Dora NodaNetlify's April 2026 repricing gave every team unlimited seats for a flat $20 — and doubled the two meters that matter most. Bandwidth went from 10 to 20 credits per GB, compute from 5 to 10 credits per GB-hour. Netlify says 98% of customers will see a flat-or-lower bill, and for most teams that is probably true. But if you run a traffic-heavy app, you are shopping in the 2% aisle now, so here is the line-by-line answer first:
| Monthly usage profile | Credits burned | What you pay on Pro | Same traffic on owned hardware |
|---|---|---|---|
| 150 GB bandwidth, light compute | ~3,000 — fits the included allotment | $20 flat | One ~$60 box, 99% idle |
| 500 GB bandwidth, 200 GB-hours compute, 5M requests, 60 deploys | 13,900 (10,900 over) | ~$100 (base + 8 auto-recharge packs) | Same ~$60 box, still bored |
| 2 TB bandwidth, heavier compute | ~47,000 (44,000 over) | ~$320 (base + 30 packs) | Same ~$60 box, finally earning its keep |
The crossover is embarrassingly low: somewhere around 300–400 GB of monthly transfer, a single Hetzner-class dedicated box starts beating Netlify's meter on raw dollars — before you count the compute. The rest of this post is the evidence behind that table: what changed, the credit-by-credit math, who the 2% actually are, and the honest caveats in both directions.
What actually changed on April 14, 2026
Start with the part Netlify led with, because it is genuinely good news for teams. Pro used to cost $20 per seat — every member and every Git contributor added another $20 a month. Now Pro costs $20 flat with unlimited members. A five-person team saves $80 a month before usage enters the picture, and Netlify framed the whole move as pricing for "3 billion builders" in the age of AI coding agents, where seat-counting makes no sense because half your contributors might not be human.
The meter changes shipped in the same announcement, and they cut the other way:
| Meter | Before | After | USD at the Pro pack rate |
|---|---|---|---|
| Bandwidth | 10 credits/GB | 20 credits/GB | $0.13/GB |
| Compute (functions + agent runs) | 5 credits/GB-hour | 10 credits/GB-hour | $0.07/GB-hour |
| Web requests | 3 credits/10k | 2 credits/10k | $0.01/10k |
| Production deploys | 15 credits each | 15 credits each | $0.10 each |
| Form submissions | 1 credit each | Free | $0 |
Pro still includes 3,000 credits a month — worth exactly $20 at the pack rate, which means the included allotment is 150 GB of bandwidth and nothing else. Burn through it and auto-recharge tops you up in $10 batches of 1,500 credits (Personal plans recharge in $5 batches of 500, a worse per-credit rate). Recharge is opt-in: you enable it, and Netlify bills your saved card in pack increments as the balance drains. On the Free plan's 300 credits there is no pack to buy — exhausted meters pause your sites until the next cycle.
Netlify's own summary of who wins and who loses is admirably blunt: the company modeled the change so 98% of customers pay the same or less, and "for the 2% of high-bandwidth or compute-heavy users whose consumption costs outpace the seat savings: we'll be reaching out to you directly." That sentence is doing a lot of work. It tells you exactly which variable decides your bill now — not headcount, but gigabytes.
The line-by-line math
Take a concrete traffic-heavy app: a marketing site plus docs plus a busy blog, mostly static, with edge functions handling search, personalization, and a few API routes. Say it moves 500 GB a month, burns 200 GB-hours of function compute, serves 5 million requests, and ships 60 production deploys (about two a day plus previews that graduate). Here is the April-2026 meter applied line by line:
| Line item | Usage × rate | Credits |
|---|---|---|
| Bandwidth | 500 GB × 20 | 10,000 |
| Compute | 200 GB-hours × 10 | 2,000 |
| Web requests | 5M ÷ 10k × 2 | 1,000 |
| Production deploys | 60 × 15 | 900 |
| Total | 13,900 | |
| Included with Pro | −3,000 | |
| Overage | 10,900 |
Overage buys in 1,500-credit packs, and 10,900 ÷ 1,500 rounds up to 8 packs: $80 on top of the $20 base, for about $100 a month. Note the quantization effect — 8 packs deliver 12,000 credits against 10,900 burned, so roughly $7 of the charge is pack granularity: you buy in $10 steps whether you need the whole step or not. At 2 TB the same arithmetic gives roughly 40,000 bandwidth credits plus ~7,000 for compute, requests, and deploys, minus 3,000 included: 44,000 over, 30 packs, about $320 a month. Bandwidth is ~85% of the bill in both cases; everything else is rounding.
