MinIO Removes Web UI Features from Community Version, Pushes Users to Paid Plans
69 points by JordiGH
69 points by JordiGH
It seems it’s a bit worse than that. They said in their Slack that the open source version of MinIO is in maintenance mode now: security fixes only.
We’re not done with the fallout of the end of ZIRP I guess
I’m sorry, can you explain what this means? Zero-interest rate policy? Are you saying that this created a stupid amount of lending that resulted in investors trying to make a quick cash, thereby increasing enshittification?
It meant that more money was put into riskier investments like venture capital funds (because the conservative investments didn’t make money), which meant that tech companies got lots of VC funding. Now this has ended, they need to focus more on making profits instead of just burning VC cash, and suddenly “get large user numbers to impress VCs by giving away stuff for free” isn’t attractive.
But isn’t that model - i.e. get loads of users and don’t worry about making a profit - still working for OpenAI and the like? Why is there still so much VC money sloshing around, enough to pay for AI companies to lose billions of dollars and still get enormous valuations, if the end of ZIRP really had that effect?
OpenAI is just frickin bizarre. Somehow Sam Altman has convinced his friends to give him MORE MONEY THAN EVER BEFORE by telling them straight up: “no, we have no path to profit and no plan for one. yes, we’re going to set your money on fire, more and more of it. consider your billions of dollars in the spirit of a donation. more money please. also, we might build AI God!! ah, that’s a forward-looking statement.”
I mean, he’s not even lying to them in any way, and they’re STILL throwing fistfuls of cash at him.
so I’m not sure that constitutes “working” as such. OTOH, ridiculous things can go on much longer than anyone could guess, so I’m not going to call the ending date.
I would consider OpenAI the anchor of the AI bubble. When it goes, the bubble goes. Only Nvidia will have made a profit.
It’s likely a vision of capitalism without human labor that is so seductive to any C-suite that they are willing to pour unlimited money without a specific business plan, because it’s like a Pascal wager, a very uncertain bid for god-like powers, a borderline mystical belief in a capitalist AGI singularity. So no, it’s not OpenAI alone.
Survival of the fittest. VC money still exists, but it’s redirected to what seems most profitable instead of the previous buckshot approach.
So has the amount of VC money looking for an investment gone down or not? Previously it was spread over many smaller ventures - the buckshot approach - and now it is concentrated in a few enormous ventures, such as OpenAI. But has the total amount changed?
I hear two narratives. One says that ZIRP is over so there’s less VC money available. The other says that there’s loads of VC money desperately looking for a home: that’s why OpenAI can continually lose billions and still raise more funding. Both narratives can’t be true.
David Gerard read the Pitchbook-NVCA Venture Monitor for Q1 2025 so you don’t have to:
https://pivot-to-ai.com/2025/05/03/in-2025-venture-capital-cant-pretend-everything-is-fine-any-more/
Thank you @David_Gerard!
Sorry if I’m asking dumb questions: I find this interesting but it is not something that I understand well at all.
The idea seems to be that during the days of ZIRP, the VC funds had so much cheap capital available that they could afford to invest in loads of different startups, most of which failed. But if only one or two succeeded, they would make such huge returns that they would more than compensate for the losses they made on all the other startups. This is like a very rich person being so rich that they can afford to buy millions of lottery tickets, so many that they’re bound to win once or twice, and the jackpot is big enough to cover the cost of all the tickets.
But now that the ZIRP is over, the VC funds can no longer afford to invest in so many startups. Instead, they need to be much more selective. This is the part that I find confusing. Back when I first worked for a tech startup, in the early 2000s, getting VC funding was a gruelling process. I don’t know how effective they really were but they certainly seemed to be choosing their investments carefully. I would have expected, now that ZIRP is over, the VC funds would go back to their pre-ZIRP investment strategies.
But @David_Gerard seems to be saying that they have forgotten how to do this. Instead of carefully choosing the startups that are most likely to succeed, they are just throwing everything at OpenAI, despite its lack of a realistic business plan and the fact that it is literally losing billions of dollars a year. Is this really right? In less than 20 years, has the whole VC industry forgotten how to select an investment?
Either I am completely misunderstanding something - which, to be fair, is quite likely - or there’s something very weird going on.
yeah, that’s about what happened.
I don’t understand how they forgot how to survive in a non-ZIRP environment. Some of them were in the VC biz before then!
I am pretty sure there isn’t a smart explanation and it’s as stupid as it looks. Perhaps there’s a smart explanation, but these are literally the same individuals who pursued Web3 and the Metaverse.
They have forgotten how to do deals that aren’t lottery tickets and are presently trying to reacquaint themselves with the idea of their portfolio companies acting like businesses that sell products and services and aren’t going to grant them a lottery win.
The latest is today’s Pitchbook report on the VC fondness for portfolio mergers - that is, taking two near-dead startups and merging them into a single bigger near-dead startup. The goal is to delay having to write a loss down on the books just a little bit longer.
https://pitchbook.com/news/reports/q2-2025-pitchbook-analyst-note-us-vc-backed-ma-outlook
Possibly I am too cynical, though I wonder if I’m cynical enough.
