This is becoming really VERY irritating. Sometimes pages load without problems, other times it takes FOREVER! Even if I’m working on a project. works 100% and then suddenly. and to be honest… Bubble is losing creditability with me… I’d rather use AI and build with code…
Grass isn’t greener, you can’t use AI to build a serious app unless you know how to code already.
2 Likes
You sure about that… 
I have just rebuild my agricultural app in React Native using claude. (Yes not a Bubble app but still) I have also build numerous ag systems for companies using ai (next.js), which I wanted to build in bubble, but the unreliability was a concern.
Note I do not know how to code, but I can think through the process logically and build processes 
If you’ve built numerous in the last 6 months (as I’d expect when AI coding got a significant upgrade), it kind of indicates the platforms aren’t that developed and you haven’t had a chance yet to see the walls you’re going to hit.
ok…
Did you select a database (Firebase, superbase, etc?) and hosted (Firebase, vercel, etc) yet? If so, what is your monthly cost so far?
@bcart0v Ive got different db depending in my requirements, from azure sql ($400 pm), azure web app(free) for all my api endpoints, xano and bubble (current bubble users). My customers that I’m busy migrating from bubble will probably go the xano route and vercel hosting. But to be honest, i have not decided on a db & hosting solution. Still looking around. I heard good things about supabase.
2 Likes
I’ve used Supabase for a few small projects, it does work really well. I just didn’t like the price jump from $25 to $599 with no in-between. Also, SOC2 is only available at the 599 plan, and HIPA is an additional fee.
I started a react/next.js project, hosted on Firebase using Fireauth, every fire option I could get. with 1 month of 3 users barely testing my app my bill was $38, So I have yet to revisit that project, I switched it over to Bubble and it’s live and running now.
I’m used to very old, object-oriented compiled languages where the application is hosted on local severs, so I wasn’t sure if that was the normal usage price or what, and I’m fairly certain I could do some enhancements to cut database queries down quite a bit but I just don’t have the time to go back. Anyway I’m rambling! Let me know what your choices are and how it works out!
1 Like
Can you give some examples of what you have build and the complexity of it and how you manage solid Claude returns?
We are busy wiring our vertical product together. Context and code quality becomes better and better but it does take more time than we wanted to get fully i18n, database integration, magic link, cookie, jwt, multi tenant etc capabilities working together like a breeze. Now busy with full infrastructure letting users to get branches in their own otap infrastructure with full undo and roll back. Also working of finalizing our workflow engine and frontend with solid js and tanstack. We are getting there and what we already have surpasses much on what we can find on the market. As soon as you define very clearly what you want, the current output of chatgpt5 and opus 4.1 is nothing short of amazing.
Most, I would say 80%, of our time is lost in making sure we have a design in place that we want and that llm models understand. Therefore we are now building our own automatic flow including review agents using ADR generated documents in JSON style. Very promising.
Would be great to hear your experience!
We are using Hetzner cloud. Talking about dollars. A good old docker or various dockers and it is amazing how performant. We do use Deno btw, flat file if possible or else Postgres or SQlite.
I realized that there is actually for almost all apps no reason to outsource hosting to Amazon and the likes. Start with a single server or two, expend as you grow. With docker, docker swarm or if you really want k8,k3 you can get very very very far
1 Like
hi all, I moved the convo to its own topic to keep things tidy
2 Likes
This topic of bubble vs ai coding is very contradicting
On one side you can build a 5-10k$ mrr app in a month and pay 1% tech (bubble itself) and not touch it again
On the other side (custom), you need a lot of time to put together an app you don’t understand, have no control over and argue with a computer all day. Plus you have to maintain everything and is likely not a solo job so there you go a salary expense out of the box
Just use what works best for the business.
Users don’t mind a little slower app if it does the job. It’s in your head that they care about UI and speed.
In reality, most expensive enterprise software is trash, has bad ui, bad user experience, not that great speed anyway. That’s why it relies on sales people, onboarding, trainings and such.
And with bubble, you can easily outsource heavy data tasks to n8n, supabase, cloudflare workers to save on WU while maintaining speed of development.
6 Likes
The main bottleneck of Bubble today lies in its native backend, particularly in its database layer.
