have made less than $10. Roughly 79% remain below $100. Thirty-two have recorded zero revenue.

This happened despite extreme copying speed. Twenty-two clone domains appeared within the first day and 189 within two days. During the busiest hour, 20 domains were registered, roughly one every three minutes.

Timing also had a measurable effect. The earliest quarter of dated clones collectively earned $9,561. The latest quarter earned $1,069, even though only about two days separated the whole measured wave.

We rarely get such a clean experiment in internet businesses: hundreds of founders copied nearly the same mechanic at nearly the same time.

The result is brutal. Building the thing explains very little of the money.

Outbid clone result	Latest measured figure
Verified boards found	~470
Boards with readable revenue	325
Original’s share of readable revenue	76.5%
Revenue across 324 measurable clones	$77,728
Median clone revenue	$20.44
Clones below $100	79.3%
Clones at $0	32

How can someone make $127,500 by advertising on a toilet?

The Million Dollar Toilet has genuinely reached $127,500 by turning one bathroom toilet into a sponsorship format people actually want to talk about.

Motion designer Nick Greenawalt projects company logos onto his toilet, films sponsored content around them and places each sponsor on a virtual 3D version of the toilet online.

The highest-paying sponsor gets the “King of the Throne” position.

Weird Revenue’s current tracking shows 25 sponsors paying between $1,000 and $17,000. The first six sponsors spent about $2,417 each on average. The latest six averaged about $8,833.

So pricing increased roughly 3.7 times as the project became more famous.

The revenue curve is also better than we would expect from a one-day joke. Weird Revenue calculates that around 70% of the project’s revenue arrived after its biggest attention peak. The tracked social posts have accumulated about 40 million views.

The toilet gives advertisers something ordinary sponsorship packages struggle to provide: a reason for people to mention the advertisement itself.

A $10,000 rectangle on a normal website sounds expensive. A $10,000 logo projected onto an internet-famous toilet becomes a story.

That difference has been worth six figures so far.

Did someone really make $30,000 by copying the Million Dollar Homepage?

Yes. 1milllionpixels.com revived a 20-year-old pixel-selling idea and has now generated about $30,765.

The site contains one million pixels divided into purchasable blocks. Prices rise progressively as more pixels sell.

For almost three months, the project remained small.

Then the creator spent $42 on an outbid.lol placement.

According to his public account, that $42 sent around 64,000 visitors and produced roughly $29,000 of sales in one day.

That one distribution event generated roughly 94% of the revenue the site has accumulated so far.

A competing pixel project makes the comparison even cleaner. Weird Revenue found ownpx.com using an extremely similar concept around the same period. It reached about $111 while 1milllionpixels.com went above $30,000.

That is roughly a 275-fold difference between two versions of essentially the same old idea.

The pixels clearly helped people understand the project. The traffic source decided the outcome.

Why are people paying for laptop stickers and literally nothing?

People currently pay for absurd internet products when the purchase gives them visibility, status or a small role in the joke.

Pay for No Reason makes this unusually obvious. The website offers payment options while promising essentially nothing in return, apart from the possibility of appearing on its leaderboard.

Its public counter recently reached $5,010.

Two people paid $1,000 each. Those two payments alone represent about 40% of everything the project has collected.

Brand My Mac uses a similar psychological trick with a more tangible product. The creator, Vincent, auctioned ten advertising positions on the back of a MacBook Pro before buying the computer. The goal was simply to make advertisers pay for the laptop.

Within days, 111 bids pushed the total to €7,018, almost three times the cost of the machine. The launch post accumulated about 4.3 million views.

Then somebody copied the concept one day later with Brand My Mac Mini.

That clone made $49.

The contrast is almost comical: one laptop-ad project gets millions of views and thousands of euros; the immediate copy struggles to sell a single cheap position.

People were buying access to a story already spreading through the internet. Remove that story and a sticker on somebody else’s computer becomes far less exciting.

Are viral joke projects actually good businesses?

Usually, no. Viral joke projects can produce excellent cash very quickly, but most have weak reasons for the same customer to keep paying.

The Million Dollar Homepage-style grid eventually fills. Brand My Mac has ten positions and an auction deadline. Pay for No Reason depends heavily on a tiny number of unusually motivated buyers. Pay-to-rank boards can keep accepting money, though bidding slows dramatically once attention disappears or the top price becomes too intimidating.

The Million Dollar Toilet looks better than most because sponsor spending has increased over time and each new advertiser produces another piece of content. Even there, the business still depends on keeping the joke interesting.

Compare that with a calorie app.

Someone can photograph food every day for years. A Hyrox athlete trains repeatedly. A person interested in astrology can request new readings indefinitely. Dating conversations keep producing new messages that need replies.

