Everyone blames motivation. The real killer has a monthly invoice.
Every developer has a graveyard.
A folder full of half-built apps. A GitHub repository with one commit from fourteen months ago. A Notion page titled “PRODUCT ROADMAP” that has not been opened since the week it was created. You know exactly what these are. You built them in a weekend burst of energy, told one or two friends. Then they quietly stopped working on them. Life got busy. Priorities shifted. The idea “just needed more time.”
That is the story most people tell themselves. Effort. Discipline. Motivation.
Those are not the real reasons.
The real reasons are sitting in your email inbox right now, disguised as invoice notifications you never want to open.
The Conversation Nobody Is Having
Here is what gets discussed in every “how to build a side project” article: picking the right idea, staying motivated, shipping fast, finding your first users. The productivity stack. The waking up at 5am.
Here is what does not get discussed: after you build the thing, you need a server to run it. A domain to point people to it. A database to store anything. An email service to communicate with users. A file storage bucket for uploads. An SSL certificate. Monitoring so you know when it breaks.
And then, once it is live and you need people to actually know it exists, you need marketing. Which in 2026 means video editing. Social media posts. Content that looks good enough to stop someone from scrolling. Everyone will tell you “just use AI” for this. They will not mention that the AI tools people actually use are not free.
Add all of that up. Do it slowly. Write down every line item.
Then think about who in your circle can absorb those costs before making a single dollar in revenue.
That is the real conversation. And nobody is having it because it is uncomfortable. It implies that starting a side project is not just a matter of will. It implies that the playing field is not level. It implies that “just build something” is advice that lands very differently depending on which bank account it lands in.
The Actual Cost of Going Live
Let us be specific. Not hypothetical. Specific.
You have built something. It works on localhost. You want to ship it.
Domain name: A .com domain runs between $10 and $15 per year through Namecheap or Cloudflare Registrar. That is the affordable part. Some premium domains cost hundreds. A .io or .dev domain costs $30 to $50 per year. You need one. Without it, you are sending people to a URL that ends in .railway.app or .vercel.app, which looks unfinished and kills first impressions.
Hosting: The free tier landscape in 2025 and 2026 is real but limited. Vercel’s free tier is excellent for frontend projects, giving you 100GB bandwidth per month with no commercial restrictions on most plan types. Netlify offers similar limits. If your project is a static site or a frontend-only app with an external API, you can genuinely stay on free tiers for a long time.
The problem starts the moment you need a backend. A Node.js server, a Python API, a background job, anything that needs to stay running 24 hours. Railway’s Hobby plan costs $5 per month plus usage. Render’s free tier exists but puts your service to sleep after 15 minutes of inactivity, which means the first user after any quiet period waits 30 to 60 seconds for a cold start. That is fine for a demo. It is not fine for something you want people to actually use.
Fly.io starts at around $3 to $5 per month for a small persistent VM.
DigitalOcean App Platform starts at $5 per month for a basic tier. A VPS (the cheapest, most flexible option) starts at $4 to $6 per month for 1 vCPU and 1GB RAM on DigitalOcean, Hetzner, or Vultr.
This is not expensive in absolute terms. But before your product earns a single ringgit, these are costs that come out of your pocket. Every month. Whether five people use it or zero.
Database: Supabase offers a free tier with 500MB of storage per project. That is genuinely useful for early projects. PlanetScale, Neon, MongoDB Atlas all have free tiers with similar small limits. The catch: on free database tiers, projects often get paused after a period of inactivity, which means your first user gets an error while the database wakes up. Supabase’s free tier pauses projects after one week of inactivity.
The moment you need your database to stay reliably awake, you are looking at $5 to $25 per month for a managed database, depending on the provider and plan.
Email: Resend, Mailgun, Brevo, SendGrid all have free tiers that cover a few hundred to a few thousand emails per month. For early projects these are fine. This one is genuinely free until you scale.
File storage: Cloudflare R2 has a free tier covering 10GB and 10 million Class B operations monthly. Supabase Storage also includes free tier limits. For small projects, this is manageable.
Total before you have a single paying user: Conservatively, $0 if you engineer around every limitation, $10 to $35 per month if you want a setup that does not embarrass you. That range sounds small. But multiply it by the 12 months it might take to find product-market fit. You have just committed $120 to $420 to an idea that has not proven itself.
For someone in Malaysia, the US, Australia, or Germany earning a comfortable salary, that is nothing. For someone fresh out of university, between jobs, or in a country where $35 per month represents a meaningful portion of discretionary income, it is a real barrier. Not a fake one. A real one.
Now Talk About the Marketing Bill
Infrastructure is the first wall. Marketing is the second one. And the marketing wall is where most people in the “just use AI” crowd go quiet.
Here is what building an audience in 2026 actually requires if you want to do it properly.
