What would this have costed had they used render or railway? Maybe, what, $200 a month?
Vercel's pricing is so ridiculously convoluted that you can't even cleanly compare usage. With render/railway/(insert provider of choice) you can at least predict that you're your biggest cost is going to be egress.
edit: I just saw that it gets 450m pageviews. I'm guessing on the upper end this costs ~$1k with railway + cloudflare?
While I used to think Railway was an amazing service, I had a production workload get broken because they removed a feature without any depreciation period or warning. I now struggle to recommend it for anything more than a hobby project. Vercel has the benifit of being big enough they have to do things properly. For reference https://station.railway.com/questions/smtp-connection-failur...
your company's mod in that thread said there was supposed to be no SMTP at all for any plan but it was enabled by a bug. then once you saw people were using it you decided to milk your bug via the most expensive payment plan.
but that's your internal dealings. from your paying customers perspective, company had a change in the environment where something was working and now it's not. which could be even okay if it was a legit bug that was fixed, but what makes it worse is instead you said "just pay us 4x more and you get it back". for some users it probably just broke
production, is there a more perfect time for blackmail right?
don't try to paint it as altruistic attempt to reduce spam in the internet, this is sleazy af
I would stay away from any startup for production workload.
Made the mistake. Never again.
Fly, railway, render. Avoid. All have weird show stopper bugs for any reasonable scale and you will fight against the platform compared to using big cloud.
And big cloud works better even in cases where PAAS is advertised as simpler (google cloud run and build is as easy to setup as railway but you have much more knobs to control traffic, routing, roll out etc)
Why not bare-cloud? Esp with AI... in 10min or less an agent can deploy almost any stack to an optimal AWS setup for a fraction of the cost of any platform.
Angelo from Railway here, Railway runs our own metal for the sheer reason to preserve margins so we can run for perpetuity.
We're nuts for studying failure at the company and Heroku's margins was one of the things we considered to be one of the many nails in that coffin. (RIP)
It is fucking CRAZY how many cloud companies don't let you set a spending limit.
I had to hunt around for a host in a suitable geography with a spending limit, almost had to go on-prem (which will happen eventually, but not in the startup phase)
Waking up to bankruptcy because of bots out of your control visiting your website seems a little nuts. Adding some other bullshit on top (like cloudflare) seems even more nuts.
Yeah I can manage all that and have the machine stop responding when it hits a spending limit -- but why would I pay for the cloud if I have to build out that infrastructure?
1. Because people vote with their wallets and not their mouths, and most companies would rather have a cost accident (quickly refunded by AWS) rather than everything going down on a saturday and not getting back up until finance can figure out their stuff.
2. Because realtime cost control is hard. It's just easier to fire off events, store them somewhere, and then aggregate at end-of-day (if that).
I strongly suspect that the way major clouds do billing is just not ready for answering the question of "how much did X spend over the last hour", and the people worried about this aren't the ones bringing the real revenue.
> and the people worried about this aren't the ones bringing the real revenue.
It's this one. If you're in a position to refund a "cost accident", then clearly you don't have to enforce cost controls in real time, and the problem becomes much easier to achieve at billing cycle granularity; the user setting a cost limit is generally doesn't care if you're a bit late to best-effort throttle them.
> I strongly suspect that the way major clouds do billing is just not ready for answering the question of "how much did X spend over the last hour", and the people worried about this aren't the ones bringing the real revenue.
See: Google's AI studio. Its built on Google Cloud infrastructure so billing updates are slow which peeves users used to instant billing data with Anthropic and OpenAI.
People act like this is an easy problem. What should a cloud provider do when you hit your limit? Delete your files from storage? Kill your database instance? Automatically terminate your VMs? Erase your backups?
500 at hetzner, they don't go up to that price, and even with their prices raised during the RAM shortage, for 500, you can still have 5 servers each with 4TB NVMe and 128GB RAM, and a Ryzen 9 7950X3D (16 cores).
Seems like their setup price has gone up from 1 month to 2.5 months. Ouch. That'll be to cover the RAM price.
The pricing is so bad I had to remove my CC details. One mistake and you wake up with a 50K bill for your personal project that was just you exploring.
Why is everyone using Vercel and the likes anyway?
Setting up a VPS with Node takes ten minutes and is miles cheaper. And it's not like you never have to debug issues with serverless configurations, which can even occasionally be harder to debug because of their proprietary natures.
>Why is everyone using Vercel and the likes anyway?
Because you literally connect to a git repo and your site is deployed, and scales with load. Compared to managing a VM, system and application packages, config, backups and then how do you scale that?
I have what is basically a demo running on Vercel free tier, there's no way I'm maintaining a VM for that lol.
If I had a serious site, same. If I had a team, then the equation would be different.
This is the Dropbox problem. People are willing to pay for convenience, and tech folks tend to underestimate how much convenience comes from seemingly simple solutions
it literally is though that's why i'm confused. you pay a flat monthly fee and get a box that runs linux. yes you might not be able to press one button and Effortlessly Deploy Your AI-Managed SaaS Infrastructure Product To Valued Customers Across The Metaverse or whatever vercel does but it only takes a couple hours to learn how to setup nginx node rsync and cloudflare (and even then i think there's some easier closer to one-click solutions)
also developing an app is something you need to be quite tech savvy to do anyway. genuinely are there really people who have the skill and patience to do that and then get stumped trying to deploy it? clearly there are since stuff like this is so popular I just don't really understand
I think the free/cheap tier is what gets people kind of hooked... it's really easy to setup something like Dokku self-hosted and run a few dozen apps on a decent rented server... Even then, there's something appealing about not having to even worry about it. Why bother setting up your own server(s) and databases when you can run in Cloudflare workers with CockroachLabs or Turso?
Even with my own server, I've explored the option(s) just to avoid potential pain down the road regarding excess load.
Sure, but my point is that 450m requests is just not a lot. Humans are bad at big numbers, and this number sounds like a scary amount of traffic, my point is that it is a very boring amount of traffic even on modest hardware.
I personally wouldn't stick it on a single VPS because we can do way better than that in terms of reliability, but the point stands that you don't need very much in terms of resources to serve this.
The post said 450 million pageviews, likely since November. If we make very generous assumptions and assume that each pageview is a megabyte (very generous based on my own experience scanning billions of websites), then that's 450TB total in traffic. If you really did 450TB per month in traffic, you would need slightly more than one gigabit line (and hence VPS), but not more than two. With Hetzner the traffic would cost you €450 or $535.
Well, https://jmail.world/jacebook-logo.png is 670KB by itself and loaded on initial load, so I think they might have blown your suggested traffic budget and still have some optimization to do.
