Oracle Made a $300b Bet on Openai. It's Paying the Price
Key topics
The Oracle-OpenAI debacle has commenters buzzing, but many argue that Oracle's real problem lies elsewhere - namely, their expiring enterprise contracts and dwindling goodwill among clients. As one commenter confirmed, their organization is actively transitioning to AWS and PostgreSQL on RDS, citing massive cost savings as the driving force. The difficulty of migrating away from Oracle is a hotly debated topic, with some saying it's a complex and high-risk endeavor, while others claim the payoff is worth it. Amidst the discussion, a tongue-in-cheek warning to "not anthropomorphize Larry Ellison" adds a dash of humor to the conversation.
Snapshot generated from the HN discussion
Discussion Activity
Very active discussionFirst comment
19m
Peak period
114
0-12h
Avg / period
21.2
Based on 127 loaded comments
Key moments
- 01Story posted
Dec 12, 2025 at 12:01 PM EST
25 days ago
Step 01 - 02First comment
Dec 12, 2025 at 12:20 PM EST
19m after posting
Step 02 - 03Peak activity
114 comments in 0-12h
Hottest window of the conversation
Step 03 - 04Latest activity
Dec 19, 2025 at 9:41 AM EST
19 days ago
Step 04
Generating AI Summary...
Analyzing up to 500 comments to identify key contributors and discussion patterns
Want the full context?
Jump to the original sources
Read the primary article or dive into the live Hacker News thread when you're ready.
Is it worth the risk/work to move everything over? For a lot of enterprises, their needs to be a huge cost savings or risk reduction. Risk usually being the most important factor the bigger the company.
My understanding is that they were relatively lucky in that most of the hard parts are in the middleware layer and rarely the DB itself - the bank has been around since the 1800s, so has a huge mishmash of technologies that go from old IBM mainframes up to more modern cloud infra. So they're already kind of used to using middleware logic to stitch together various data sources.
The funny thing is that my contact there said the primary impetus is that they see the writing on the wall for a lot of their "legacy" Sun hardware, and figure if they're going to have to redo a lot of it, they may as well re-architect the rest. They'll still be oracle DBs running in the bank for a looong time, but there'll be less and less of it.
The other half of the equation is I wonder what their new contract rate is. Even if 0% were migrating off, if 0% were migrating to then the net rate would still be negative because of business/app attrition.
https://www.youtube.com/watch?v=-zRN7XLCRhc?t=2300
Working: https://www.youtube.com/watch?v=-zRN7XLCRhc&t=2300s
~38ms0s
We've implemented aggressive desktop monitoring and blocked downloads from Oracle to avoid the Java subscription. Where it's needed, an OpenJDK distribution is used.
Where we must still use Oracle database, in some small, bespoke legacy use cases (heavy PL/SQL), we've moved to RDS with license included to avoid the direct relationship with Oracle. I get it, a big RAC customer will have a harder time, but they'll also likely have alternatives (e.g. SAP implementation to HANA).
I know of at least one vendor (Hyland) who's dropping Oracle support and providing a migration path to MS SQL. Shame not a FOSS database, but still a trend away from Oracle.
The team defined requirements, ran an RFP and demo process and did site visits to clients of each company. The SAP reference clients weren't exactly thrilled with SAP, the product was too complex and too expensive, but it was rock solid and SAP was a reliable partner. The Oracle reference clients had the usual complaints about features and flexibility, but their real beefs were that Oracle was a predatory and untrustworthy partner.
Oracle made claims in their RFP response that were proven false in the demos and site visits, confirming the claims from reference clients about the company's ethics. In contrast, SAP's RFP responses were validated by the team's due diligence.
So management decided to go with SAP. In response, a senior Oracle person tracked down all of the company's board members and made outrageous claims of incompetence against the company's executives, and alluded ominously about bad faith and conflicts of interest.
Oracle was completely hostile and off the rails when they figured out they lost the deal. I will never, ever do business with Oracle.
Unfortunately, while the SAP application seemed solid, the organization went with their HANA database which was astronomically expensive, and had a bad habit of returning different and provably incorrect results to the same deterministic SQL query every time it ran, and then the entire database would crash for all users.
Again, you're burying the lede here.
It's like the Linux fanboi stating without evidence that Windows will just accept any user name without a password, and then refusing to elaborate on that claim. Like... wat?