Video content. Short-form video is the highest-reach organic channel on the internet right now. TikTok, Instagram Reels, YouTube Shorts. If you want people to discover your product without paying for ads, this is where the attention is. Editing a video requires either hours of your own time or a tool to help. CapCut is free for basic editing. DaVinci Resolve is free and professional-grade, but has a steep learning curve. These are genuinely free options that work.
Graphic design. Canva’s free plan gives you access to thousands of templates and covers most basic design needs. It is real, it is usable, it is enough for early-stage content. The free tier limits AI uses to around 50 per month. The Pro plan is $15 per month and unlocks considerably more. For someone doing daily marketing content, the free tier gets constraining fast.
AI writing assistance. ChatGPT’s free tier exists. It is capable. For occasional use, brainstorming, or writing a handful of posts per week, the free version covers most of what you need. ChatGPT Plus costs $20 per month. Claude’s free tier is available. Both free tiers are meaningful, but they have daily limits that kick in when you are trying to produce consistent content at volume.
Social media scheduling. Buffer has a free plan covering three social channels, ten scheduled posts. That is limiting but workable. Publer, Later, similar tools have free tiers too.
SEO tools. Google Search Console and Google Analytics are free. Ahrefs, Semrush, Moz are not. For early-stage projects, the free options are sufficient.
So the honest answer to “is AI free for marketing” is: the free tiers are real and more capable than most people admit. A disciplined builder can market a product using only free tools. But the free tiers have limits, daily caps, feature restrictions. The moment marketing becomes your primary growth channel and you need consistency and volume, you start bumping into those walls regularly. ChatGPT free, Canva free, CapCut free, Buffer free, Google Analytics free. That stack works. It just requires more time per output than the paid versions.
The larger issue is that producing consistent marketing content takes time even with free tools. And time is the resource that people working a full-time job while building a side project have the least of.
The Actual Problem: Two Costs Stacked on Top of Each Other
Here is the real structure of the problem.
Infrastructure costs money before you earn anything. You pay to exist online. You pay for a database to hold your users. You pay for uptime. These are fixed costs that run every month regardless of traction.
Marketing costs time. Even with free tools, producing good content takes hours. Hours you would otherwise spend sleeping, being with family, or doing literally anything else after a full workday.
Most “side project failure” articles blame the builder. Not enough discipline. Not enough consistency. Not enough early morning hustle. What they do not say is that the person who built the project also has a day job, a family or friends they owe time to. Infrastructure costs are also pulling $15 to $35 out of their account every month for something that is not earning anything yet.
When you look at it that way, the question is not “why do most side projects fail?” The question is “why does anyone manage to ship one at all?”
A Quick Collection of Things People Say That Sound Like Advice But Are Not
Before we go further, let us appreciate the greatest hits. The sentences that show up in every comment section, every LinkedIn post, every reply to someone asking how to get started. Delivered with complete sincerity by people who have never once questioned whether they apply universally.
“If you really wanted it, you would find a way.”
Translation: I had savings, a support system, no dependents, a network that gave me my first clients. You do not have those things. Therefore you do not want it enough. The logic is airtight.
“Stop making excuses and just start.”
Start what, exactly? The $5 VPS? The $20 ChatGPT subscription? The Canva Pro for the thumbnails? The domain? The legal entity you need to take payments? “Just start” assumes the starting line is free. It is not. But sure. Just start.
“I built my first product with zero money.”
This one is technically true for a surprising number of people. What they leave out: they built it on their employer’s MacBook Pro, in their rent-covered apartment, during the two hours after dinner that they could afford to spend because they were not also working a second job or raising children alone. “Zero money” and “zero cost” are different things. One of them counts the laptop.
“Wake up at 5am. That is when successful people work.”
The implication being that anyone who has not shipped a product simply did not get up early enough. Not that they worked a closing shift until midnight. Not that their child woke up at 4am. Not that they are in a timezone where 5am means their coworkers in San Francisco are still awake expecting responses. Just. Wake. Up. Earlier.
“Successful people do not watch Netflix.”
I once saw this posted by someone who then shared a twelve-tweet thread about their Peloton ride, their cold plunge, their green smoothie, their morning pages, their dog walk. Activities that collectively represent about three hours of a morning. All of which require time, money, space, equipment, a body that cooperates. But sure. The people who have not shipped their app are the ones watching Netflix.
“You do not need money to start. You just need a laptop and internet.”
A laptop costs $300 to $1,500. Internet in Malaysia costs RM60 to RM200 per month. Electricity runs the whole thing. This is not free. The person who says it owns all three already and stopped noticing the cost years ago.
“I had nothing when I started.”
Stated by someone who then reveals, three paragraphs later, that they had a computer science degree, two years of industry experience that gave them the skills to build without hiring, plus a spouse who covered rent for six months. “Nothing” does a lot of heavy lifting in that sentence.
“Opportunities are everywhere. Most people just do not see them.”
Most people see them fine. They are looking at them while calculating whether they can afford to try. That is a different problem. It photographs worse on Instagram, so it does not make the carousel.