Because inexplicably, there's random pixel-level noise baked into the blue area. You can't see it unless you crank up contrast, but it makes the bitmap hard to compress losslessly. If you remove it using threshold blur, it doesn't change the appearance at all, but the size is down to 100 kB. Scale it down to a more reasonable size and you're down to 50 kB.
None of this is due to "modern web development". It's just about a dev not checking reasonable asset size before deploying/compiling, that has happened in web, game-dev, desktop apps, server containers, etc. etc.
This should be an SVG (a few kb after proper compression) or if properly made as a PNG it'd probably be in 20-ish kb.
The dev not having the common sense to check file size and apparently not realising that the PNG format was being grossly misused for this purpose (by not even having a single tone of white for the J and the corners, let alone for the blue background) is modern web development.
What is that noise actually? It's clearly not JPEG artifacts. Is it dithering from converting from a higher bitdepth source? There do appear to be very subtle gradients.
I would bet it's from AI upscaling. The dark edges around high contrast borders, plus the pronounced and slightly off-colour antialised edges (especially visible on the right side of the J) remind me of upscaling models.
If it's large enough for say 2x or more "retina" usage... a very minor soften filter, then color reduction to get well under 256 colors (often can do 32-64) quantization and oxipng w/ zopfli can go a long way... just getting down to a palette over rgb for png brings down sizes a lot... palette reduction to around 32 colors does a bit too. Just depends on the material.
That said, the actual size of this image is massive compared to where it needs to be, and looking at it, should definitely be able to quantize down to 32-64 colors and reduce the size to even 4x the render size... let alone just using svg, as others have mentioned.
Oh, ding ding! Opening in a hex editor, there's the string "Added imperceptible SynthID watermark" in an iTXt chunk. SynthID is apparently a watermark Google attaches to its AI-generated content. This is almost certainly the noise.
Fair enough, I just loaded some pages and some of them are even bigger than 2MB. But then again those static resources would be cached client-side. So unless you have 450 million unique visitors who only ever go to one URL on your site, you are looking at significantly less per pageview. I reloaded the frontpage with caching enabled and it was ~ 30kB of data transfer.
Genuine question: How is that a value proposition when Cloudflare offers a CDN for free with unlimited bandwidth, that you could just put in front of the sweaty VM in Helsinki?
Not trying to be obtuse, I really don't get how other providers can compete with that, I can't imagine Vercel's CDN is so significantly superior to make it worth it.
For that matter, the entire site could be in a Cloudflare worker with all the content in R2 (no egress fees, just storage). Likely to barely exceed the $5 baseline price for a paid account. (not sure on the storage costs, but likely well under $100/mo)
Yes, and I didn't mean to imply that a single VPS is all you needed. But I wanted to put things into perspective for the other posters who claimed that you couldn't possibly serve a site like this from a single machine, purely in terms of performance.
For high traffic check places like datapacket (no affiliation), or slightly cheaper, places like onlyservers (no paid affiliation, was a customer) or even find a transit provider and a colo and a server yourself. For $535 a month or less you can get a pretty good amount of bandwidth.
It's Vercel, it's their business model. They have all the minnows eating for free, and then once you pass a threshold, you're a whale with a whale–sized bill.
Say what you want about Elon but X is where all the investors and tech execs are. Nobody is going to sign up for threads because they saw it link to a picture in a HN post
unfortunately it's also where the signal is distorted by the noise of a thousand blue check hangers-on who add nothing to the comment thread besides "great job, Mr CEO!", "this is a game changer", and other unquestioning platitudes.
I know there's a lot of questions why it's so expensive, but can I just extol the work done by Riley and team?
Since the Epstein files dropped they've cloned gmail, gdrive, gmessages, amazon orders, transcribed court proceedings (yes with AI), fights, facebook, and imessages.
It's an insane amount of work. They added the latest batch of files, photos, videos in like 2 weeks. And he's keeping up files that the justice department took down.
jmail has made it so much easier for everyone to explore the files.
I don't know how Riley has planned to monetize this or if it's simply for the public good. I can totally understand not wanting to optimize for cost from the outset. And I see a lot of abject criticism on every social media platform rather than constructive.
How does this work from an accounting perspective? They write off a bad debt, but the actual loss is likely multiple orders of magnitude less. Do they only get to write off up to the actuals?
It's simply discounting the fees for that one user to zero.
(It's not writing off a bad debt, which is technically different)
So: your costs are still X but now your revenue is Y instead of Y + (that one user's fee which likely wasn't going to get paid anyway)
You pay taxes on Y - X (profit).
So, really, their costs just increased by whatever it cost to deliver that data (likely zero depending on how they're billed for it), and their revenue didn't change at all.
Turning a no-collect situation into a PR positive.
To be fair: it really depends on their datacenter environment; if they're physically hosting, this is probably a rounding error. But, if instead, they're actually running on top of AWS or another hyperscaler and paying 9 cents per gigabyte for traffic, then their bandwidth bill could actually be quite substantial and they're just passing that along to the customer. In that case, this could be actually quite generous of them.
I don't really understand why he'd say he'd cover the costs personally... like, Vercel can just write it off, what's the significance of him paying for it?
I'm not the first to point this out but the website in question, which is mostly static, could easily be hosted on a VPS for at most a couple hundred dollars a month.
Rather than linking to a threads post that is a screenshot of the x.com post with little to no commentary, we should be linking to the original x.com post
Considering that Twitter doesn't show the original post for non-logged-in users, the screenshot on Threads actually provides a better reading experience for most people!
What would this have costed had they used render or railway? Maybe, what, $150 a month?
Vercel's pricing is so convoluted that you can't even compare usage. With render/railway you can at least predict that your biggest cost is going to be high.
Why wouldn't it be good to be associated with publicly exposing pedophiles, cannibals, murderers, and rapists? That seems to be a very good thing to be opposed to them.
Its good PR. He had some pretty bad PR recently that caused a lot of people to boycott the service. I assume this is him trying to regain trust or something?
1) Covering the ~$50k hosting bill for Jmail on Vercel sounds generous, but a self-hosted VPS on Hetzner could serve the same purpose for ~€30/month, which is orders of magnitude cheaper and avoids vendor lock-in.
2) This comes as the CEO of Vercel, Guillermo Rauch, is already facing community backlash for publicly supporting Israeli Prime Minister Netanyahu, a move that’s led to boycotts and migrations off the platform among developers. All my homies hate Vercel.
I'm new to webdevelopment and was using Vercel because people told me it was good, but I was unaware that the company supported the genocide. What other similar services there are that you would recommend?
Railway/Render can easily host Next.js applications
You could also get a VPS from Digital Ocean or Hertzner, run open-source PaaS like Coolify, Dokploy, Caprover, etc.