I am pretty sure there isn’t a smart explanation and it’s as stupid as it looks.
It seems that way but I keep hoping that it isn’t. I mean, I have very little experience in this area. There’s a long tradition of people, especially scientist/engineer/nerds like me, expounding confidently on topics about which they don’t really know anything. I don’t want to be one of those people. But I really struggle to see how this could be anything but phenomenally stupid.
Worse, it’s not just stupid, it’s so wasteful. At a time when our societies are having to make difficult trade-offs because we’re feeling so resource-constrained, here we have a system that appears to be burning through hundreds of billions of dollars for no gain at all. It is baffling, to say the least.
What I know about financing can be written in large type on the back of a business card, but it’s entirely possible that the 25 years since the dotcom bubble has cycled out an entire generation of VCs who actually knew what they were doing, replacing them by people just following the script written by financing Facebook.
If you’re telling your investors you’re gonna invest in “normal” companies but that their expected return will just half of what A16Z is promising, they’re not gonna give you any money.
I started just trying to make fun of bitcoins, then I had to find out how this stuff worked and pretty soon I was reading Matt Levine every day and appearing as an expert at conferences?? good lord
OpenAI is very different. They mainly lose money on ChatGPT, but it’s not really lost money, because they in turn accumulate fresh daha to further train their models. Data that none of their competitors have access to.
OpenAI is also different because AI is a major geopolitical factor at the moment and unless you’ve been living in a cave lately, you must have noticed that geopolitics is much more important than money these days. ChatGPT is an incredible intelligence gathering channel and cutting access to AI APIs would make US sanctions hurt that much more. The only other country that can compete with US companies when it comes to bulk training data access is China, via their social media alternatives like TikTok and RedNote. You can imagine the geopolitical implications of that too.
But consider: No it isn’t. Their moat is spending more money than anyone else, and DeepSeek just made that strategy look stupid. Sam Altman is now frantically spinning plates to keep it going while he can.
That’s unfortunate, but the existing state is already really stable and I would feel comfortable with describing it as functionally “done” for the usecases I’ve used it for.
I have, unfortunately, once again been blessed by my aversion for FOSS software with too fancy of a homepage.
If anyone is looking for an alternative, my personal favorite is Garage
Not vouching since I haven’t used it but there is also seaweedfs.
Haha, the CLI is called weed — that would be enough of a reason for me to try it vs. Ceph!
I run SeaweedFS, basically for my Takahe instance. I even maintain a COPR with an EL9 (I run it on Rocky Linux 9) package. (Which I should update soon, they released a new version four days ago.)
I picked it up because someone in the Internet Archive published some very interesting benchmarks and the technology looks fancy. And it was trivial to set up (once I made the RPM) and works perfectly for me, but that’s likely because I’m just storing 220mb on it…
It really is a pattern. The best software for long-term use comes from those old, ugly plain HTML websites with bad text flow and unaligned UI elements.
I’m wondering why this kind of « staple » in the industry isn’t picked up by the Apache Foundation to create an alternative ?
I expect it would be massively successful.
This is not how ASF works, it doesn’t “pick up” projects, it doesn’t employ engineers to work on its projects. Somebody would have to propose this as new project under the ASF umbrella, in that process donating the IP, trademarks etc. to the foundation, demonstrate willingness to put resources behind the project, and attract contributors from other parties.
I have heard it joked before that the ASF is the retirement home for defunct corporate foss projects. There’s a grain of truth in that.
Garage looks interesting, and I have a couple use cases for this. Are you aware of similar projects that standardize around other object storage APIs eg. Azure blob? asking for a friend :)
I wish, but unfortunately the S3 API has just won over all of the better alternatives at this point.
There was a gateway someone made that translated between a few of the object storage protocols, but I don’t know if Azure Blob was included in that nor can I find it (thanks google)
I wonder if we need an “enshittification” tag.
Apparently the whole topic is off-topic here. :-\
It’s technically business news, so yes, off-topic.
But tags define what’s on topic. Is it not a truism to say that a non-existent tag is off topic? Its non-existence is proof of such.
Enshitification has affected every company I’ve worked for more times than I’m likely to remember. Tags are a great way to organize news and events under themes. I believe it would be useful to have a community-organized catalog of enshittification events, because they simultaneously and profoundly affect so many people in the technology community.
I’m not sure @caboteria was serious about their suggestion, but had they been, I have no doubt it would make for an interesting discussion, like most proposals do.
I’d be far more interested in enshitification than culture, finance, or vibecoding – which are all on topic here.
But tags define what’s on topic.
Not really. If the finance or art tags didn’t exist the submissions under them wouldn’t be off-topic, and in reverse just because they exist doesn’t mean all finance and art is on-topic, but rather only the parts of it that fit into what’s on-topic for the site.
no, what is on-topic defines what tags there are.
