olieidel

Haha, thanks for linking to my website! Happy to answer any questions.

Skimmed this thread and I'm quite surprised how few people are aware of the permanent establishment problem.

Quick primer on the permanent establishment problem: You would technically have to pay taxes for your company in the place you're living (not Estonia). For example, if you're living in Germany, your Estonian OÜ would technically have to file for German taxes, too, because it's being run from Germany and it now has a permanent establishment in Germany.

So, roughly speaking, the Estonian OÜ is only useful if:

- You are in a country which doesn't have a permanent establishment problem and maybe even offers tax benefits for foreign companies (e.g. the non-dom rules of Malta, Cyprus, etc.)

- You are in a country which doesn't have a permanent establishment problem because they don't crack down on foreign company ownership (e.g. most developing countries)

In all other countries, the Estonian OÜ is likely going to cause you many tax headaches in the long run. In practice, this means:

a) Your local tax authorities don't notice or don't understand, and you're still fine, even though you'd need to file for taxes;

b) Your local tax authorities crack down on you and you need to go looking for a very expensive international tax advisor versed in Estonian and your local tax law.

A lot of slightly misleading information here.

- What you're talking about is working as a freelancer (Freiberufler), which is a subset of a sole proprietor (Einzelunternehmen); these are different business types than a GmbH, and the article is about setting up a GmbH.

- Only setting up a GmbH if you want to "invest money" is overly simplistic. People might choose a GmbH due to limited liability, a more solid shareholder structure, etc. Slightly more generalized, you do have a point that of course a sole proprietorship might be a much better and simpler choice for one-person "companies" who mainly offer software freelancing services.

- Not setting up a separate bank account for your business, GmbH or otherwise, is strongly advised against. Not only by literally 100% of tax advisors, but also by the tax office themselves.

Yup. Most of the time, you simply don't need a GmbH.

I wonder why German people often think they need to found one - maybe because people in the US have a low threshold of founding an LLC, but that's because it's easy and cheap; or maybe due to the German Angst of "I will immediately get sued, so I need limited liability".

Sure, there are lots of middlemen who are more than happy to take your money to navigate this broken process, and tax advisors are probably #1 on the list :)

That being said, even throwing money at a tax advisor won't reduce the 16 steps to 1. More like 5-10.

You'll still have to to go the notary yourself, you'll still be opening the bank account yourself, you'll still be subject to Handelsregister spam and fake invoices, etc., etc.

Thanks! Super interesting.

Yeah, they were alright - definitely a bit of a "we love to take a lot of your money" vibe, but they did have some actual knowledge.

Some of the other tax advisors in this area are way more shady - like, "do you have 30k€ in your bank account for our fixed-fee consulting" shady.

Still, I'd really love to have a more transparent and affordable choice here. Currently, to my knowledge, none exists.

Awesome. I'd be really curious which lawyer you were referring to - if you like, feel free to reach out privately (e.g. Telegram channel link on the website) :)

Ha, that's my website - thanks for posting it (and for linking to it - allaboutberlin is awesome!).

Indeed, the valuation for the purpose of exit tax is 13.75 * (avg. of yearly profits for the past 3 years); and that valuation is taxed at approximately 30%.

So, as an example, if you own 100% of a company which makes 200k€ yearly profits, your back-of-the-envelope exit tax is 200k€ * 13.75 * 0.3 = 825k€.

A few quick notes:

- If your startup is not profitable but has raised VC money (we're on the YC website after all), the tax office likes to take the VC valuation (!) instead of the valuation resulting from the 13.75 multiple. So, if the valuation in your last round was €10M, then that's your valuation for the purpose of exit tax (back of the envelope: You own 50%, €10M valuation: 0.5 * €10M * 0.3 = €1.5M exit tax; huge problem for early-stage founders who usually don't have liquidity).

- You can deduct a CEO salary from that (yearly) number if you haven't been paying yourself a salary yet - realistically, up to 150k€ / year. So if your profit is up to 150k€ / year, you can reduce it to near zero for the purpose of exit tax valuation.

- You can also supply your own company valuation, but it has to be done by a "Wirtschaftsprüfer" - this costs around 10k€ per company; if you have shares >1% in multiple companies, this means costs of n * 10k€. This is often prohibitive.

- There's a whole tax advisor industry around this exit tax topic, and it feels very shady. I've written up all my notes from (paid) tax advisor calls and shared them on my website for free (linked by in parent comment).

