Revenue tiers: what each one costs
$100k → $1B MRR. For each tier: users needed, variable + fixed costs, team, and net margin. Adjust ARPU and active% to stress-test.
$1.42 = two-tier mix (cheap consumer + GPU-priced predict/train). $19 = Pro. $199 = Dedicated. $1500 = Enterprise.
Share of registered users active in a given month.
Upstream infra (Gemma, RunPod, CF) as a share of revenue. 45% is today's ratio.
P&L by revenue tier
| Tier | Users | MAU | MRR | Variable | Fixed | Net | Margin |
|---|---|---|---|---|---|---|---|
| $100k MRR | 352.1k | 70.4k | $100.0k | $45.0k | $15.0k | +$40.0k | 40% |
| $1M MRR | 3.5M | 704.2k | $1.0M | $450.0k | $200.0k | +$350.0k | 35% |
| $10M MRR | 35.2M | 7.0M | $10.0M | $4.5M | $3.5M | +$2.0M | 20% |
| $100M MRR | 352.1M | 70.4M | $100.0M | $45.0M | $30.0M | +$25.0M | 25% |
| $1B MRR | 3.5B | 704.2M | $1.0B | $450.0M | $250.0M | +$300.0M | 30% |
What each tier needs
$100k MRR
3-5 warm GPUs, D1 DB, status page, on-call rotation.
$1M MRR
Multi-region, GPU reservation contract, SOC2, EU data residency.
$10M MRR
Global inference fleet, enterprise tier, procurement motion, dedicated SRE.
$100M MRR
Multi-cloud, platform BU, on-prem option, acquisitions.
$1B MRR
Category leader / IPO-ready. Infra + R&D as % of revenue.
Revenue vs cost across tiers (log-scale)
Critical levers
- ARPU is the primary lever. At today's $1.42, reaching $100M MRR needs 352.1M registered agents. Raising ARPU to $5 (enterprise tier) drops that to 100.0M.
- Variable cost % should fall with scale, not stay flat. GPU reservations beat serverless above ~$100k MRR. Drop the slider to 35% to see the impact.
- Fixed cost escalates non-linearly. $100k MRR = ~$15k fixed (15%). $1B MRR = ~$250M fixed (25%). At hyperscale you pay for R&D and compliance as much as for infra.
Assumptions
Users and MAU are back-computed from the target MRR at the given ARPU and active%. Variable cost is a flat share (it actually improves with scale, which is why the slider exists). Fixed cost bundles infra, team, and platform investment.
Team sizes are rough industry benchmarks for B2B infra companies at each MRR tier. The $1B row assumes public-company-grade R&D spend (~25% of revenue) — below that level, revenue stalls.