Digital Ocean has app platform that’s lets you host these applications but more experience than VPS
Sealos has a $7 and $25 plan and work with Next.js
Just a few options. If you’re looking to leave Next.js completely, check out Tanstack Start. It’s by the creator of React-Query (defacto way to handle fetching data in Next.js). Still some rough edges but I think it could overtake Next.js once a bit more mature.
Do you really think a $30 hetzner host can sustain that level of traffic performantly? Don't get me wrong, I love hetzner, but I would be very surprised if the numbers work out there.
It shouldn't. The issue is that most developers would rather spin up another instance of their server than solve the performance issue in their code, so now it's a common belief that computers are really slow to serve content.
And we are talking about static content. You will be bottlenecked by bandwidth before you are ever bottlenecked by your laptop.
To be fair, computers are slow if you intentionally rent slow & overpriced ones from really poor-value vendors like cloud providers. For people who started their career in this madness they might be genuinely unaware of how fast modern hardware has become.
With a 2025 tech stack, yes. With a 2005 tech stack, no. Don't use any containers, no[/limited] server-side dynamic script languages, no microservices or anything like that.
Considering the content is essentially static, this is actually viable. Search functions might be a bit problematic, but that's a solvable problem.
Of course you pay with engineering skills and resources.
Is there any feasible way to implement search client-side on a database of this scale?
I guess you would need some sort of search term to document id mapping that gets downloaded to the browser but maybe there's something more efficient than trying to figure out what everyone might be searching for in advance?
And how would you do searching for phrases or substrings? I've no idea if that's doable without having a database server-side that has the whole document store to search through.
I think the key thing here is the context and size; the searchable content of even a lot of e-mails is quite dense and small. I'm not a search expert but I'd look at precalculated indexes on very short substrings (3-4 characters maybe?), have the client pull those it needs for a particular query and then process client-side from there. (Doesn't even need figuring out in advance what people will search for, though that'd certainly improve things.)
Theoretically, just thinking about the problem... You could probably embrace offline first and sync to indexeddb? After that search would become simple to query. Obviously comes with it's own challenges, depending on your user base (e.g. not a good idea if it's only a temporary login etc)
there's been demos of using SQLite client-side, with the database hosted in S3, and HTTP range requests used to only fetch the necessary rows for the query.
there might be some piece I'm missing, but the first thing that comes to mind would be using that, possibly with the full-text search extension, to handle searching the metadata.
at that point you'd still be paying S3 egress costs, but I'd be very surprised if it wasn't at least an order of magnitude less expensive than Vercel.
and since it's just static file hosting, it could conceivably be moved to a VPS (or a pair of them) running nginx or Caddy or whatever, if the AWS egress was too pricey.
There are several implementations of backing an Sqlite3 database with a lazy loaded then cached network storage, including multiple that work over HTTP (iirc usually with range requests).
Those basically just work.
Containers themselves don't, but a lot of the ecosystem structures around them do. Like having reverse proxies (or even just piles of ethernet bridges) in front of everything.
Or if you go ping pong across containers to handle a single request. That will certainly make a laptop unable to handle this load.
No it won't. This is static content we're talking about. The only thing limiting you is your network throughput and maybe disk IO (assuming it doesn't fit in a compressed RAM). Even for an "around the globe roundtrip" latency, we're still talking few hundred msec.
Some cloud products have distorted an entire generation of developers understanding of how services can scale.
I just fired up a container on my laptop... running on kubernetes... running in a linux VM. It's lightly dynamic (no database or filesystem I/O).
While I've also got enough other stuff running that my 15 min load average is at 4 and I've got 83% RAM used ignoring buffers/caches/otherwise.
I went and grabbed a random benchmarking tool and pointed it at it with 125 concurrent connections.
Sustained an average of 13914 reqs/s. Highest latency was 53.21ms.
If there are 10,000 people online at any given time hitting the API on average once every 3 seconds (which I believe are generous numbers), you'd only be around 3.3k reqs/s, or about 24% of what my laptop could serve even before any sort of caching, CDN, or anything else.
So... if a laptop can't serve that sort of request load, it sounds more like an indictment of the site's software than anything.
I think it’s more helpful to discuss this in requests per second.
I’d interpret “thousands of people hitting a single endpoint multiple times a day” as something like 10,000 people making ~5 requests per 24 hours. That’s 0.5 requests per second.
A laptop from 10 years ago should be able to comfortably serve that. Computers are really really fast. I'm sorry, thousands of users or tens of thousands of requests a day is nothing.
There may be a risk of running into thermal throttling in such a use-case, as laptops are really not designed for sustained loads of any variety. Some deal with it better than others, but few deal with it well.
Part of why this is a problem is that consumer grade NICs often tend to overload quite a lot of work to the CPU that higher end server specced NICs do themselves, as a laptop isn't really expected to have to keep up with 10K concurrent TCP connections.
I would use a $100/mo box with a much better CPU and more RAM, but I think the pinch point might be the 1Gbps unmetered networking that Hetzner provide.
They will sell you a 10Gbps uplink however, with (very reasonably priced) metered bandwidth.
A profitable customer? How would Hetzner know if you're profitable or not?
I've hosted side projects on Hetzner for years and have never experienced anything like that. Do you have any references of projects to which it happened?
I am not sure how one even gets 250TB/mo through a 1Gbps link. In any case, completely saturating your networking for the full month is outside most people's definition of "fair use".
Yeah but they still advertise with unlimited traffic.
"All root servers have a dedicated 1 GBit uplink by default and with it unlimited traffic"
https://docs.hetzner.com/robot/general/traffic/
Even before the Vercel CEO supporting a genocidal maniac. Vercel as a platform has been silently giving open source projects a "fuck you, pay me" when it comes to renewing benefits.
Have seen it happen to smaller projects and even pointed it out when Vercel took static sites down.
2nd point resonates with me, how come he wants to cover expenses, while being connected to Israeli PM and Epstein is connected to Israel through Ehud Barak.
Barak and bibi are political enemies (or at least we're when Barack was a relevant political figure) and besides that I haven't seen anything suggesting that his connection with bibi is more than the one meeting that was publicized.
This is offtopic honestly, but I'm curious if I've been using this phrase wrong for my whole life. Doesn't "order of magnitude" refer to steps of powers of ten?
$50000 vs €30. (or €42066.30 vs €30 if I normalize the currency) 5x10^4 vs 3x10^1.
You have it right, perhaps the original poster was referring to it in a more colloquial manner, in the sense that against 200 million in revenue, 50,000 and 30 are in the same ballpark?
> in the sense that against 200 million in revenue, 50,000 and 30 are in the same ballpark
I don't understand how those are in the same ballpark? I thought saying something is in the same ballpark suggested that they are similar in scale, and the implication is that little-leauge does not play in the same ballpark as a NBA team. They are in the same category (baseball), but not at all the same level.
