Why (Almost) Everything Became a Subscription

Every time I set a price for ResistGate↗ or Amethyst↗, I run into the same contradiction — and it's the same one the whole industry is stuck in.

The subscription economy grew 435% over the last decade, approaching a $1.5 trillion market by 2025. And in the same breath: the average U.S. household cut its paid subscriptions from 4.1 to 2.8 in a single year, and 67% of people cancelled at least one subscription in 2025.

That's the whole game. The economics increasingly demand recurring revenue. Users increasingly resist it. This series is my attempt to hold both truths at once as a solo founder — starting with why the shift happened at all.

The old deal: free forever, figure out money later

The early internet ran on one logic: give everything away, build the audience, monetize later with ads. That was rational at the time. Adoption was exploding, venture money was cheap, and — this is the part people forget — the marginal cost of serving one more user was basically zero. A download costs nothing to duplicate. Attention was the scarce resource, not compute.

That world is gone.

What changed: every user now costs money

Modern products don't live on a CD. They live on cloud infrastructure, where every API call, every database query, every stored file is a line item on a bill. There's a real cost to serving a "free" user, and it has to be recouped somewhere.

For AI products it's brutal. A single capable-model call runs real money and scales directly with usage. OpenAI spent roughly $8 billion in 2025, with inference — not training — as the defining constraint; its 2024 inference bill alone was about $2.3 billion, roughly 15x the cost of training GPT-4.

Even my two little extensions have a per-user cost: the entitlement API, the Postgres database, the servers that verify a license. "Free forever for everyone" isn't free — it's a subsidy, and someone always pays for it.

The math that makes one-time purchases lose

Three numbers explain why the industry abandoned buy-it-once pricing:

• LTV should be at least 3x CAC to be viable.
• Median CAC payback stretched to 18 months in 2024, up from 14 the year before.
• If it takes 18 months just to earn back what you spent acquiring a customer, a one-time purchase mathematically can't get there — you need them to stay.

Subscriptions fix both sides at once: they raise lifetime value (recurring payments plus expansion) and create the retention that makes payback possible. That's also why investors pay a premium — around 11.5x revenue for public SaaS — for recurring revenue: it's predictable and it compounds. Today 27% of B2B SaaS revenue comes from expansion of existing accounts, and expanding an account costs about half what acquiring a new one does.

What this means for a solo founder

I didn't put ResistGate and Amethyst behind Pro because subscriptions are fashionable. I did it because the alternative — free for everyone, ads or nothing — doesn't fund the work, and it quietly pressures you toward the one thing I refuse to do: sell user data. Amethyst is local-first on purpose precisely so I never have to.

But that "2.8 subscriptions and falling" number is the warning label. Recurring revenue is necessary. It is not permission to gate value badly. The rest of this series is about that difference — what to make free, how to price it, and how not to become the company people cancel first.

The lesson: recurring revenue became survival math, not a growth trick. The winners aren't the founders who gate the most features — they're the ones whose product delivers more value the longer you stay.

I'm Orlando. I build browser extensions and AI tools from Venezuela, including ResistGate and Amethyst. This is part 1 of a series working through digital business models from a solo founder's chair — next: what to make free, and what to charge for.