If you are here because a nubela.co/proxycurl call started failing and you need something to put in its place, I want to be useful and I am going to disappoint you first: I am not the replacement, and I am not going to pretend to be.
That is the whole reason this page exists. Everything below is dated, and each claim says where it comes from.
What the record shows
LinkedIn Corporation v. Nubela Pte. Ltd. et al, case 3:25-cv-00828, Northern District of California, before Judge Charles R. Breyer. Filed 24 January 2025. The named defendants are Nubela Pte. Ltd., Proxycurl LLC, Steven Goh and Bach Le.
LinkedIn filed claims including breach of contract, fraud and deceit, violation of the Computer Fraud and Abuse Act, unfair business practices under California law, a Lanham Act claim, and misappropriation. The docket's nature-of-suit code is 840 Property Rights — Trademark.
The case was terminated by a FINAL JUDGMENT ON CONSENT entered 25 July 2025.
⚠️ Read that sentence precisely, because the precision is the point. A consent judgment is an agreed outcome entered by the court. Nobody was found liable. No court decided whether the claims were right. Anyone telling you "the court ruled that scraping LinkedIn is illegal" is reading something into this record that is not in it — and I would rather lose your click than be that page.
The evidence level, stated
I did not read PACER. Justia returned 403 to me. What I have is the docket as reproduced by a public case tracker, saved on 28 September 2026, and I checked that the saved file contains the case number, the judge, the filing date, "FINAL JUDGMENT" and "Terminated".
That is a court record reproduced by a third party, not the court document itself. It is good evidence. It is not the same thing as the filing, and I am not going to blur the two.
What the founder said, in his own words
Proxycurl's CEO, Steven Goh, published "Proxycurl Shuts Down. Thank you." on 4 July 2025 — three weeks before the judgment was entered. The relevant passages, verbatim:
"In January earlier this year (2025), LinkedIn filed [a lawsuit] against Proxycurl. Today, we are shutting Proxycurl down."
"Regardless of the merits of LinkedIn's lawsuit, there is no winning in fighting this."
He gives two reasons, and neither is about whether he would have won:
"The American Rule, which means that even if we were to win the lawsuit, we would not be able to claim legal fees."
"LinkedIn, owned by Microsoft, has more or less an unlimited war chest."
And the number that makes it concrete:
"...grow Proxycurl to a ~$10M revenue business before we had to shut it down to comply with the legal settlement with LinkedIn."
The thesis: you can lose without the merits being decided
This is not "scraping is dangerous, be careful." That framing is both lazy and wrong about this case.
Look at what the record actually contains. The nature-of-suit code is trademark. The claim list is led by breach of contract and includes Lanham Act and misappropriation — those are about agreements and about names, not about whether reading a public page is lawful. Reading public data is one thread in a six-claim complaint, and it is not the one the docket is filed under.
And look at what the settlement did to their vocabulary. On 28 September 2026 I fetched nubela.co/proxycurl — it answers HTTP 200, 116,572 bytes, and it says "Proxycurl is no longer in service." In that page's HTML, the string "LinkedIn" appears zero times. The string "Professional Social Network" appears 28 times. Their own product names went with it: what was once a LinkedIn scraper API is documented as a "Database of public Professional Social Network profiles."
The name was erased. You can read that as a settlement term or as caution afterwards; either way it tells you the dispute reached the trademark, not only the data.
So the lesson is not about proxies or rate limits. It is this: a business with roughly $10M in revenue stopped operating without any court deciding it was in the wrong, because the arithmetic of defending was worse than the arithmetic of closing. The American Rule and an opponent's balance sheet did the work that a verdict never had to do.
What this means if you depended on them
Practically, for anyone with Proxycurl in production:
- The shutdown was not abrupt, by the founder's account — he wrote that he had been "assisting our existing customers with deboarding Proxycurl" over the preceding weeks and months. If you were a customer, you likely got warning. If you inherited the code from someone who left, you did not.