- There are various setups to "avoid" it (all outlined on the website). None of those setups is easy, and none of them is free. Still, if you're e.g. faced with a potential exit tax of 825k€ like in the example above, any setup which might cost less than that might be theoretically worthwhile.

- If you leave Germany and return within 11 years, you get the exit tax back - so if that's your plan, you could "just" take out a loan and it mainly becomes a liquidity problem.

- Historically, there has been a strong tendency for Germany to tighten its exit tax laws over time.

- Different people have (vastly) different opinions on how "good" or "fair" this tax is.

- Discussing the exit tax has become quite a common topic among German founders nowadays.

Roads are salted, everything else is not.

Berlin is a great place to observe policies with good intentions, yet negative second-order effects.

Distributing free potatoes will likely cause waste somewhere else, as e.g. people will buy less potatoes in supermarkets. The waste just becomes less visible as supermarkets dispose of food every day.

Another current exhibit is the prohibition of using salt for removing snow and ice from the pavements because it's "bad for plants and the ground water". While that is true to some degree, the Berlin policy conveniently ignores all second-order effects: Sidewalks are more slippery, more people get hurt. I see people slipping on snow-compacted ice almost every day. How many trees have to be saved to make it worthwhile for more people breaking their bones?

You can apply for an exemption though, e.g. if you plan to use salt on a driveway to a hospital. Processing fees for such an exemption are up to 1.4k€ [1].

The rent cap is another one. But let's go there another day..

[1] https://www.berlin.de/umwelt/themen/natur-pflanzen-artenschu...

That won't work, for multiple reasons:

- I assume you're referring to founding a UG with 1€ in Germany. If you truly found it with 1€, it'll technically be bankrupt shortly after founding it because the founding costs are around 1k€ (notary etc.), and you haven't made any revenue yet. It's generally recommended to found with at least 2-3k€.

- 10 days is not a reasonable timeframe for founding a German UG. Optimistic timeline: Instant notary appointment, 14 days for corporate registry, instant bank account, 14 days for tax ID. Total of 28 days.

In most developed countries, you should incorporate at your place of residence. That's because incorporating in a foreign country will introduce tax issues regarding the so-called "permanent establishment".

As an example, if you'd be living in Germany and choose to incorporate in Estonia, then it's likely that your Estonian company has a permanent establishment in Germany, because you (as an owner and managing director) are performing work there. This leads to your company having to file for Estonian and (!) German taxes, which quickly becomes a headache.. and potentially expensive, as you'd need to rely on international tax advisors if you run into problems.

But even besides that, the main problem is that you still are hit with German bureaucracy, even if you incorporate in Estonia - you have to file for German taxes, potentially register your Estonian company (= its permanent establishment in Germany) in the German corporate tax registry, etc.

I've looked into this fairly in-depth and also discussed it with people in the Estonian e-Residency team. They largely confirmed my analysis, which I wrote up here [1].

The conclusion, unfortunately, seems to be that incorporating in another country (e.g. Estonia) only is viable if a) you're not living in a developed country which follows up on tax payers and where their businesses are located, b) you're actually living in Estonia or c) you have a sufficient budget for actually setting up a (physical) establishment there so that you don't run into the dual-permanent-establishment problem.

[1] https://eidel.io/estonias-e-residency-is-awesome-and-sucks-t...

Interesting to see this on HN. I was part of the research group which published this back in 2015 [1], I think we were the second group worldwide to publish this.

So, first off, this is not new. The linked publication here mainly seems to be explaining a potential mechanism of how it might happen.

Some quick notes to aid in a constructive discussion - bear with me, it's been a while and I've left research and since worked as a software developer, chuckle:

- Different gadolinium agents have vastly different "buildup" characteristics - some are better, some are worse. Biochemically, the ones where the gadolinium was trapped in harder "complexes", those were more stable (less accumulation). I suck at biochemistry, so all of those words may be wrong.

- If you'd want to over-engineer this, you could indeed select your MRI hospital / practice based on which gadolinium agent they use.

- Unless you're getting a ton of MRIs (think multiple sclerosis monitoring etc.), you probably won't be affected.

- Most MRIs are without contrast agent anyway, so you probably won't be affected.

