The $1M solo SaaS playbook in 2026: workflows > platforms, usage pricing > subscriptions

Been digging into how solo founders are actually hitting $1M ARR without hiring in 2026. A few things stood out enough that I wanted to share:

The stack that keeps showing up: Next.js/SvelteKit on Vercel, Supabase for Postgres/auth/storage, GPT-4o or Claude 3.5 Sonnet via API (rarely fine-tuned), Stripe for usage-based billing. Total infra cost at $50K MRR: $400-1,200/month. No DevOps hire needed.

The bigger shift is pricing, not tech. Flat $49/mo subscriptions cap you at needing 1,700+ customers to hit $1M — not realistic solo. The ones actually getting there use usage-based pricing ($0.10-2.00 per unit processed), which apparently pushes net revenue retention to 127% vs 103% for seat-based models.

Real example from the piece: a founder built a support-ticket auto-tagger with GPT-4, charged $0.08/ticket, ended up at $960K ARR with just 50 clients averaging $1,600/month each.

The failure mode that caps most solo devs below $500K: building a “platform” (multi-role permissions, integrations, extensibility — needs a team) instead of a “workflow” (one input, one output, one job done well).

Full writeup with the pricing math and go-to-market breakdown: newstide.news/million-dollar-one-person-saas-the-…

Curious if this matches what people here are seeing — is usage-based pricing actually taking over, or is that cherry-picked survivorship bias?