Cloud
Post #7502, on Dec 2, 2025 in TG
Vibe Coding Founders Shocked by First Cloud Resource Bill
Description
A meme showing Snoop Dogg at what appears to be a sporting event, lifting his sunglasses with a shocked and bewildered expression. He's wearing a white jacket with an NBC/Snoop Dogg logo. The top text reads: 'Vibe coding founders when they vibe deploy and get their first bad vibes Cloud Resource Bill'. The meme captures the moment of reckoning when startup founders who've been building apps through 'vibe coding' (using AI to generate code without deep technical understanding) encounter the financial reality of cloud infrastructure costs for the first time. The shocked expression perfectly conveys the sticker shock of seeing auto-scaled cloud bills
Use J and K for navigation
Comments
28Comment deleted
The vibe was immaculate until AWS decided the monthly bill should also have main character energy
same for me with anthropic api bills Comment deleted
im moving to deepsek Comment deleted
What about self-hosted? Comment deleted
Imagine smiling Jensen Huang Comment deleted
Nope… I prefer to get AI processes 1s slower than adding 3 zeroes to my bills Comment deleted
Unless we deal with a life-or-death situation, speed is cool, but we need to understand (as a society) that sometimes we need to slow down Comment deleted
I know a project with compilation time of 8 hours. Do you really want that?) Comment deleted
me when I compile entire codebase instead of breaking it down to components Comment deleted
good luck breaking down compiler to components Comment deleted
Sure! Why not? Or are you talking about specific compiler what was designed monolithic? Comment deleted
that specific compiler had to be monolithic Comment deleted
I don't want anything, just saying that GPUs are expensive and maybe (just maybe) the final user should wait while a non-GPU server runs an AI process Comment deleted
just order your own circuit with parallel processing from tsmc Comment deleted
currently high memory server would cost you more than a GPU 😁 Comment deleted
It depends on the provider Comment deleted
Also, I am not sure if 8h of compilation time is good either (because I don't know the details of the project)… maybe optimise the app instead? Comment deleted
Btw, main thing about cloud computing is flexibility. If today your app got a lot of users, you just scale up. And if tomorrow they move out, you just scale down. But if you have consistent user base, self host will be much cheaper. Comment deleted
the main thing about cloud computing is not giving a f about infra. selfhost is way cheaper in any case Comment deleted
Nope, its not. Imagine ordering top end gpu cluster only to realise that product unpopular and gpu sit idle 99% of time. Comment deleted
Imagine buying a hall of mainframes to start the bank only to realise you need none of them Comment deleted
Overcommitment is a CTO problem, not the solution choice Comment deleted
tech failures aren't uncommon that's why we know how to estimate Comment deleted
Instead of thinking about AWS, GCP or Azure (the big three) why not thinking about cheaper providers that give you a preset CPU and RAM? It's the same exact thing Cloud providers are super-expensive because they want you to fall into the K8S trap (which, in the end, launches even more servers, aka more expensive invoices) Comment deleted
Exhale Comment deleted
it's really not, they have shareholders above their heads constantly setting higher and higher revenue goals each quarter and they do care about issue of servers not being used as well, they can't simply downscale themselves either. But one or two failed projects canceling their purchase aren't going to be their end Comment deleted
an actual cheap provider will provide you a rack space, an electric outlet and network connection. Comment deleted
Where I work, we have just cut 60% of the total expenses by leaving the big cloud providers, because we don't have any needs that require their expensive infrastructure 🤷♂️ Comment deleted