That repetition changes the economics completely.

A stupid viral site can be an amazing project.

A stupidly simple product connected to a behavior people repeat can become a much bigger company.

Are stupid-looking mobile apps making more money than joke websites?

By a massive margin. The strongest simple consumer apps today operate one or two orders of magnitude above the viral websites we just examined.

Cal AI passed $30 million in annual revenue in under two years, according to figures MyFitnessPal gave TechCrunch when it announced the acquisition.

TrustMRR’s RevenueCat-verified listings currently show an anonymous Hyrox app collecting $30,655 over its latest 30 days and an AI fortune-telling app collecting $80,920.

STFU, an AI app aimed at people flirting online, generated another $10,979 during its latest 30-day period.

None of these products requires a long explanation.

Photograph your food.

Train for Hyrox.

Get a fortune reading.

Figure out what to reply to someone you are flirting with.

That clarity becomes extremely valuable in consumer apps because someone can understand the pitch from a three-second TikTok or App Store screenshot.

Simple app	Core pitch	Recent verified revenue
Cal AI	Photograph food to track calories	$30M+ annual revenue
Project A	AI fortune telling and astrology	$80,920 in 30 days
AI Hyrox app	Train specifically for Hyrox	$30,655 in 30 days
STFU	AI help with digital flirting	$10,979 in 30 days

How did Cal AI make $30 million from taking photos of food?

Cal AI made more than $30 million a year by taking an existing habit people hate and removing most of the annoying work.

Calorie tracking was already a huge market. MyFitnessPal had spent years proving that millions of people wanted to know what they were eating.

The tedious part was logging everything.

Cal AI gave users a much easier entry point: take a photograph and let the app estimate the calories.

The company reached more than 15 million downloads and over $30 million in annual revenue in under two years. MyFitnessPal eventually acquired it while keeping the Cal AI app alive as a separate product.

The team had only seven employees plus a small group of contractors at the time the deal was announced.

There is another detail we found especially interesting. Founder Zach Yadegari has said photo scanning represented only around 30% of calories logged inside the app.

The flashy AI feature brought people through the door, while most actual usage happened inside a broader calorie-tracking product.

That is the interesting part. A stupidly simple feature can do the acquisition work even when the full business underneath it is much more complete.

Can an AI fortune-telling app really make $80,000 in 30 days?

Yes. An anonymous AI fortune-telling app currently verified through RevenueCat generated $80,920 during its latest 30 days.

TrustMRR lists the project as Project A. It offers tarot readings, astrology, clairvoyance and traditional coffee-cup readings through a mobile app.

The numbers are much larger than the concept initially suggests.

Project A reports more than 50 million downloads and around one million active users. Its current subscription MRR is $57,971 from 8,577 active subscriptions.

The founder also says the app makes significant advertising revenue that is missing from the RevenueCat number.

That means the $80,920 figure understates total revenue.

Fortune telling also has unusual economics for an AI product. The customer usually wants an interesting, personalized experience rather than a fact that can be objectively audited for accuracy.

That leaves plenty of room for repeated use.

A user can ask about love today, money tomorrow, a relationship next week and another tarot spread afterward. Each reading creates another reason to open the app.

For all the attention going toward AI productivity tools, a machine-generated tarot reading is currently outperforming a huge number of serious-looking SaaS products. Yeah, really.

Can a Hyrox app really make $30,000 a month?

Yes. A mobile app built specifically around Hyrox currently shows $31,438 in MRR from only 1,761 active subscriptions.

The RevenueCat-verified TrustMRR listing shows $30,655 collected over its latest 30 days.

That works out to roughly $17.40 of 30-day revenue for each active subscription.

Nothing about “fitness app” sounds new. Workout trackers, training plans and digital coaches have existed for years.

Hyrox gives the product a much sharper reason to exist.

Someone preparing for a Hyrox race has a specific event, exercises, distances, performance benchmarks and training schedule in mind. A generic fitness app has to convince that person it can do everything. A Hyrox app only needs to convince them it understands Hyrox.

RevenueCat’s broader subscription data supports the opportunity. Health and Fitness remains one of the strongest consumer subscription categories, including one of the highest median trial-to-paid conversion rates in its latest large-scale report.

The Hyrox app shows something pretty simple: an old product idea can become interesting again when the audience gets much narrower.

Can an AI flirting app really make $11,000 in 30 days?

Yes. STFU generated $10,979 over its latest 30 days by helping people with digital flirting.

TrustMRR currently verifies STFU through RevenueCat and shows 610 active subscriptions, about $6,969 in subscription MRR and roughly 23,000 users.