- The last I heard was that the clinical implications were still being investigated - like, yeah, you do see a buildup of gadolinium in patients who 1) get certain gadolinium agents and 2) have a ton of MRIs, but what does that mean they'll suffer any clinical consequences from this? Not sure. I heard that there was a paper (.. somewhere) which at least showed a correlation with worse MS outcomes of people who had a high buildup, but then again, cause-effect here is not clear as people with worse MS tend to have more MRIs, too (correlation != causation).

[1] https://pubs.rsna.org/doi/full/10.1148/radiol.2015150337

Pretty much all political parties loudly announce that they'll reduce bureaucracy, but, judging by the outcomes, not much has happened so far.

That being said, it's probably overly simplistic to blame political parties for this - there's a lot of e.g. county/state-level bureaucracy in Germany which gets in the way of making any sort of constructive changes. It's a bit like blaming the CEO of a bloated company for not making it "agile" in a short period of time. Sure, leadership is important, but the reality is, it's.. complicated.

Agreed. As mentioned in another comment, I think it'd be fair to levy the exit tax when you actually sell your company in the future. Like, if I ever sell my business, I'd be happy to pay my fair share of German taxes on said business, even if I'd no longer be a tax resident of Germany.

The current implementation which essentially simulates a "virtual" sale of your business once you leave the country is pretty terrible, as most normal humans don't have that sort of cash on hand because, well, they actually didn't sell their business at that point in time.

Interesting pointer on Canada - thanks!

Yes, that's true, but the implementation is.. not very elegant.

In theory, the exit tax should ensure that Germany gets the taxes of the sale of your company. So, if you ever sold your company once you're no longer in Germany, Germany wouldn't get those taxes, so it charges you immediately once you leave Germany in a sort-of "virtual" sale.

This, of course, sucks tremendously because you actually haven't sold your company, and "normal" people don't have this sort of cash on hand.

Other countries have "smarter" exit tax implementations and only charge you when you actually sell your company in the future. I think that's pretty fair. It also doesn't hinder people from leaving the country.

What about a software company founded in Germany by someone who grew up in another country, and accordingly got their education elsewhere?

What if that company is a remote company which hires people all over the world, and none of those people benefited from the {education|peace|law enforcement|trust} in Germany?

I do agree with you, in principle, that a company is somewhat coupled to the country it was founded in. The exact nature of that coupling, however, is not that simple, I would say.

Reality is complicated, I suppose :)

Good points!

1. Yeah, valid - I was assuming the default case of "you founded your company in Germany and are moving away at some stage". In that case, you could deduct the initial share capital (often €25k) from the valuation, as that was your "purchase price". In most cases, that doesn't lead to a significantly different outcome.

But yeah, if you actually bought shares of an existing company at a certain (higher) price, than of course the "taxable delta" might change your calculation.

In that respect, I was wrong as I assumed everything would get taxed. This is only roughly the case when you founded the company yourself in Germany, as mentioned above. Thanks for the correction!

2. True! As mentioned in my post, you can also pay someone to assess the value of your shares, which would most likely result in a valuation lower than 13.75x. You will have the additional costs of getting that assessment though, and you'll have to convince the authorities that your assessment is closer to the truth than the default valuation which is based on 13.75x.

Also:

- A printer (the most important equipment of any German startup founder)

- Envelopes for letters

- A stamp with your company name (some companies and agencies you deal with require you to stamp things, because a stamp obviously proves, beyond any doubt, that you are acting on behalf of your company, because obviously no one would be able to create a similar stamp with your company's name on it, right)

- A virtual office address at a coworking space (because you're receiving physical mail, and also there are weird tax reasons not to register your company at your home address)

- A mail-scanning service (because you don't want to walk to the coworking space every few days to pick up your physical mail)

- A mail-forwarding service (so that the mail gets forwarded from your virtual office address, which now has exactly no purpose at all, to your mail-scanning service)

Yup, this is possible. It would have to be at some fair market value, and you'd (obviously) have to tax that in Germany. And depending on how much you trust your buddy, you might or might not have to draft up some complicated legal framework that you indeed have the right to buy back your company at some stage :)

True! The purpose of my post was more to zoom in on the very specific case of people with small businesses, and not explain exit tax in general.

I wrote up another post with more generic notes on the exit tax [1] which might be a better post to compare to your link.

The minor benefit of my post is that I don't have an incentive to sell you expensive tax advice, chuckle..

