# why I had to shut down slate (it cost too much money)

You know that feeling when you pour months into building something, get it working exactly how you want, then reality comes crashing down when you realize you can't afford to keep it alive? Yeah, that was Slate. Classic high school startup founder move.

I spent way too many months building a job platform that was supposed to fix all of LinkedIn's problems. No paywalls, no AI-generated spam posts, just clean tools for finding actual work. I got the authentication working, built out the resume editor, set up the dashboard. For a moment there, it felt legit. Like, actually legit.

This was all happening between Scout meetings (I'm the SPL of my troop), mountain bike practice, AP Physics homework, and French class. As anyone who knows me knows, I'm a world-class procrastinator, so most of this got built at 2am when I should've been sleeping or doing actual homework. But building felt more important than conjugating French verbs, which probably tells you something about my priorities.

Then I sat down and actually crunched the numbers on what it would cost to run it for real. Ouch. Reality hit harder than my AP Physics midterm.

# The Numbers Everyone Ignores

When you start building web apps, everyone talks about AWS free tier and Vercel's hobby plan. What they don't tell you is just how quickly you outgrow those free options. Like, way quicker than you'd think.

Here's what running Slate would actually cost me each month (the brutal reality check):

Backend hosting (AWS):

EC2 t3.micro instance: ~$8/month (barely enough power, but it works)
RDS PostgreSQL db.t3.micro: ~$15/month (if you want real database reliability)
S3 storage for resumes/files: ~$2-5/month depending on how many people actually use it
Data transfer out: another $5-10/month once you actually get users (the "if you build it, they will come" tax)

Frontend hosting (Vercel):

Hobby plan works until you need more than 100GB bandwidth (which you'll hit faster than expected)
Pro plan: $20/month per member (just me, but still)
Serverless function costs add up really fast with actual API usage

Total minimum monthly: ~$50-60

And that's if everything goes perfectly and you stay tiny. Add monitoring tools, email services for notifications, and anything beyond basic functionality, and you're looking at $75-100/month easily. This is all before you even have users. Once people start actually using it, costs go up based on usage.

# Why $50/Month Is Actually Impossible For Me

I'm in high school, remember? I don't have $50 to throw at a project every month, especially one that doesn't have a single user yet. That's actually a ton of money when you're 17.

We're talking $600 a year here. To put that in perspective:

That's about 60 hours of minimum wage work (which I don't really have time for with Scouts, biking, and school)
Roughly two months of gas money (I drive a lot for Scout events and mountain bike races)
Most of what I saved from working at summer camp (seriously, my actual savings)

I'm already juggling being SPL of my Scout troop (the responsibility part is real), mountain bike team, AP Physics (don't get me started on the labs), and French class. Adding a $50/month expense for something that might not work at all? That's just not happening.

The math was brutal: to run Slate, I'd need either users paying me money (which ruins the whole "free platform" concept) or find some other way to make money to subsidize it. Maybe ads? But then it's not free anymore and becomes just another LinkedIn clone with ads.

I had neither option. So shutdown it was.

# The "Free" Options I Looked Into

Free hosting providers (the desperate hunt for free resources):

Heroku's free tier: RIP, they killed it (thanks, Heroku)
Railway gives you $5 credit but that's maybe a week of runtime (not sustainable at all)
Render's free tier sleeps after inactivity (terrible UX - users hate apps that take forever to wake up)
Netlify/Vercel are free for frontend, but you still need backend hosting (meaning more costs)

Possibilities I considered:

Google Sheets as a database (weird but functional - I actually used this for CampusLink initially, it's janky but works)
Supabase free tier (generous limits, but you'll eventually hit them)
PlanetScale's free tier (was promising but they discontinued it, because of course they did)

The real problem with all free-tier approaches: you're always one policy change away from your entire project breaking. These companies change their "free" terms constantly. You end up constantly migrating or accepting that your project has an expiration date. Big tech giveth, big tech taketh away.

# What I'd Do If I Actually Had Money

If I had, say, $200-300 to actually get Slate off the ground properly, here's the smarter approach I'd take:

Months 1-2: Stay lean as possible

Vercel for frontend (free, as long as you stay under usage limits)
Supabase for database + auth + storage (free tier works initially, just watch those limits)
Focus entirely on getting 50-100 real beta users (probably by hanging up flyers everywhere and doing old-school "door-to-door" outreach)
Total: $0 (the dream scenario)

Months 3-4: Prove it's worth paying for before scaling

If users actually stick around and engage, upgrade Supabase: ~$25/month
Keep Vercel on free as long as possible
Add basic analytics to see what people actually use
Total: ~$30/month

Months 5-6: Scale up only if there's real traction

If we have 100+ active, engaged users, move to proper production hosting
AWS with proper database setup: ~$60/month
Email service for notifications: ~$10/month
Monitoring tools so I know when things break: ~$15/month
Total: ~$85/month

The key difference: I would've spent that first $50-60 proving people actually wanted Slate before committing to ongoing monthly costs. Instead, I built the whole thing first, then realized I couldn't afford to let anyone actually use it. Classic mistake - build everything first, then figure out if anyone wants it. Super smart, Kadin.

# The Actual Takeaway

Let's be clear: Slate didn't fail because of hosting costs. It failed because I built something expensive before finding out if anyone actually wanted it. The cost breakdown was just the harsh reality check that made me face facts.

The real problem was that I spent months building in isolation instead of talking to potential users from day one. The monthly hosting bill would come due whether I had 0 users or 100. And I definitely wasn't going to get 100 users without actually running the platform for people to use. The classic startup chicken-and-egg problem, except the chicken costs $600 a year and I'm broke. Can't exactly play that game.

# What Actually Works (Proven Example)

After killing Slate, I built MarkSite (the static site generator that's powering this blog right now). Total hosting cost? Zero dollars. It's just static files on GitHub Pages. No servers, no databases, no ongoing costs beyond my time.

Here's the thing that messed with my head: MarkSite has been way more useful to me than Slate ever was. It's simpler, costs nothing, and I actually use it regularly. Sometimes the best projects are the ones that don't require a credit card (which I don't even have - it's just a debit card with about $50 on it).

Building Slate, though it failed, definitely taught me to think way harder about infrastructure costs before committing to a specific architecture. Could I have built a job platform as a static site with client-side features? Probably not the full vision I had. But maybe a simpler MVP that wouldn't require a database? Maybe something smaller and more focused?

These are questions I should've been asking in week 1, not month 3. 🤦‍♂️ Live and learn, I guess.

The $120 hosting bill might not sound like much, but when you're 17 and that's most of your summer camp savings, it actually hurts. My mom asked why I was spending money on "computer stuff" and I couldn't come up with a good answer that made sense to her (or honestly, to me).

On the bright side, I probably learned more about building sustainable products from failing with Slate than I would've from it succeeding. That's what I keep telling myself anyway. Makes that $120 sting a bit less.