The Story
After selling DuckMath for $120K, I knew the next thing had to be bigger. DuckMath taught me that students on school Chromebooks are an underserved audience with real demand. The problem was that browser games only go so far - students wanted to play real games like Roblox, Clash Royale, and Toca Boca, but those require native apps that school Chromebooks can't install.
The solution: stream the games from the cloud. Run Android games on remote servers and pipe the video to any browser. No downloads, no installs, no hardware requirements. That's MaddoxCloud.
I built it in about a month of 10-hour days and officially launched in December 2025.
By The Numbers
How It Works
MaddoxCloud runs Android emulator instances on remote servers. When a user picks a game, the server launches an Android instance with that game, captures the video and audio output, and streams it to the user's browser in real-time via WebRTC. The user's keyboard, mouse, and touch inputs get relayed back to the server, so it feels like the game is running locally.
The result: you can play Roblox, Clash Royale, or any Android game on a school Chromebook, a crappy laptop, or even a phone browser. All you need is an internet connection.
If you want the gory detail of how a system like this maps a mouse onto a touchscreen, I took a competitor's client apart line by line in how CloudMoon turns a keyboard and mouse into Android touches.
Demo
Tech Stack
The infrastructure behind MaddoxCloud is the most complex thing I've built. Here's what powers it:
- Streaming: WebRTC for low-latency video/audio streaming from server to browser and input relay back
- Android Emulation: Server-side Android emulator instances running game APKs on cloud compute
- CDN/Security: Cloudflare for DDoS protection, caching, and global distribution
- Frontend: Web application served through the browser
- SSL: Google Trust Services, domain validated
The hardest part isn't the code - it's the infrastructure scaling. Each user needs their own Android instance running on a server, which means compute costs scale linearly with users.
The Business Model
MaddoxCloud ran on a freemium model. Free users got 10-minute play sessions and waited in a queue when servers were at capacity. Paid users got unlimited play time with no waitlist.
The waitlist was both a blessing and a curse. At peak times, queues hit 12,000+ people. That's incredible demand validation, but it also meant users got frustrated waiting.
Marketing
I applied the same playbook from DuckMath: aggressive short-form video across every platform. TikTok, Snapchat, YouTube Shorts, Instagram Reels. The content targets the same audience - students who want to play games on their school Chromebooks.
One Snapchat post - "Schools tried to ban maddoxcloud.com so I did 15 jumping jacks, destroyed their firewall, and now I'm back in" - pulled 357 likes and showed the kind of viral, relatable content that resonates with students. When schools block the domain, that just becomes more content.
The Exit
MaddoxCloud was sold for $15,000. With 50K users, $500 MRR, and real technical infrastructure behind it, the platform had proven its value. After DuckMath's $120K exit, this was another successful sale - my second project acquisition.
DuckMath proved I could build something millions of people use. MaddoxCloud proved I could do it again with harder tech and harder scaling. Two exits before 21 - on to the next one.