[1] https://eidel.io/notes-and-hacks-on-germanys-exit-tax/

While rather sarcastic, your comment does hit an interesting point: How much does the infrastructure and society of any given state contribute to the "building" of a company?

I'd argue that, for software companies, not very much; at least if you contrast it with a hardware company. If you're, say, forging steel, you're using roads, trains, a lot of electricity, you've got an industrial plant, worker unions, public accident insurance, etc., etc. - a significant chunk of state-associated infrastructure is a part of your business, and was a part of your business when you built it.

But for software companies? I mean, you need a stable internet connection, good mobile phone coverage (tricky in Germany sometimes), rule of law, efficient bureaucracy (e.g. when hiring people), good banks which don't lose your money, electricity, etc. - none of these "infrastructure factors" feel as big as the ones for a hardware business.

On the contrary, for a software business, one could argue that Germany is actively hostile to you: Founding a company takes weeks / months and is expensive (notary), most processes are still paper-based, hiring people (especially internationally) is a huge pain, mobile internet is spotty, residential internet has outages. Charging customer credit cards via Stripe exposes you to a rabbit hole of VAT bureaucracy - all companies I've met so far rolled their own, broken software stack to somehow match up their Stripe + VAT charges with their internal bookkeeping software (e.g. Datev). A huge mess. It doesn't end there.

But I may be wrong.

Exactly. And they, by default, use a very high multiple (13.75) for calculating the value of your shares.

Author here. Sure, here are the sources:

- First off, your assumption is wrong that only the increase in value gets taxed. No, the entire value of your holding gets taxed, see § 6 Abs. 1 Satz 1 Außensteuergesetz (AStG) [1].

- The factor 13.75 originates from the calculation method called "vereinfachtes Ertragswertverfahren" (~ simplified earnings-based method), which itself is defined in Bewertungsgesetz (BewG), § 11 Wertpapiere und Anteile [2]

- Factor 13.75 is defined in Bewertungsgesetz (BewG), § 203 Kapitalisierungsfaktor [3]

- The tax rate of 42% is the marginal tax rate in Germany (at least below €250k income, beyond that it's 45%) - so the assumption here is that, in the year in which you leave Germany, you've already had some salary income (say, €90k) which bumps you into the marginal tax rate for any additional income on top of that.

[1] https://www.gesetze-im-internet.de/astg/__6.html

[2] https://www.gesetze-im-internet.de/bewg/__11.html

[3] https://www.gesetze-im-internet.de/bewg/__203.html

revenue != profit

People tend to forget this outside of the Tech / VC / YC bubble.

OpenAI is losing a brutal amount of money, possibly on every API request you make to them as they might be offering those at a loss (some sort of "platform play", as business dudes might call it, assuming they'll be able to lock in as many API consumers as possible before becoming profitable).

The big question here will be what will happen next: Serving LLMs will likely become cheaper (as the past has shown). But will that lead to companies like OpenAI becoming profitable? Or will that lead to all platform providers lowering their prices again, offering them at a loss again? Or will that lead to everyone self-hosting their own LLMs because serving them has become cheaper not only financially, but computationally? That's the big question.

In the meantime, OpenAI is bleeding money.

Yeah, 100% agreed!

Citation #1:

Navigate to https://ec.europa.eu/budget/financial-transparency-system/an...

1. Enter "subject of grant or contract" = "eudamed" [enter]

2. Receive the numbers until 2022 inclusive (2023 is incomplete).

3. Extrapolate conservatively.

Citation #2:

We received documents as part of a "freedom of information act" request (the EU version, named differently) which we published here. Those include numbers for 2022 and the head count, among other things.

https://openregulatory.com/articles/beudamed-better-eudamed

Compare this with Cloudflare:

1. Open website

2. See the IP 1.1.1.1. Copy.

Done.

I do get your point - and, sure, the EU website is not catastrophically terrible. But, damn, if I'm looking for a DNS, I just want the IP, I don't want five options and the mental overhead of having to determine why the hell I now am faced with five options for a DNS, which one I should choose, how they differ, etc., etc.. Add all of the IPv4/6 stuff on top of that, and.. oh man, I feel like you've lost 90%+ of interested people already.

Sigh.

1. Looks like the cost of this project so far is already ~€1M. Does it really take you a million Euros to set up a DNS server?