At a big enough scale, previously large differences are effectively 0.
50k/mo is 600,000/yr vs 360/yr at 30/mo. Thats existential for a 1MM/yr company. Neither register on a balance sheet for a 1B/yr company. They are both closer to 0 than being a major cost.
But saying that 200 million and 30 are in the same ballpark is not true in 99.99% of contexts.
Even 50k and 30 I would not say are in the same ballpark. I've worked for major corps and of course a cost saving of 50k/month would not register for the overall company but it probably would for my team. A saving of 30/month is probably not worth spending any considerable amount of time on in most non-personal contexts.
Insane to me a bill that large for what is effectively hosting static content. He could dump the entire thing on S3 and even with cloudfront it would be fraction of that.
Public files needing to be distributed to a huge population of interested persons? Sounds like the perfect situation for an oldschool torrent. That's how large data leaks were handled back in my day. 450TB is peanuts for perhaps ten thousand peers on fast residential connections.
This is an absolutely ludicrous bill for that number of page views. Even if it was built in the dumbest way possible (I'm not saying it was) this would be ridiculous. This is not as good a look for Vercel as they think it is. I would be embarrassed to shine more light on this pricing if I was the CEO.
What would this have costed had they used render or railway? Maybe, what, $200 a month?
Vercel's pricing is so ridiculously convoluted that you can't even cleanly compare usage. With render/railway/(insert provider of choice) you can at least predict that you're your biggest cost is going to be egress.
edit: I just saw that it gets 450m pageviews. I'm guessing on the upper end this costs ~$1k with railway + cloudflare?
Railway is getting so good I'm not sure what Vercel brings to the party anyway.
While I used to think Railway was an amazing service, I had a production workload get broken because they removed a feature without any depreciation period or warning. I now struggle to recommend it for anything more than a hobby project. Vercel has the benifit of being big enough they have to do things properly. For reference https://station.railway.com/questions/smtp-connection-failur...
SMTP is gated behind the $20/mo Pro plan to reduce spam on the internet.
It sounds like you were running a production workload on the Hobby plan
that's rich, from Railway employee.
your company's mod in that thread said there was supposed to be no SMTP at all for any plan but it was enabled by a bug. then once you saw people were using it you decided to milk your bug via the most expensive payment plan.
but that's your internal dealings. from your paying customers perspective, company had a change in the environment where something was working and now it's not. which could be even okay if it was a legit bug that was fixed, but what makes it worse is instead you said "just pay us 4x more and you get it back". for some users it probably just broke production, is there a more perfect time for blackmail right?
don't try to paint it as altruistic attempt to reduce spam in the internet, this is sleazy af
You were too cheap to pay $20/mth for the Pro plan... That says a lot more about you than Railway.
theres also a ton of smtp-> mailservice proxies.0
I would stay away from any startup for production workload.
Made the mistake. Never again.
Fly, railway, render. Avoid. All have weird show stopper bugs for any reasonable scale and you will fight against the platform compared to using big cloud.
And big cloud works better even in cases where PAAS is advertised as simpler (google cloud run and build is as easy to setup as railway but you have much more knobs to control traffic, routing, roll out etc)
Jmail is itself an experiment, that doesn't need to be production quality.
And their CEO doesn't post selfies with war criminals
OOTL, what is this a reference to?
His photo with a prime minister of a country with a famous arrest warrant in The Hague.
You know, the same country whose former prime minister is this person: https://jmail.world/person/ehud-barak
https://finance.yahoo.com/news/vercel-ceo-takes-selfie-israe...
Why not bare-cloud? Esp with AI... in 10min or less an agent can deploy almost any stack to an optimal AWS setup for a fraction of the cost of any platform.
AWS is still expensive as fuck, just go for a VPS or dedicated server at that point
Every single mentioned service is either an AWS or GCP abstraction.
Angelo from Railway here, Railway runs our own metal for the sheer reason to preserve margins so we can run for perpetuity.
We're nuts for studying failure at the company and Heroku's margins was one of the things we considered to be one of the many nails in that coffin. (RIP)
(my rant here: https://blog.railway.com/p/heroku-walked-railway-run)
thanks for the correction
Fascinating, thanks for chiming in.
Pretty sure Hetzner don't share infrastructure with either of those.
Wake me up when GCP allows you to spending limits
It is fucking CRAZY how many cloud companies don't let you set a spending limit.
I had to hunt around for a host in a suitable geography with a spending limit, almost had to go on-prem (which will happen eventually, but not in the startup phase)
Waking up to bankruptcy because of bots out of your control visiting your website seems a little nuts. Adding some other bullshit on top (like cloudflare) seems even more nuts.
Yeah I can manage all that and have the machine stop responding when it hits a spending limit -- but why would I pay for the cloud if I have to build out that infrastructure?
grumble.
2 reasons basically.
1. Because people vote with their wallets and not their mouths, and most companies would rather have a cost accident (quickly refunded by AWS) rather than everything going down on a saturday and not getting back up until finance can figure out their stuff.
2. Because realtime cost control is hard. It's just easier to fire off events, store them somewhere, and then aggregate at end-of-day (if that).
I strongly suspect that the way major clouds do billing is just not ready for answering the question of "how much did X spend over the last hour", and the people worried about this aren't the ones bringing the real revenue.
> and the people worried about this aren't the ones bringing the real revenue.
It's this one. If you're in a position to refund a "cost accident", then clearly you don't have to enforce cost controls in real time, and the problem becomes much easier to achieve at billing cycle granularity; the user setting a cost limit is generally doesn't care if you're a bit late to best-effort throttle them.
> I strongly suspect that the way major clouds do billing is just not ready for answering the question of "how much did X spend over the last hour", and the people worried about this aren't the ones bringing the real revenue.
See: Google's AI studio. Its built on Google Cloud infrastructure so billing updates are slow which peeves users used to instant billing data with Anthropic and OpenAI.
People act like this is an easy problem. What should a cloud provider do when you hit your limit? Delete your files from storage? Kill your database instance? Automatically terminate your VMs? Erase your backups?
Try it out. Implementation is always harder than conjecture
I do. Every day, for at least 5 services.
Are you able to bypass the aws web app entirely via the command line?
yea i mean i basically have claude code do everything with aws cli.
if they had used hetzner Cloud servers, probably like 500 a month lol
With cloudflare? Less than 100, easily.
500 at hetzner, they don't go up to that price, and even with their prices raised during the RAM shortage, for 500, you can still have 5 servers each with 4TB NVMe and 128GB RAM, and a Ryzen 9 7950X3D (16 cores).
