There are many posts about how much it costs to start a startup. Almost all are written from San Francisco, assuming five-person teams, shared offices, and seed rounds. Almost none are written from Bogotá, assuming a solo founder, owned home, and zero external investment. This post isn't my receipts — it's the categories that matter and the ones that don't, based on what I've been learning.
The floor: absolutely necessary
There's a set of services you pay every month without exception. The good news for someone starting alone is that almost all of them have generous free tiers at the beginning. In categories:
- Hosting and deploy (Vercel, Fly, Railway). Starts free, scales to a single-developer paid plan when your product hits the limits.
- Database (Postgres on Neon, Supabase, Railway). Same: free up to certain volume.
- Queue and cache (Upstash, Redis Cloud).
- Object storage (Cloudflare R2, AWS S3, Backblaze B2). Cents per month up to TB of traffic.
- Transactional email (Resend, Postmark).
- Payments (Wompi in Colombia, Stripe in markets with presence). No fixed fee, per-transaction commission.
- AI APIs if you use them (Anthropic, OpenAI). The most volatile cost because it tracks real usage.
If you add up the cheapest paid plan from each category, you're looking at a couple of hundred dollars a month for a serious one-product operation. If you take advantage of free tiers in the early months (which is reasonable), you go much lower.
Domains: cheap, but more than it looks
I bought domains on Cloudflare at cost (no markup). Each main domain costs between $10 and $40 per year depending on the extension. Multiply by the number of products you have, add a defensive one if your brand justifies it, and project to 10 years. The bill is still modest, but it's no longer negligible.
The rule: I only register defensives if the brand justifies it. I don't buy a .com for each product if the .co, .io, or .app is already clear. Every extra is a decision, not a default.
Hardware
I started with the laptop I already had. I bought nothing new for many months. When the time came for an additional monitor, I bought it used. No new mechanical keyboard, no iPad, no camera. The rule: if I can do it with what I have, I don't buy. Every expense you avoid at the start is runway you preserve.
What you don't see in stacks
Services almost every founder forgets to list but which add up:
- GitHub Pro for personal Copilot and private repos.
- Family password manager (1Password, Bitwarden).
- Personal Notion (free for one person).
- Linear (free for small projects).
- Figma (free) or Affinity Designer (one-time purchase, no subscription).
- Adobe Creative Cloud: NO. Adobe's annual subscription would cost more than my entire tooling stack combined.
The difference between reasonable tooling and expensive tooling at year-end is hundreds of dollars easy. For a bootstrapped operation, choosing that stack well is as important as choosing your backend language.
Coffee, lunches, and "co-working"
I work from home most of the time. Once a week I go to a neighborhood café. That's my "alternate office". It costs a fraction of what a real coworking costs, and for someone who works alone, it's exactly the amount of socializing I need in a week.
In the US, that same weekly "alternate office" is several times more expensive, and a dedicated coworking runs $250-400 per month. Here, prices for coffee, lunch, and decent connection drop dramatically. It's one of the real and underrecognized advantages of staying in Latam to build.
The US comparison
If I were doing this from San Francisco with the same operation, the software categories would stay roughly the same (SaaS charges the same everywhere). But cost of living changes radically:
- Coffee/co-working: 5-10x more expensive.
- Housing where you can work in peace: multiply by 5 to 10 depending on the city.
- Health: in Colombia, a reasonable health plan costs a fraction of a basic private plan in the US.
On basic cost of living alone, an equivalent operation from Colombia costs less than 20% of the same operation from San Francisco. That's not a detail — it's the factor that decides how much runway you have to find product-market fit without external pressure.
The conclusion that matters
The cost of bootstrapping from Colombia is low compared to the US model, but it isn't zero. And the important thing isn't to minimize every dollar. That turns you stingy, careless about things that do matter (like work tools), and ends up costing you more in frustration than what you save on bills.
What does matter is understanding which categories scale with your product and which are fixed. Cloud infrastructure scales with real usage — it's fine for it to grow when the product grows, that's the signal something is working. Tooling, domains, hardware are fixed costs: worth keeping low at the start to have room to maneuver when something genuinely needs investment.
Distance is asymmetry — and on the money side, the asymmetry favors whoever builds from here.