(Just did a quick research in the EU's financial transparency system [1], I entered "dns4eu" in the subject field. €3m budgeted, €1M used already, most of it going to a company named "Whalebone sro" (?))

2. Why does every EU-funded software project have such a terrible website? As a visitor, you get the impression that the designers took great care to obfuscate the actual product as much as possible, while throwing in random text blurbs, useless buttons and boxes.

Stuff like:

Looking for a fast, secure, and privacy-focused way to browse the internet? You're in the right place.

Yeah, sure.. just give me the product?

Reading on..

Learn everything you need to know about DNS4EU Public Service – including where it's located, how to easily set it up on your device, and what configuration options are available to best suit your needs.

Yeah, sure.. just give me the product?

Compare this with the UI of Cloudflare's 1.1.1.1 [2] which gives visitors exactly what they need. It's awesome.

---

It's hard not to be cynical about EU projects, this one included. I've had the questionable pleasure of diving deep into EU software projects and their funding when analyzing and rebuilding [3] the EU medical device database [4], a simple database with ~500k entries (~10GB on disk), which has burned €45M (!) so far and employs a team of ~50 people. Link to website with budget tracker [5].

[1] https://ec.europa.eu/budget/financial-transparency-system/an...

[2] https://one.one.one.one

[3] https://openregulatory.com/articles/beudamed-better-eudamed

[4] https://ec.europa.eu/tools/eudamed/#/screen/home

[5] https://eudamedburn.com

Help me understand this: European companies have already been interested in choosing for European alternatives instead of AWS, Azure, etc. for years.

Sure, this interest has recently gone up significantly, but it already was there, especially in the area I've been working in (hospitals / medical devices).

But here's what I can't wrap my head around: Why do the European alternatives suck so much?

- Pretty much all cloud providers seem (very) subpar to AWS. Some are better in a very isolated way, e.g. Hetzner offers better prices for cloud instances, bare metal servers and object storage. But they don't offer anything else. OVHcloud's web interface is terrible. Telekom Cloud runs Huawei hardware and software (!) and the dev experience is really bad.

- There's no real Cloudflare alternative.

- There's no real Google Workspace / Microsoft Teams competitor (Mail + Calendar + Drive).

- No search engine / Google alternative.

Or, let me rephrase this from another angle: In the US, startups regularly build serious competitor products to some of the incumbents mentioned above. Look at what fly.io and render.com did to Heroku, I think that's very impressive. Or hey.com slowly building a Gmail alternative, very cool.

So it's possible, technically.

But why not in Europe?

I can't get a satisfactory explanation for this observation. We certainly have smart people and good universities. Is it about funding? Or about the culture of not promoting entrepreneurship? Bureaucracy? GDPR?

I genuinely don't know. What do you think?

+1 on this. I did my thesis on Glioblastoma-related imaging stuff [1]. The state of the art at the time (~2016) was that, realistically, none of the current treatments were "great", unfortunately. In short, you have 1) surgery, 2) chemotherapy, 3) radiation. Those treatments did extend survival in studies, but the overall survival of Glioblastoma patients was (tragically) still very bad at 12-24 months, and none of those therapy options were a cure.

As a side note, I recommend the book "Being Mortal" from Atul Gawande. The TLDR here is that our healthcare systems tend to overtreat patients, especially those with cancer who actually have a rather bleak prognosis, because it's easier for a physician to simply order all treatments and tell the patient "all good here, good luck" instead of taking the time to sit down and have a (long) conversation about the bleak prognosis and which options are actually still worth it. By "worth it" I mean that there are trade-offs to each treatment option, and it takes some very careful weighing whether each one provides a net benefit for your friend's individual situation. E.g. surgery might extend survival by X months, but might also create, worst case, new disabilities. So now you're faced with the very difficult decision of whether to potentially live for a shorter time with less disabilities, or for a longer time with more. There's no perfect answer, but having this sort of discussion is a good step which many patients unfortunately never take. I think this is a failure in our healthcare systems and maybe in the education of physicians.

Now, if I personally had a Glioblastoma, on top of the standard of care (surgery probably makes sense etc.), I think the ketogenic diet would currently be my best shot. Yeah, sure.. it's mostly only case reports so essentially anecdotal evidence, but it does look promising.

Good luck for your friend!

[1] https://scholar.google.com/citations?user=tinu7tYAAAAJ&hl=en