The founder says there is no ad spend and operating costs remain minimal. Instagram, TikTok and YouTube are the main acquisition channels.

STFU is interesting because ChatGPT can already help somebody write a dating message.

People still pay for a dedicated app.

The product arrives with the exact framing the user wants: help me handle this conversation. There is less prompting, less deciding what the AI should do and less feeling that a general chatbot is overkill for a tiny problem.

The timing helps too. Someone staring at a message from a person they like may care much more about getting an answer in the next thirty seconds than about how technically impressive the software is.

A narrow emotional problem can support surprisingly high willingness to pay.

Did TrustMRR really turn a six-hour leaderboard into $330,000?

Yes. TrustMRR started as a six-hour revenue leaderboard and has now generated just over $300,000 in product revenue plus roughly $30,000 in acquisition fees.

Marc Lou published a detailed update this week explaining what happened during the first nine months.

The first TrustMRR version had basically two columns: total revenue and MRR. Founders connected a read-only Stripe key so their revenue could be verified.

Lou launched it after a discussion about fake startup revenue screenshots. His launch post reached 2.9 million views, and the site quickly sold its initial advertising inventory.

Then traffic cooled.

Instead of treating the leaderboard as the finished product, Lou kept trying small additions. Games, extra pages and directories produced mixed results. The big change came when someone asked whether a startup listed on TrustMRR was for sale.

That led to a marketplace.

TrustMRR now reports 154 founders acquired through the platform, 56,174 registered buyers and more than 2,000 startups listed for sale. Its best month reached $44,000 at a reported 90% margin, with 23 acquisitions completed during that month.

The original stupid idea turned out to be a very good way to gather verified startups and people curious about their revenue.

Once those two groups were in the same place, selling companies became a much more valuable transaction than ranking them.

TrustMRR may be the best example in this article of what happens when somebody follows a stupid project far enough to discover the serious business hiding underneath.

Does vibe coding make stupid projects easier to win with now?

Vibe coding makes stupid projects much easier to build these days, while winning has become harder because everybody else received the same advantage.

RevenueCat’s latest State of Subscription Apps dataset covers more than 115,000 apps and $16 billion in revenue. It shows how dramatically software supply has changed.

Around 2,000 new subscription apps were launching each month in early 2022.

That figure has climbed above 14,700.

So monthly supply has grown more than sevenfold in roughly four years.

The new apps have barely shifted where the money sits so far. Apps launched before 2020 still generate 69% of subscription revenue in RevenueCat’s dataset. Apps launched in 2025 or later account for only 3%.

The odds of reaching meaningful scale remain harsh. Only 4.6% of newly launched subscription apps reach $10,000 in monthly revenue within their first two years.

AI apps also show a fascinating trade-off. RevenueCat finds that AI-powered apps generate 41% more revenue per payer over their first year, while users churn about 30% faster.

The outbid clone wave gives us the tiny-project version of the same story. Hundreds of people could reproduce the original within hours. That flood of supply produced a median clone with twenty dollars of revenue.

Cheap software creation clearly opens markets that once looked too narrow to bother with. It also destroys most of the advantage that came from simply being capable of building the software.

These days, “I can make this” has very little scarcity left.

So what stupid projects are making a lot of money now?

The stupid projects making the most money today tend to be ridiculously easy to explain, attached to an existing desire, and built around a distribution loop or repeat behavior that is much harder to copy than the product itself.

For pure internet absurdity, outbid.lol is still the standout. Paying to move higher on a webpage generated around a quarter-million dollars. The Million Dollar Toilet has turned bathroom ad space into a six-figure sponsorship business. A revived pixel grid crossed $30,000 after one tiny advertising purchase unlocked tens of thousands of visitors.

For durable money, consumer apps are much more impressive. Cal AI made food logging feel as simple as taking a photograph and grew beyond $30 million a year. Project A wraps tarot and astrology around AI and currently brings in more than $80,000 over 30 days before counting its advertising revenue. A Hyrox-specific training app has passed $30,000 MRR. STFU makes five figures by helping people flirt.

TrustMRR sits somewhere between both worlds. It began as a tiny leaderboard, attracted a large founder audience and eventually found a much bigger business in startup acquisitions.

The clone data gives us the warning label. Roughly 470 pay-to-rank boards have already been found. The typical measurable clone made about twenty dollars.

So building something stupid is a terrible strategy by itself.

The much better question is whether the stupid idea has an unfair way to spread, catches somebody at exactly the right moment, removes an annoying step from something they already do, or gives them a reason to come back repeatedly.

When one of those conditions is present, a project that sounds ridiculous for thirty seconds can make more money than software that took two years to build.