“If I can do it, anyone can.”
The most democratically phrased way of not understanding survivorship bias. Said by someone who did it, to the vast group of people who did not, as though the shared outcome of “human with a laptop” was the only variable in the equation.
The sentences above are not said by bad people. Most of them are said by people who genuinely want to help and genuinely believe what they are saying. The problem is that every single one of them was written from a position that solved the money problem before it asked the discipline question. They have confused their own finishing line with everyone else’s starting line.
The Class Problem Hidden Inside Every Build-in-Public Post
This is the part that genuinely bothers me.
The “indie hacker” community, the build-in-public Twitter threads, the “I quit my job to build a SaaS” LinkedIn posts: they are almost universally written by people who could absorb the early costs. Someone with savings. Someone with a partner who earns income. Someone in a country where $200 per month in infra and tools is genuinely pocket change. Someone whose family situation means a failed side project is a learning experience, not a financial setback.
Those stories get shared. Those stories go viral. Those stories become the template for what “building a side project” looks like.
The person who killed their idea after month two because the database bill hit on the same week as an unexpected car repair does not write a Medium article about it. They just stop. The idea goes back into the folder. The GitHub repo stays at one commit.
This is not to say that money is the only reason projects die. Plenty of well-funded projects fail too. But the success stories that circulate are systematically biased toward people for whom money was not the constraint. Which makes the advice that comes from those stories systematically useless for the people it most needs to reach.
What Actually Works With Close to Zero Budget
None of this means giving up. It means being clear-eyed about the constraints and building a strategy that fits inside them.
Start with free infrastructure and use it honestly. Vercel free tier for frontend. Supabase free tier for database. Cloudflare for DNS and SSL. Resend for email. You can get a real product live for the cost of a domain ($10 to $15 per year). The limitations are real: Supabase projects pause after a week of inactivity, Vercel’s free tier is non-commercial for certain use cases, Railway’s “free” tier is actually a one-time $5 credit. Understand these limits. Engineer around them. Do not pretend they do not exist.
Use a $5 to $6 VPS if you need persistent backend. Hetzner is the cheapest reputable provider in Europe. For $4 to $6 per month you get a real server with 2GB RAM, persistent uptime, no sleep timeouts. You manage it yourself, which means learning enough Linux to deploy a Docker container. That knowledge is not optional anymore for anyone building backend software seriously.
For marketing, use free tools but compress the time cost. Do not produce video and written content and graphics simultaneously if you cannot sustain all three. Pick one channel. Post consistently on that one channel for 90 days before adding another. One well-maintained TikTok account or one LinkedIn presence built with CapCut and Canva free beats three half-maintained channels every time.
Do not pay for ChatGPT Plus until you are earning money from the project. The free tier is genuinely capable for the volume of content a person working part-time on a side project needs. Same for Canva. The upgrade is worth it once you are generating revenue and the time savings become measurable. Before that, it is overhead.
Validate before you build anything that costs money. Put up a landing page with a waitlist using Carrd (free for basic sites) or a Notion page with a Typeform. Tell people what you are building. See if they give you their email before you spend a dollar on infrastructure. If nobody signs up, you have saved months of work and hundreds in hosting bills. If 50 people sign up, you have evidence worth spending money on.
Build smaller. The instinct is to build the full product. The smarter move is to build the smallest version that proves the idea works. A spreadsheet with manual processes running underneath it, sold as a tool. A newsletter before a community platform. A service before a SaaS. The service generates revenue immediately. The SaaS generates revenue eventually.
The Real Answer to “How Do People With Less Money Grow?”
Slowly. Honestly. With fewer tools than the tutorials suggest.
There is no version of this that pretends the system is designed for you if you are starting from a small budget. It is not. The tools are built for people who can afford them. The success stories are told by people who could absorb early failure. The advice is calibrated for someone with a financial cushion.
What is available to you if you do not have that cushion is time, specificity, the competitive advantage of building something close enough to your own problem that you do not need market research to know if it will work.
Build the thing you yourself desperately need. Tell the specific people who also desperately need it where to find it. Do not spend money on infrastructure until someone has given you money first. Use free tiers as runways, not permanent homes.
Pieter Levels, one of the most cited indie hackers in the world, built Nomad List in 2014 starting as a public Google Sheet. No server costs. No database bill. A spreadsheet with manually entered data, shared with the people who needed it. It eventually became a platform earning over $1.5 million in annual revenue. The starting point was a document that cost nothing to make.
That is not an accident. That is a strategy. It is also, notably, more sustainable than the “spin up a VPS, buy a domain, set up a database, subscribe to ChatGPT Plus, start a TikTok, post daily” plan that every side hustle tutorial implicitly assumes is the baseline.
The baseline needs to be rebuilt from a smaller assumption. What can you build today, this week, that costs you nothing but time and proves the idea works before a single dollar leaves your account?
Start there. Grow from there.
The infrastructure bill comes later, once the idea has already proven it is worth paying for.