Seems like their setup price has gone up from 1 month to 2.5 months. Ouch. That'll be to cover the RAM price.
The pricing is so bad I had to remove my CC details. One mistake and you wake up with a 50K bill for your personal project that was just you exploring.
I don't think cost is the dominating concern here.
Why is everyone using Vercel and the likes anyway?
Setting up a VPS with Node takes ten minutes and is miles cheaper. And it's not like you never have to debug issues with serverless configurations, which can even occasionally be harder to debug because of their proprietary natures.
>Why is everyone using Vercel and the likes anyway?
Because you literally connect to a git repo and your site is deployed, and scales with load. Compared to managing a VM, system and application packages, config, backups and then how do you scale that?
I have what is basically a demo running on Vercel free tier, there's no way I'm maintaining a VM for that lol.
If I had a serious site, same. If I had a team, then the equation would be different.
This is the Dropbox problem. People are willing to pay for convenience, and tech folks tend to underestimate how much convenience comes from seemingly simple solutions
this is convenience for tech folks and the price isn't a few bucks a month but 1000x.
It's like if Dropbox was an rsync server (no app) and it cost $10,000 a month for 1TB of space. Think it would still take off?
They spend $$$ on marketing to execs. A couple of months ago my CEO came to me with a $30k Vercel app I could have done for $1500 on our own hardware.
It’s just that simple!
https://www.explainxkcd.com/wiki/index.php/2501:_Average_Fam...
it literally is though that's why i'm confused. you pay a flat monthly fee and get a box that runs linux. yes you might not be able to press one button and Effortlessly Deploy Your AI-Managed SaaS Infrastructure Product To Valued Customers Across The Metaverse or whatever vercel does but it only takes a couple hours to learn how to setup nginx node rsync and cloudflare (and even then i think there's some easier closer to one-click solutions)
also developing an app is something you need to be quite tech savvy to do anyway. genuinely are there really people who have the skill and patience to do that and then get stumped trying to deploy it? clearly there are since stuff like this is so popular I just don't really understand
I think the free/cheap tier is what gets people kind of hooked... it's really easy to setup something like Dokku self-hosted and run a few dozen apps on a decent rented server... Even then, there's something appealing about not having to even worry about it. Why bother setting up your own server(s) and databases when you can run in Cloudflare workers with CockroachLabs or Turso?
Even with my own server, I've explored the option(s) just to avoid potential pain down the road regarding excess load.
Most people don't know what VPS and Node are.
450 million pageviews on Vercel = $46,000
450 million pageviews on a single 16c/32t OVH box with nginx and a 3 Gbps connection = $245
450m page views in 1 month is only 173 requests per second. You can do that on a much cheaper box.
You're assuming an extremely consistent traffic rate, with zero spikes....
Sure, but my point is that 450m requests is just not a lot. Humans are bad at big numbers, and this number sounds like a scary amount of traffic, my point is that it is a very boring amount of traffic even on modest hardware.
I personally wouldn't stick it on a single VPS because we can do way better than that in terms of reliability, but the point stands that you don't need very much in terms of resources to serve this.
Sure, you are only proving my point even further. I just happen to have a personal policy of "no boxes in prod under 16 cores". :)
why?
Having someone on-call 24/7 to keep that $245 box running: $400K.
To run a $245 box, you'd need an entire team with Kubernetes experience — at least one Scrum Master included. Under $2M, I’d say it’s impossible.
Yeah... lol, because cloud setups never fall over. Ever.
The post said 450 million pageviews, likely since November. If we make very generous assumptions and assume that each pageview is a megabyte (very generous based on my own experience scanning billions of websites), then that's 450TB total in traffic. If you really did 450TB per month in traffic, you would need slightly more than one gigabit line (and hence VPS), but not more than two. With Hetzner the traffic would cost you €450 or $535.
Did I get something wrong?
Well, https://jmail.world/jacebook-logo.png is 670KB by itself and loaded on initial load, so I think they might have blown your suggested traffic budget and still have some optimization to do.
How is that image 670 KB!? Definitely some optimization low-hanging fruit there.
Edit: dang, even pngcrush can't get it below 580 KB. Disappointing performance on PNG's part.
Because inexplicably, there's random pixel-level noise baked into the blue area. You can't see it unless you crank up contrast, but it makes the bitmap hard to compress losslessly. If you remove it using threshold blur, it doesn't change the appearance at all, but the size is down to 100 kB. Scale it down to a more reasonable size and you're down to 50 kB.
Modern web development never ceases to amaze me.
None of this is due to "modern web development". It's just about a dev not checking reasonable asset size before deploying/compiling, that has happened in web, game-dev, desktop apps, server containers, etc. etc.
This should be an SVG (a few kb after proper compression) or if properly made as a PNG it'd probably be in 20-ish kb.
The dev not having the common sense to check file size and apparently not realising that the PNG format was being grossly misused for this purpose (by not even having a single tone of white for the J and the corners, let alone for the blue background) is modern web development.
Right, so you mean that this is unique and inherent to web dev and specifically modern web dev.
What is that noise actually? It's clearly not JPEG artifacts. Is it dithering from converting from a higher bitdepth source? There do appear to be very subtle gradients.
I would bet it's from AI upscaling. The dark edges around high contrast borders, plus the pronounced and slightly off-colour antialised edges (especially visible on the right side of the J) remind me of upscaling models.
Not even the white is pure. There are at least #FFFFFD, #FFFFFB and #FEFEFE pixels sprinkled all over the #FFFFFF.
If it's large enough for say 2x or more "retina" usage... a very minor soften filter, then color reduction to get well under 256 colors (often can do 32-64) quantization and oxipng w/ zopfli can go a long way... just getting down to a palette over rgb for png brings down sizes a lot... palette reduction to around 32 colors does a bit too. Just depends on the material.
That said, the actual size of this image is massive compared to where it needs to be, and looking at it, should definitely be able to quantize down to 32-64 colors and reduce the size to even 4x the render size... let alone just using svg, as others have mentioned.
I'd bet that it's AI generated, resulting in the funky noise.
Oh, ding ding! Opening in a hex editor, there's the string "Added imperceptible SynthID watermark" in an iTXt chunk. SynthID is apparently a watermark Google attaches to its AI-generated content. This is almost certainly the noise.
Make it an SVG and it's down to 1kb.
I got it down to 200KB using normal png encoder and just limiting number of colors, but it should be replaced with tiny SVG.
Fair enough, I just loaded some pages and some of them are even bigger than 2MB. But then again those static resources would be cached client-side. So unless you have 450 million unique visitors who only ever go to one URL on your site, you are looking at significantly less per pageview. I reloaded the frontpage with caching enabled and it was ~ 30kB of data transfer.