The tagging system works this way for three reasons:
It keeps the site on-topic by only allowing a predefined list of tags. These tags represent what most of the users of the site want to read, so content that does not fit into any of those categories should not be submitted. It also keeps stories organized and more easily searchable.
Emphasis mine
what is the topic?
MinIO’s board and investors are on their webpage, if you want to know who to blame. Let’s follow the money! In Jan 2022 they got a big pile of Series B funding from Intel Capital which got spun off into its own company in Jan 2025. So basically, Intel Capital was considered under-performing and Intel no longer wanted to pay for its mistakes, and now they have to sink or swim on their own and are doing this by squeezing harder on their investments to get enough money to do… whatever it is they do.
But wait, there’s more! Looks like Intel Capital was mainly there to invest in Cool Technology that might, you know, buy Intel’s products for their own products – my inference is it wasn’t really there to make money, it was there to fund R&D. Turns out the decision to get rid of Intel Capital was made by Intel’s new interim CEO’s, put in by Intel’s board after they fired the old CEO. So what is really happening is Intel is axing their R&D efforts to try to make money now. I dunno why they put the old CEO in charge of Intel anyway, ‘cause while he seemed to do a good job as Intel’s CTO up until 2009?, from 2012 to 2022 he was CEO of VMWare, which, you know, has just done great during that time, and totally isn’t a limping husk of the juggernaut it once was.
Meanwhile Intel’s board itself is a vivid rogue’s gallery of 11 interesting characters:
I would say “who gives these clowns their money”, but the answer is… me, since a major investor in Intel is Vanguard, which also manages my personal retirement fund. Welp.
Of no relevance to your larger point but BlackRock is primarily an index fund manager like Vanguard; Blackstone is the leveraged buyout vampire. In case you happen to have mixed them up. Not that index managers can’t also be objectionable.
Good call, thanks for the correction. Looks like BlackRock is not particularly blameless in various things, at least as far as Wikipedia thinks. Though I admittedly got them confused at first with Blackwater, the literal murderers-for-hire, so I’m not the most reliable narrator either.
Wouldn’t be surprised if executives of all three companies go to each other’s new year’s parties, tbh.
I’m am pretty much confused about the current state of open-source. I’m a big fan of FOSS, but I would like to mention some of my observations regardless.
It generates a lot value for the economy and makes technology accessible. However, this technology, that is made free and accessible intentionally, is used by companies to generate more and more revenue with higher margins rather than driving innovation. There is a contradiction, FOSS is not primarily meant for companies to generate higher revenues (maybe it is indirectly?). This balance results in the maintainers of the FOSS projects to pay the higher price, while consumer to pay the lowest prices. This results in a competitive advantage for the FOSS consumer, which eventually pushes the costs of producing FOSS even higher.
As companies that are primarily trying to generate revenue exist, I find it natural that the same motivation spreads to the FOSS environment, as providing free tools and services for those companies to operate with higher margines could make the FOSS maintainer feel foolish.
Finally, this doesn’t mean I support the decision.
Note: I really like this working paper by Harvard trying to esimate value of Open-Source on economy.
I was considering deploying Minio at work, but it seems like that’s likely a no-go.
For my needs, I think https://github.com/versity/versitygw might actually end up being a better option.
Unlike Minio which handles the full lifecycle of storage, Versity Gateway is an S3 -> Filesystem gateway. Not really the same thing, except that I can use S3-compatible tools to access the storage on my log archive servers. Less migration work too, since existing workflows can keep using the files as files.
But no sync/replication/multi-server support. Ceph might be a better option if you want that.
I did deploy it at work, and it was a mistake. Bad performance and unusable (site) replication (if one of the replicated instances goes offline for a bit, all changes that happen in that time frame will not be caught up when it comes back online).
I don’t see any mention of replication at all in the docs? I would have assumed that it just inherits the replication setup of whatever storage backend you use.
I was referring to site replication which replicates writes to one instance of Minio to another one. It’s the only viable setup to have a redundant copy/high availability.
Ah, given the context I thought you were talking about Versity, not Minio. That makes more sense then.
Thank you for pointing out this project i’ve been looking for s3 to local file system wrapper as that’s extremely useful for development
If someone needs an alternative, there’s the awesome Garage https://garagehq.deuxfleurs.fr/
It take another approache to resiliency tho, no erasure coding but duplication. In exchange it’s made to work on spotty internet and common hardware.
Might very much fit many non-hpc non-ia usecases :) !
I’m not familiar with this tool, but can’t somebody fork the last version with the UI? If I had rtfa first, I would have seen that they have done so already…
It’s crazy to me that people were using a free tool from a startup without anticipating something like this.
Databases have always been something to be extra careful with in terms of relying on startups.
Development tools, APIs, even frameworks, all of those are things that aren’t too difficult to switch out to an alternative should the need come. Rarely pleasant, but doable. But switching out a database? Without downtime? Basically impossible. They’ve got you trapped and they know it. Oracle has been living off of this effect for decades.