SAP HANA may have its faults, but I've never heard of pervasive data corruption as one of them.
Did you enter the correct NSAKEY username?
If anyone has to deal with this, I highly recommend Palisade Compliance for consulting. Ex-Oracle people who do not sell licenses, only consult on compliance and represent you during an audit.
Oof. That's a new standard for shitty company: when ex-employees build a business around protecting customers from their former employer.
E.g. you're an IT admin at Big Co overseeing software contracts. You can often get interesting insights by looking at things like how aggressive their sales reps are with end of quarter discounts (how desperate are they to meet numbers that quarter?). Or if you see a company completely dropping the ball within your org, but on CNBC you constantly hear how great the company is by pundits and analysts -- maybe you know something the pundits don't.
Often times the consensus view of a stock trails reality by a few weeks to a month - there's a lot of non-public but also non-confidential information that isn't readily available to analysts, but exposed to employees of customers/vendors/partners/end-users.
TLDR: when stock picking or day trading, pick companies within the niche of the world where you're a SME.
The example that my business school professor gave was that if you’re riding in an elevator with two executives and they talk about how they’re going to miss numbers and trade it’s not insider. If one of them tells you specifically, it is.
That's why I always shout my inside information within earshot of my financial adviser but never actually place any trades myself.
It takes companies 3-5 years for migration of these products, all of which are not CapEx funded and so get minimal resourcing without prioritisation by leadership.
The vendor isn't incentivized to fuck you over on renewal pricing as soon as the implementation is complete.
And because of the size of the contract, the customer has more leverage at renewal time.
TL;DR - these 10 year enterprise deals with Oracle allowed companies to save money in the short run and get predictable annual licensing fees. It also bought them time to get more of their application portfolio off of Oracle so when it comes time to re-up they'll negotiate those fees down.
Think about how hard it would be for you to switch from iPhone to Android. Now multiply that by 10000. That's how hard it is to switch enterprise software.
There was a recent big company that posted on Twitter about "shutting down our last Oracle server" and that was the last thing in a multi-year process or something like that.
Coordination is sometimes harder than the technology itself.
- Is it possible for a 3 person team to manage 1000 distinct Kubernetes clusters?
- No way in hell!
- What if we hypothetically pay you $2M salary each?
- Well, let me think about it, we could figure this out...
That's a joke. but unironically you could manage 1000 Kubernetes clusters with automation. why not?
only after the move is complete and assuming it's as successful as you think it would be. What usually happens is the migration takes on a life of its own and is a multi-year if not multi-decade project. It sucks up so much money and effort that a business could be using to actually build their business vs migration to a different database. Meanwhile, the account execs of the old system know you're moving off of it so say good bye to any kind of contract discounts or special treatment during emergencies.
There's entire graveyards of failed enterprise system migrations. The most likely outcome is eventually a compromise has to be made and now you have two systems to maintain and license, the legacy one, and the new one. With the promise of eventually getting off the old one but it never happens.
It took like 5 or 6 years and that $10M represents the cost of only 10 months of operations on Z.
So in such situation, I'd be tempted to actively oppose this initiative.
Even if you do move mountains and make it happen, suddenly any outages after the transition become your fault. “This never happened on the old system.”
Imagine switching between Firefox and Chrome. Between Ford and Toyota. Between Seagate and Western Digital. Between USB-C and Lightning.
>Market cap of half a trillion.
>Somehow they're "in trouble".
Mega LMAO.
None of those were in business since 1977 (w/ the exception of Nokia, which I would argue is still a successful company today, I wouldn't put it on that list).
None of those were ever valued (even close to) half a trillion, even adjusting for inflation.
Your second point is right on the spot! Its valued. By what? By others, right? Somebody says a company has a value, which might not reflect its worth. As mentioned by some other commenters, Oracle has a lot of competition. Good competition. That's why I wrote it needs to change in order to stay competitive.
Is this news to anyone in here?
So while it's hard to call a company with $50+ billion in revenue a failure, they're not nearly what they once were. That's the direction I see Oracle going.
https://stockanalysis.com/stocks/hpq/revenue/
What I do see is orgs choosing other Oracle apps like ERP which sneak the Oracle RDBMS in as part of the bundle.