- The endpoints were broad, and that is what makes replacement hard: People, Company, Contact, Reverse Email Lookup, Work Email Lookup, Jobs, School, Search and Customer APIs, plus LinkDB and a Google Sheets add-on. The person fields ran to work history, education, languages, volunteering, gender and birth date; the contact fields to personal email, phone and work email; there was a salary range field.
- The team is still working — on NinjaPear, a different product. The people did not vanish; the LinkedIn surface did.
Why nobody should be promising to replace LinkedIn today
Here is the part the other "Proxycurl alternative" pages will not tell you: the vendor promising you a drop-in LinkedIn replacement is selling you a risk you cannot see from the outside.
Not because they will necessarily be sued. Because the case above shows the outcome does not require them to be wrong — it requires them to be outspent. When your provider's survival depends on not being targeted rather than on being right, you are building on a dependency whose failure mode is instant and total, and which no amount of good engineering on your side can mitigate.
Proxycurl was, by the accounts of the people who used it, a well-run API. Backward compatibility, good documentation, transparent pricing, real uptime — the founder lists those himself as what got them to $10M. Being good at the job was not the variable.
What I do instead, and it is a different thing
I run a social-data API over ten platforms: Instagram, TikTok, Threads, GitHub, Bluesky, Mastodon, Linktree, Pinterest, SoundCloud and Medium.
None of them is LinkedIn, and none of them is going to be. That is a decision, written down and not revisited: LinkedIn is where the relevant litigation in this sector comes from, and the case above is the reason the door stays shut here. If LinkedIn data is what you need, I am not your vendor and nothing on this page is trying to convince you otherwise.
And my own exposure, which I am not going to hide
It would be comfortable to end there, as though the distinction makes me safe. It does not.
I name platforms all over this site. There is an /instagram-api page, a /tiktok-api page, a /pinterest-api page. The vector that reached Proxycurl included trademark and contract terms, not only the question of reading public data — and naming a platform across a commercial site is exactly the surface where a trademark claim lives. Their settlement erased the name from their own documentation. I use names openly.
My position differs, and I think the difference is real:
- I collect logged out, never with an account. Meta v. Bright Data (2024) went against Meta on logged-out collection, and the court explicitly left the logged-in scenario open. That is the line my collection rule is built to stay on the right side of.
- I never touch LinkedIn, which is where this sector's cases come from.
- I hold no platform account for collection, so there is no terms-of-service agreement of mine to breach in the way a contract claim needs.
But "my position differs" is not "I am immune", and I am not going to write it as though it were. The American Rule applies to me too. My balance sheet is smaller than Proxycurl's was. If I am ever on the receiving end of this, the arithmetic Steven Goh described will be worse for me than it was for him — and I will not have had $10M in revenue first.
This can happen to anyone doing this work. It can happen to me. I would rather say that on my own comparison page than have you discover I was quiet about it.
Who measured this
I did, on 28 September 2026, and I sell in this category — so check the primary sources yourself rather than taking my summary of them.
The live site and the founder's post are saved with their dates in this repository, because a page that cites a shutdown should not depend on the shut-down site still answering next year.
I also run an hourly archive of what's rising across public platforms, and I am building my own collection rather than reselling someone else's.
Sources. The Proxycurl site read at nubela.co/proxycurl on 28 September 2026 — HTTP 200, 116,572 bytes; the "LinkedIn" and "Professional Social Network" counts are from that file. The founder's post, "Proxycurl Shuts Down. Thank you.", published 4 July 2025 at nubela.co/blog/goodbye-proxycurl — all quotations are verbatim from it. The case details are from the public docket record for LinkedIn Corporation v. Nubela Pte. Ltd. et al, 3:25-cv-00828 (N.D. Cal.), as reproduced by a public case tracker and read the same day — not the court filings themselves, which I did not access. Meta v. Bright Data (2024) is cited from my own legal notes and is not re-verified here.