Isn’t part of Vercel’s value proposition a robust global CDN in front? Seems quite a bit different than one sweaty VM in Helsinki.
Genuine question: How is that a value proposition when Cloudflare offers a CDN for free with unlimited bandwidth, that you could just put in front of the sweaty VM in Helsinki?
Not trying to be obtuse, I really don't get how other providers can compete with that, I can't imagine Vercel's CDN is so significantly superior to make it worth it.
For that matter, the entire site could be in a Cloudflare worker with all the content in R2 (no egress fees, just storage). Likely to barely exceed the $5 baseline price for a paid account. (not sure on the storage costs, but likely well under $100/mo)
Yes, and I didn't mean to imply that a single VPS is all you needed. But I wanted to put things into perspective for the other posters who claimed that you couldn't possibly serve a site like this from a single machine, purely in terms of performance.
Some people don't realize how big machines get. A single ordinary server can have a 4x100Gbps connection and 256 physical CPU cores.
That's not worth 45k. It's barely worth anything for a typical website, tbh.
For high traffic check places like datapacket (no affiliation), or slightly cheaper, places like onlyservers (no paid affiliation, was a customer) or even find a transit provider and a colo and a server yourself. For $535 a month or less you can get a pretty good amount of bandwidth.
well each view of an 'epstien file' is a pdf with images embeded so i think your 1mb might be not so generous.
Isn't Jmail a static site? How could the bill be $47k?
There's a searchbar, and the amount of content you search through is far more than you could do with clientside search.
Definitely not just static content.
That's a good question. As someone bootstraping a few projects on Vercel this post has me looking over at the pricing sheet more closely.
It's Vercel, it's their business model. They have all the minnows eating for free, and then once you pass a threshold, you're a whale with a whale–sized bill.
Bandwith costs
It would be $0 if they spent 10 minutes just throwing Cloudflare CDN in front. They don't even need to move off Vercel.
Tech bros and VCs need to eat, that's how.
There is still a helluva lot of data to transfer to the client, I'm sure it's being stored somewhere.
And with the entire world perusing this archive, I'm sure the costs will be very high, regardless of provider.
Unreadable without an instagram/threads account
no account needed -> https://www.threads.com/@qa_test_hq/post/DUkC_zjiGQh/media
It's also dumb because the original interaction happened on X https://x.com/rauchg/status/2020984434338693622
Say what you want about Elon but X is where all the investors and tech execs are. Nobody is going to sign up for threads because they saw it link to a picture in a HN post
unfortunately it's also where the signal is distorted by the noise of a thousand blue check hangers-on who add nothing to the comment thread besides "great job, Mr CEO!", "this is a game changer", and other unquestioning platitudes.
X links are also very limited today, and logged out visitors wouldn't see any reply to the tweet.
Change x.com to xcancel.com, has never failed me
Dang maybe fix the link?
>Say what you want about Elon but X is where all the investors and tech execs are
You could say the same thing about certain Island not to long ago, its where business and world leaders are!
With noscript active, I was able to see most of it.
>Vercel's CEO offers to cover expenses of 'Jmail' as it has become the number 1 site for tracking the Epstein files
and the expense is 46,486 USD. He said he is happy to cover expenses and that Vercel worked good for your needs.
creator of Rev (Vercel Mobile) here
you may find it useful to check on costs (among other useful stuff like widgets)
https://github.com/revcel/revcel
Boo is the only think i can imagine when I hear about Vercel
I know there's a lot of questions why it's so expensive, but can I just extol the work done by Riley and team?
Since the Epstein files dropped they've cloned gmail, gdrive, gmessages, amazon orders, transcribed court proceedings (yes with AI), fights, facebook, and imessages.
It's an insane amount of work. They added the latest batch of files, photos, videos in like 2 weeks. And he's keeping up files that the justice department took down.
jmail has made it so much easier for everyone to explore the files.
I don't know how Riley has planned to monetize this or if it's simply for the public good. I can totally understand not wanting to optimize for cost from the outset. And I see a lot of abject criticism on every social media platform rather than constructive.
This was there (way) before the last batch of files was released.
But your point stands, the amount of work they've put into this is remarkable.
How does this work from an accounting perspective? They write off a bad debt, but the actual loss is likely multiple orders of magnitude less. Do they only get to write off up to the actuals?
It's simply discounting the fees for that one user to zero.
(It's not writing off a bad debt, which is technically different)
So: your costs are still X but now your revenue is Y instead of Y + (that one user's fee which likely wasn't going to get paid anyway)
You pay taxes on Y - X (profit).
So, really, their costs just increased by whatever it cost to deliver that data (likely zero depending on how they're billed for it), and their revenue didn't change at all.
Turning a no-collect situation into a PR positive.
To be fair: it really depends on their datacenter environment; if they're physically hosting, this is probably a rounding error. But, if instead, they're actually running on top of AWS or another hyperscaler and paying 9 cents per gigabyte for traffic, then their bandwidth bill could actually be quite substantial and they're just passing that along to the customer. In that case, this could be actually quite generous of them.
You deduct the expenses you paid, not the income you hoped to earn.
Marketing probably, unless thew CEO pulls out his credit card
I don't really understand why he'd say he'd cover the costs personally... like, Vercel can just write it off, what's the significance of him paying for it?
Personal brand building? Wanting Vercel to stay out of politics? A vague attempt at diffusing the focus on Vercel pricing?
Really hard to tell.
Assuming that he wants to keep Vercel out of politics is a somewhat wild take considering what he’s posted in the past.
Yes, because accounts payable are valued at recognized revenue, and aren't being revalued at cost when written off.
Alternatively, bill the costs under the PR department as a marketing campaign.
I suspect this sort of thing is some of the best marketing money can buy anyhow, so it's a bit of a no-brainer.
I'm not the first to point this out but the website in question, which is mostly static, could easily be hosted on a VPS for at most a couple hundred dollars a month.
Directly to CDN. Put it in a CloudFront distribution and it would be a fraction of a fraction of that Vercel bill.
Remember kids, they're incentivized to get you to build something to burn as much compute as possible.
threads.com wrapper for some reason. The actual origin post is on x.com: https://x.com/rtwlz/status/2020957597810254052
Guillermo reply: https://x.com/rauchg/status/2020984434338693622
the image makes it easier to see the top post and response together. also i guess the threads comments add extra context
Rather than linking to a threads post that is a screenshot of the x.com post with little to no commentary, we should be linking to the original x.com post
https://x.com/rauchg/status/2020984434338693622
or to a nitter instance, where you can actually read responses without signing in:
https://xcancel.com/rauchg/status/2020984434338693622 or https://nitter.net/rauchg/status/2020984434338693622
Considering that Twitter doesn't show the original post for non-logged-in users, the screenshot on Threads actually provides a better reading experience for most people!