Now run the old-versus-new comparison the way a team lead should, because the seat savings are real money. Under the old meters, that 500 GB workload burned about 8,400 credits (bandwidth at the old 10/GB, compute at 5/GB-hour, requests slightly pricier at 3/10k). The meter doubling adds roughly $37 a month at the pack rate.
A five-seat team saves $80 in seats, so it still comes out ~$43 ahead — a card-carrying member of the 98%. A solo dev with the same traffic saves $0 in seats and eats the full $37 increase. Headcount is the shock absorber. Big team plus modest traffic: you won. Small team plus heavy traffic: check your dashboard before you celebrate.
The 2% test
Netlify named the losers precisely — high-bandwidth or compute-heavy — so the test for whether you are one is equally precise. Pull your last three months of bandwidth and function usage, multiply by the new rates, subtract 3,000 included credits, and compare the overage against your old seat bill. Three profiles land in the 2% almost by definition:
- Media-adjacent static sites. Image-heavy marketing pages, video posters, downloadable assets, and docs with embedded media chew bandwidth without touching compute. A docs site doing 800 GB a month in pageviews and asset downloads was background noise under per-seat pricing; under the new meters it is 16,000 credits before a single function runs.
- Edge-function workloads with real traffic. Personalization, A/B testing, auth checks, and search-as-you-type at the edge all bill compute per GB-hour and bandwidth on the way out. This is the workload Netlify most wants to host — it is also the one the doubled meters tax twice.
- Agent-built, agent-operated apps. Netlify's own pitch is that AI agents are the new builders, and agents deploy often, invoke functions constantly, and never sleep. Fifteen credits a deploy and 10 per GB-hour add up fast when your contributor ships forty times a day and your support agent polls its own endpoints.
The community reaction has been exactly what you would expect from a meter-doubling sold as a price cut. The Hacker News thread on the announcement reads like a group invoice review, and at least one viral teardown ("when cloud rent comes due") frames credits as the industry's next extraction layer after subscriptions. Some of that is theater — per-seat billing was its own racket, and nobody mourns the Git-contributor seat. But the structural complaint is fair: seats were at least predictable, while credit meters convert every traffic spike into a billing event. Budgeting used to be headcount times twenty dollars. Now it is a function of your CDN graph.
The same workload on owned hardware
Here is the other column of the ledger. A Hetzner-class dedicated box — Ryzen-grade CPU, 64 GB of RAM, NVMe storage — rents for roughly $50–65 a month after 2026's DRAM-driven price bumps, on a gigabit uplink with around 20 TB of included transfer. Amortize the box over its included bandwidth and the marginal gigabyte costs about three-tenths of a cent. Netlify charges thirteen cents. That is a ~40x per-gigabyte multiple, and it means the 500 GB workload from the previous section consumes 2.5% of one box's monthly transfer budget while costing $100 on the meter.
This is the part where honest accounting demands the caveats, because raw bandwidth multiples flatter self-hosting and everyone knows it. The $60 box does not come with a global CDN, DDoS absorption, atomic deploys with instant rollback, preview environments per pull request, or somebody else's pager going off at 3 a.m. Price a weekend of your own ops time at consulting rates and Netlify wins back several months of meter in one incident you never have. For bursty traffic — launch days, Hacker News hugs, seasonal retail — the meter is also genuinely the right shape: you rent the spike instead of provisioning for it.
But notice what the caveats all have in common: they are about operations, not capacity. None of them disputes that the underlying compute and transfer cost dollars while the meter charges tens of dollars. The question was never whether Netlify's platform is worth something — it plainly is. The question is whether it is worth a 5–10x markup at your specific traffic level, and the April repricing moved that break-even point down to a few hundred gigabytes. Teams that crossed it used to need real scale to justify leaving. Now a busy blog qualifies.
The verdict
Netlify's trade is coherent on its own terms: seats were friction for exactly the agent-assisted, everybody-builds future the company is betting on, and funding unlimited seats by doubling usage meters is the obvious way to pay for it. Most customers genuinely benefit — any team large enough that seat savings outweigh meter growth, which is most teams with normal traffic. The 98% claim deserves its hearing, not just its skepticism.
But averages are not bills. If your app is traffic-heavy and your team is small — the indie hacker with a hit, the docs site with real readership, the agent-built service that never sleeps — run your own numbers before assuming you are in the 98%. Pull three months of bandwidth, multiply by $0.13, add compute at $0.07 per GB-hour, and compare against a box you own. Somewhere around 300–400 GB a month, the answer flips, and it flips hard: past that line you are not paying for a platform anymore, you are paying a bandwidth tax with a platform attached.
Running the numbers and not liking them? Bex.co is the open-source, AI-native Render alternative — push a git repo, get a running HTTPS service on machines you own, where bandwidth ships in bulk with the box instead of by the credit. Star the repo on GitHub or deploy your first app today.
