What one more customer costs you
Airtable's Business plan costs $45 per seat per month. Supabase charges $0.00325 per monthly active user after the first hundred thousand. I pulled both numbers off the vendors' own pricing pages this morning, and the distance between them is not a discount war. The two companies are counting different things. One meters people, the other meters work, and that choice decides whether your software gets cheaper or more expensive as it succeeds.
One platform meters people, the other meters work
Airtable's Team plan is $20 per seat per month billed annually, or $24 month to month. Business is $45 and $54. Nothing on that pricing page scales with how hard your app works. It scales with how many humans you let near it. Add a warehouse supervisor who opens the base twice a week and you have added $540 a year.
Supabase Pro starts at $25 a month and includes 100,000 monthly active users, then charges $0.00325 per user beyond that. Cloudflare Workers runs the front of the stack on a $5 monthly minimum with 10 million requests included and $0.30 per additional million. Requests to static assets are free and uncounted. Egress has no line item at all.
I am not claiming these are equivalent products. An Airtable seat buys a grid and an automation engine that someone non-technical can operate on a Tuesday afternoon without filing a ticket. That is real, and for a lot of teams it is worth every dollar. The point is narrower: the meter sits on a different part of the system, and meters compound.
The record ceiling that costs less than a cent
Airtable caps a Business base at 125,000 records, and a Team base at 50,000. The pricing FAQ says plainly what happens next: you keep your data, you get a notice, and you cannot add another record until you upgrade.
Price the same data as storage. 125,000 rows at 500 bytes each is about 62 MB. Supabase Pro includes 8 GB of disk and charges $0.125 per gigabyte past that, so those 62 MB would cost under a cent a month if you were billed by the gigabyte at all, which at that size you are not. The ceiling that pushes a ten person team from $200 a month to $450 a month is sitting on top of roughly eight tenths of a cent of storage.
That gap is the product working as designed. Record limits are a pricing instrument rather than an infrastructure constraint, and they are a fair instrument for a company selling a collaboration tool to departments. They stop being fair to you when your record count is the thing your business does. A booking platform adds rows for a living.
Run the curve out to two thousand customers
Say you want your customers to see their own data. Airtable sells that as Portals, an add-on starting at $150 for 15 guests per month on Business. Two hundred customers needs fourteen blocks of fifteen, so $2,100 a month before you have paid for a single internal seat.
Take the same two hundred people to Supabase Pro plus a Worker: $25 and $5. Two hundred monthly active users is 0.2% of what the plan already includes. Now push to two thousand customers. Airtable's portal line goes to something near $20,100 a month. Supabase and Cloudflare still bill $30, because two thousand is 2% of an allotment you bought at the start.
I have walked people through that table who did not believe it until the second pass, and the reaction is usually the same quiet recalculation. They had been reading the seat price as a software cost while it was behaving like a cost of sale. On Stay World Class we replaced Webflow and Xano with Next.js, NestJS and Supabase, using a four stage ETL that preserved every legacy ID as a PostgreSQL UUID so nothing broke at cutover. Lighthouse went from 55.91 to 91 out of 100 and LCP got 77% faster. Performance was the headline. The flat unit cost was what the finance side brought up a year later.
When the seat price is the right price
If your user count is bounded by something real, seats are honest pricing and you should keep paying them. An internal ops tool for a 25 person company tops out at 25 seats. That is $1,125 a month on Airtable Business forever, and a custom build of the same tool costs more than that in attention alone before anyone opens a laptop. Flat unit economics only pay you back if the units grow.
The same goes for anything where the editing experience is the product. If five coordinators live in a grid all day, reordering columns and firing automations, you are buying the interface, not the database. Rebuilding that interface well is expensive and most teams underestimate it by a factor of three.
The number worth computing is your cost per active user, this month, with the live pricing page open. If it is falling as you grow, stay where you are. If it climbs every time you sign a customer, you are paying a toll on your own success, and that is worth knowing a year before it starts to hurt rather than the week it does.