It seems like sometimes it does and sometimes it doesn't, and I can only imagine popularity is somehow the reason.
You're missing a large part of the conversation and context if you don't at least link to the source.
Only if you already have a threads account. On mobile it makes you sign in
Agreed, was slow to load and I just had to find the source post on X to view the real conversation "thread".
This is the first Threads link I've ever seen here. Is that what Threads mainly is, reposting X screenshots and starting a sidechain conversation?
for fucks sakes, don’t link twitter
Use nitter or xcancel
What would this have costed had they used render or railway? Maybe, what, $150 a month?
Vercel's pricing is so convoluted that you can't even compare usage. With render/railway you can at least predict that your biggest cost is going to be high.
Is that good PR?
Doesn't seem to be a good idea to be associated with that.
Why wouldn't it be good to be associated with publicly exposing pedophiles, cannibals, murderers, and rapists? That seems to be a very good thing to be opposed to them.
Its good PR. He had some pretty bad PR recently that caused a lot of people to boycott the service. I assume this is him trying to regain trust or something?
1) Covering the ~$50k hosting bill for Jmail on Vercel sounds generous, but a self-hosted VPS on Hetzner could serve the same purpose for ~€30/month, which is orders of magnitude cheaper and avoids vendor lock-in.
2) This comes as the CEO of Vercel, Guillermo Rauch, is already facing community backlash for publicly supporting Israeli Prime Minister Netanyahu, a move that’s led to boycotts and migrations off the platform among developers. All my homies hate Vercel.
I'm new to webdevelopment and was using Vercel because people told me it was good, but I was unaware that the company supported the genocide. What other similar services there are that you would recommend?
Railway/Render can easily host Next.js applications
You could also get a VPS from Digital Ocean or Hertzner, run open-source PaaS like Coolify, Dokploy, Caprover, etc.
Digital Ocean has app platform that’s lets you host these applications but more experience than VPS
Sealos has a $7 and $25 plan and work with Next.js
Just a few options. If you’re looking to leave Next.js completely, check out Tanstack Start. It’s by the creator of React-Query (defacto way to handle fetching data in Next.js). Still some rough edges but I think it could overtake Next.js once a bit more mature.
Do you really think a $30 hetzner host can sustain that level of traffic performantly? Don't get me wrong, I love hetzner, but I would be very surprised if the numbers work out there.
Isn’t it just serving static content and the content fitting in RAM? If so your laptop can serve it just fine even.
A laptop would have a hard time serve thousands of people hitting a single endpoint multiple times a day.
It shouldn't. The issue is that most developers would rather spin up another instance of their server than solve the performance issue in their code, so now it's a common belief that computers are really slow to serve content.
And we are talking about static content. You will be bottlenecked by bandwidth before you are ever bottlenecked by your laptop.
To be fair, computers are slow if you intentionally rent slow & overpriced ones from really poor-value vendors like cloud providers. For people who started their career in this madness they might be genuinely unaware of how fast modern hardware has become.
With a 2025 tech stack, yes. With a 2005 tech stack, no. Don't use any containers, no[/limited] server-side dynamic script languages, no microservices or anything like that.
Considering the content is essentially static, this is actually viable. Search functions might be a bit problematic, but that's a solvable problem.
Of course you pay with engineering skills and resources.
Is there any feasible way to implement search client-side on a database of this scale?
I guess you would need some sort of search term to document id mapping that gets downloaded to the browser but maybe there's something more efficient than trying to figure out what everyone might be searching for in advance?
And how would you do searching for phrases or substrings? I've no idea if that's doable without having a database server-side that has the whole document store to search through.
I think the key thing here is the context and size; the searchable content of even a lot of e-mails is quite dense and small. I'm not a search expert but I'd look at precalculated indexes on very short substrings (3-4 characters maybe?), have the client pull those it needs for a particular query and then process client-side from there. (Doesn't even need figuring out in advance what people will search for, though that'd certainly improve things.)
I did say you pay with engineering, didn't I? :)
Theoretically, just thinking about the problem... You could probably embrace offline first and sync to indexeddb? After that search would become simple to query. Obviously comes with it's own challenges, depending on your user base (e.g. not a good idea if it's only a temporary login etc)
there's been demos of using SQLite client-side, with the database hosted in S3, and HTTP range requests used to only fetch the necessary rows for the query.
there might be some piece I'm missing, but the first thing that comes to mind would be using that, possibly with the full-text search extension, to handle searching the metadata.
at that point you'd still be paying S3 egress costs, but I'd be very surprised if it wasn't at least an order of magnitude less expensive than Vercel.
and since it's just static file hosting, it could conceivably be moved to a VPS (or a pair of them) running nginx or Caddy or whatever, if the AWS egress was too pricey.
There are several implementations of backing an Sqlite3 database with a lazy loaded then cached network storage, including multiple that work over HTTP (iirc usually with range requests). Those basically just work.
SRE here, Containers are not causing any performance problem.
Maybe the perception comes from all the Mac and Windows devs having to run a Linux VM to use containers.
Containers themselves don't, but a lot of the ecosystem structures around them do. Like having reverse proxies (or even just piles of ethernet bridges) in front of everything.
Or if you go ping pong across containers to handle a single request. That will certainly make a laptop unable to handle this load.
No it won't. This is static content we're talking about. The only thing limiting you is your network throughput and maybe disk IO (assuming it doesn't fit in a compressed RAM). Even for an "around the globe roundtrip" latency, we're still talking few hundred msec.
Some cloud products have distorted an entire generation of developers understanding of how services can scale.
I just fired up a container on my laptop... running on kubernetes... running in a linux VM. It's lightly dynamic (no database or filesystem I/O).
While I've also got enough other stuff running that my 15 min load average is at 4 and I've got 83% RAM used ignoring buffers/caches/otherwise.
I went and grabbed a random benchmarking tool and pointed it at it with 125 concurrent connections.
Sustained an average of 13914 reqs/s. Highest latency was 53.21ms.
If there are 10,000 people online at any given time hitting the API on average once every 3 seconds (which I believe are generous numbers), you'd only be around 3.3k reqs/s, or about 24% of what my laptop could serve even before any sort of caching, CDN, or anything else.
So... if a laptop can't serve that sort of request load, it sounds more like an indictment of the site's software than anything.
I think it’s more helpful to discuss this in requests per second.
I’d interpret “thousands of people hitting a single endpoint multiple times a day” as something like 10,000 people making ~5 requests per 24 hours. That’s 0.5 requests per second.
A 6 core server or laptop can easily serve 100K requests per second, so 259B requests per month. 576x more than their current load.