Anyone using Oracle purely as a database is going to migrate to PostgreSQL eventually, but there are a lot of orgs where the database is just one part of a wider Oracle ecosystem with world-class vendor lock-in features.
I worked at a midsize that was core internet infra, where we had an in house OS and ODM hardware and FOSS DBAs. The one Oracle DB and Oracle HW was slipped in the door through finance for ERP as you say. Although I suspect that would be cloud hosted these days.
It’s not even really a “chase,” it’s a question of “if I’m building something new, what am I choosing?”
Eventually that momentum can turn into the old thing being worth actively removing.
The languages that get a lot of airtime on HN like Rust, Go, and OCaml are way down in a tier of languages that get a lot of blog posts but enjoy relatively little traction in reality.
Hotspot is the current choice for high performance programs, but is Rust lower performance in some way or are the only downsides related to its younger age?
It’s perhaps useful look at what languages brand new projects are being started with rather than just looking at what languages large established companies like Netflix are choosing.
Companies couldn't care less about the underlying platform or language, they want reliability, stability and tons of easy to find people who can work with it from Day1. Java delivers all that, and will keep delivering for upcoming decades. Big businesses and big money love this (or hate the least out of IT stacks).
It's very surprising to hear you say this, as it's so contrary to my experience.
From the smallest programs (Computer Language Benchmarks Game) to pretty big programs (web browsers), from low-level programs (OS kernels) to high-level programs (GUI Applications), from short-lived programs (command-line utilities) to long-lived programs (database servers), it's hard to think of a single segment where even average Java programs will out-perform average C, C++, or Rust programs.
I hadn't heard of QuestDB before, but it sounds like it's written in zero-GC Java using manual memory management. That's pretty unusual for Java, and would require a team of experts to pull off, I'd think. It also sounds like it drops to C++ and Rust for performance-critical tasks.
I think hybrid implementations, where a project enjoys the beneficial aspects of the language runtime at large, but delegates small, critical functions to other languages, makes sense. That keeps the C, C++, or Rust stuff contained to boundaries that are ponderable and doesn't let those language platforms dictate the overall architecture of the program.
You make a fair point about smart pointers, and median "modern C++" practices with STL data structures are unimpressive performance-wise compared to tuned custom data structures, but I can't imagine that idiomatic Java with GC overhead on top is any better.
A decade ago, a good ~80% of applicants chose to use it or C#.
I personally don't have any issues with working with it, but nobody's learning it outside of work.
On the other hand, it is quite easy to learn, so there's that going for it.
Adding AI to the oracle infrastructure cancer will certainly a boon to it's business model. Sure it might kill 10-20% of it's customers, but if it can become a pure AI parasitic play and spread it's seed, it's going to grow.
People dont realize that capitalism is size agnostic: As long as you can sell 1 boner pill for $1 million, you only need one customer rather than say 1 million pills for 1$. And, isn't it easier to keep one customer happier if they pay your bills?
Single-sign on, in-person support, certificated software, offering training courses to onboard people, undeletable logs, help with upgrading major versions..
All from a single vendor so you can pick up the phone, yell "fix it" and go on with your day.
Unless they decide to ~~extort~~audit you.
Under most circumstances, you should still pick non-oracle-DB on Oracle Cloud Infrastructure.
It was particularly bad because it was a very small family business with equally small customers. And they all had to buy oracle licenses first, which made us insanely expensive without making money lol.
Fun in hindsight
It blew my mind at the time. Oracle is so widely hated among developers, entirely justifiably, that this guy's take really shocked me. I've literally never heard another glowing recommendation for that company before or since.
We used Clarion and MSSQL7 on windows because it was cheap. Since we started making real money, some figured we could finally afford Oracle and Sun (back when they were different).
I was a junior so my job was to evaluate the migration of one of our sql servers to oracle to test it out. I talks with the Oracle team who helps people plan purchases. They took my transaction level (~100M/year) and size (1-2GB/year) and came back with $1M for the system. This replaced a functioning $10k server. And we had maybe a dozen that would have to eventually move.
When I told them the current server was $10k, they revised their estimate to $100k. I recommended we not move.
I left the company a little while later and I think they ended up buying lots of Oracle.
Companies have money and don’t mind spending on useless stuff.
https://openjdk.org/projects/babylon/articles/auto-diff