The platform suffers from inefficient queries and high processing costs, especially in scenarios involving relational databases with large tables. Unlike solutions such as Supabase or even traditional SQL databases, Bubble does not allow the creation of custom views or optimized queries (e.g., joins, select fields, refined where clauses). In practice, this forces the application to fetch all records from a table, which directly impacts:
-
High latency on read operations;
-
Limited scalability, as complexity grows linearly with the number of records;
-
Increased costs, since Bubble’s pricing model is based on operations and resource consumption, and the lack of query optimization drives costs unnecessarily higher.
This set of limitations has made Bubble’s backend unviable for medium and large-scale applications. This is not an isolated perception—many developers in the community have already migrated away from Bubble’s backend due to its poor cost/performance ratio.
Furthermore, Bubble’s current focus on AI features does not address these structural issues. While these features may attract top-of-funnel users, the lack of backend improvements severely undermines the experience for those building scalable, production-ready applications.
At present, Bubble is positioning itself more as a frontend-as-a-service platform. However, it does not provide native integration with high-performance databases (Postgres, Supabase, Neon, etc.), forcing developers to rely on workarounds—often fragile and unreliable—just to keep their systems running.
This situation compromises the platform’s long-term technical sustainability. To regain credibility with the developer community, Bubble must:
-
Implement query engine optimizations (e.g., support for views, advanced filters, indexing).
-
Reduce the operational costs of database queries.
-
Offer native integrations with high-performance databases.
Without these improvements, Bubble’s backend will remain a scaling bottleneck, limiting the platform’s adoption in projects that require efficiency and growth.
2 Likes
My personal opinion…
I don’t think you can find a better platform than Bubble to do what it is intended to do.
If you have an app idea, it’s the best place to easily get it up and running and test out your idea.
I don’t think it’s meant to be a global Uber-type building platform where they spend 10s of thousands of dollars a month to operate and have dozens of engineers.
With Bubble, you use their engineers, etc.
I do think you could easily do a taxi service app for your area (example), and even expand it to other nearby areas. As far as going global…if you were at that point, I would hope you would be making enough from the app that it wouldn’t be a hard decision to just go with an in-house team.
I’ve tried the AI builders. Knowing code, my assessment of them is they’re great for a demo app…but, if you get to the point of having multiple concurrent users, you’ll have major problems. The code on most of these AI builders is hacky at best, but works great for a demo.
As far as stability and uptime, I do think that is something that should always be a priority.
Just my thoughts.
5 Likes
This is not entirely correct.. unfortunately it’s not as you say.. but bubble remains bubble for my part.
Wha…? It’s that time of the year again?
I would have agreed with you a year ago, but a lot of badly optimized Bubble apps are just down to bad DB planning and implementation. There’s a lot that you can squeeze out of Bubble. I stack Bubble with other tech and build plugins to get around limitations but believe that youcan scale Bubble apps.
There’s this IMO - childish - perception that scaling in Bubble means just forking out money for an Enterprise plan and/or keeping it all in Bubble. You can and should stack tech for scale. It’s like cooking, you get most of your ingredients from one source but then there will be other sources that specialize in specific ingredients. You want the best stuff in your final dish.
This and this again. Everyone gangsta till something breaks in production. I almost shat myself once with a CF worker that buffer overloaded because I decided “Meh, the code works”. So I didn’t check that Gemini didn’t take the memory limit of workers during runtime, into consideration.
I’ve been to a few Vibe Code events with founders that vibe coded their funded MVP. Guess what none of them vibe coding and hired devs to rebuild.
There’s always the argument that Bubble is for MVPs. I’d argue that vibe coded apps are only for MVPs.
7 Likes
I’m about to set up my first virtual machine with digital ocean thanks to insights you’ve shared regarding using a tech stack that bubble is part of.
4 Likes
Glad to hear that 
It’s important to not pay more than what you need, cause these things do add up when you’re not looking. Find a service that fits, not fit to the service and you can manage costs.
Which is why I prefer services like Cloudflare because they are not middleware like say Supabase. I get granular control, it’s as close as I can get to running on prem servers. Hence why it can be cheaper.
Also, unlike large umbrella businesses like Amazon, Google and Microsoft, Cloudflare just does Cloudflare stuff. Big 3 are pretty stingy with their pricing IMO.
Side note: Man those VM prices. Why Digital Ocean and not something like Hostinger? Hostinger is literally less than half the price as you scale. That said, I’m still new to Hostinger.
hetzner!
Hetzner too