A laptop from 10 years ago should be able to comfortably serve that. Computers are really really fast. I'm sorry, thousands of users or tens of thousands of requests a day is nothing.
It all depends of course, but generally no, a laptop could handle that just fine.
There may be a risk of running into thermal throttling in such a use-case, as laptops are really not designed for sustained loads of any variety. Some deal with it better than others, but few deal with it well.
Part of why this is a problem is that consumer grade NICs often tend to overload quite a lot of work to the CPU that higher end server specced NICs do themselves, as a laptop isn't really expected to have to keep up with 10K concurrent TCP connections.
I was serving >50K accounts gaming forum with ~10K daily users hitting us multiple times a day on a Pentium 133 desktop LAMP server back in 2000.
Kids these days :-)
Lol yes? It's all reads. If it can all fit in ram, great. Otherwise an SSD will do fine too.
You could probably serve it from the quad-core ARM64 inside the SSD controller, if you were trying "for the lulz".
If it's mostly static, just cache it at the http level e.g. cloudflare which I believe wouldn't even charge for 450m requests on the $20 plan at least
yes
and if it doesn't spawn up another $30 instance and add another RR entry to the dns
serving static content scales horizontally perfectly
I would use a $100/mo box with a much better CPU and more RAM, but I think the pinch point might be the 1Gbps unmetered networking that Hetzner provide.
They will sell you a 10Gbps uplink however, with (very reasonably priced) metered bandwidth.
For sure, even cheaper if you cache effectively.
No . Hetzner would terminate your server as you are not a profitable customer.
A profitable customer? How would Hetzner know if you're profitable or not?
I've hosted side projects on Hetzner for years and have never experienced anything like that. Do you have any references of projects to which it happened?
Because you are using an incredibly large amount of bandwidth for €30 a month.
They offer unlimited bandwidth with their dedicated servers under a “fair usage” policy.
The bandwidth costs would be higher than what you pay monthly, so they would simply drop you.
You are probably using very little bandwidth, so it doesn’t matter in your case.
However, I assume Jmail consumes a very large amount of bandwidth.
We handle 200x their request load on two Hetzner servers.
I have heard of hetzner terminating customer relationships if too many legal complaints are filed against your VPSes.
But not because of being "not a profitable customer". Mind sharing some links here?
https://lowendtalk.com/discussion/180504/hetzner-traffic-use...
I am not sure how one even gets 250TB/mo through a 1Gbps link. In any case, completely saturating your networking for the full month is outside most people's definition of "fair use".
Yeah but they still advertise with unlimited traffic. "All root servers have a dedicated 1 GBit uplink by default and with it unlimited traffic" https://docs.hetzner.com/robot/general/traffic/
Even before the Vercel CEO supporting a genocidal maniac. Vercel as a platform has been silently giving open source projects a "fuck you, pay me" when it comes to renewing benefits.
Have seen it happen to smaller projects and even pointed it out when Vercel took static sites down.
So they have always had a bad rep in my opinion.
2nd point resonates with me, how come he wants to cover expenses, while being connected to Israeli PM and Epstein is connected to Israel through Ehud Barak.
Isn't he going to ask for a "favor"?
Barak and bibi are political enemies (or at least we're when Barack was a relevant political figure) and besides that I haven't seen anything suggesting that his connection with bibi is more than the one meeting that was publicized.
$50k and €30 are of the same order of magnitude.
This is offtopic honestly, but I'm curious if I've been using this phrase wrong for my whole life. Doesn't "order of magnitude" refer to steps of powers of ten?
$50000 vs €30. (or €42066.30 vs €30 if I normalize the currency) 5x10^4 vs 3x10^1.
You have it right, perhaps the original poster was referring to it in a more colloquial manner, in the sense that against 200 million in revenue, 50,000 and 30 are in the same ballpark?
I took it as a joke about the USD/EUR exchange rate ;)
> in the sense that against 200 million in revenue, 50,000 and 30 are in the same ballpark
I don't understand how those are in the same ballpark? I thought saying something is in the same ballpark suggested that they are similar in scale, and the implication is that little-leauge does not play in the same ballpark as a NBA team. They are in the same category (baseball), but not at all the same level.
At a big enough scale, previously large differences are effectively 0.
50k/mo is 600,000/yr vs 360/yr at 30/mo. Thats existential for a 1MM/yr company. Neither register on a balance sheet for a 1B/yr company. They are both closer to 0 than being a major cost.
But saying that 200 million and 30 are in the same ballpark is not true in 99.99% of contexts.
Even 50k and 30 I would not say are in the same ballpark. I've worked for major corps and of course a cost saving of 50k/month would not register for the overall company but it probably would for my team. A saving of 30/month is probably not worth spending any considerable amount of time on in most non-personal contexts.
Or dump the EML for everyone to import into their own clients.
That feels gross. I'd hate to have the risk of something that was meant to be redacted and not and now you have that on your own client.
Would have been much cheaper in the first case on Cloudflare
$46k for 470m page views.
That seems extremely expensive. What the heck?
Is he using Vercel Functions as well?
I think this is where some SPA + a few instances of a Node.js server + Redis would be much cheaper.
I'd say you can probably serve this much on $1k/month? It's simple content. It's not like it needs to do complex business logic in the backend.
in other words, "we know our product is overpriced as hell, so i will pay for it to avoid further exposure of our pricing model".
this seems like an unreasonably unchartiable reading of a relatively chill and nice situation
I'm not sure I would describe the discourse around Vercel and its CEO as "relatively chill and nice". Things are perceived in context.
Recent and related:
Show HN: Jmail – Google Suite for Epstein files - https://news.ycombinator.com/item?id=46339600 - Dec 2025 (363 comments)
Insane to me a bill that large for what is effectively hosting static content. He could dump the entire thing on S3 and even with cloudfront it would be fraction of that.
Source: https://x.com/rauchg/status/2020984434338693622
(https://xcancel.com/rauchg/status/2020984434338693622)
It's a screenshot of this Twitter thread [1] for those who can't view Threads on mobile because it forces you to sign in.
[1] https://xcancel.com/rauchg/status/2020984434338693622
Garbage engineering begets garbage bills
Public files needing to be distributed to a huge population of interested persons? Sounds like the perfect situation for an oldschool torrent. That's how large data leaks were handled back in my day. 450TB is peanuts for perhaps ten thousand peers on fast residential connections.
This is an absolutely ludicrous bill for that number of page views. Even if it was built in the dumbest way possible (I'm not saying it was) this would be ridiculous. This is not as good a look for Vercel as they think it is. I would be embarrassed to shine more light on this pricing if I was the CEO.
It's common to hear rumors about SF CEOs and their NDAs with young (but legal) ladies. I hope there's no irony here, g.