What does DeepL's $2 billion valuation mean for #localization and language services market: --------------------------- Today DeepL announced a new record round of $300 million at a $2 billion valuation. Except for Verbit, no one from language services market has raised that much venture capital, and no company in our field has been valued that high. With revenue projected at $100 mil, DeepL is now higher on information than a billion-dollar translation company, and it has likely just dethroned Google Translate. DeepL has become the biggest helper for professional translators, and their biggest existential threat. In the past, I've been somewhat skeptical about DeepL, but they always proved me wrong, time after time. Just look at the cons: 1) ChatGPT is 10x cheaper than DeepL, 10x more customizable, and has 3x the traffic. Can it replace DeepL? 2) IT people buy Google Translate, Amazon and Microsoft Translators bundled with hosting, and cognitive services from the world's largest Cloud providers. DeepL doesn't have that luxury 3) The number of languages is only 30, while Google, Alibaba and Microsoft support 150+, and cover many more nations around the world 4) DeepL didn't have a big marketing or sales team until a couple of years ago, and if you wanted to contact their sales or support, you had to wa-a-ait. 5) No custom models, no on-prem, no adaptive, no quality estimation API, no OCR lens, no speech recognition, no Skype Translator, and other sophisticated features. So why do clients buy DeepL by the million, and investors believe in it despite the looming threat of ChatGPT? Is it the best performance of the transformer model in the languages they doubled down on? Is it the reliability of the document translation experience that makes DeepL attractive to non-localization and non-IT folk, like your friendly neighborhood lawyer? Is it the clever understated and minimalist marketing that makes DeepL appear like they don't solicit? Looking forward to hearing your ideas in the comments. For now: long live the king!
Renato Beninatto
2mo
You beat me to it, Kostya. It's been a busy day here. Your commentary on DeepL's $2 billion valuation is spot on. This significant milestone underscores the rising importance of function-specific AI models in the localization and language services market. DeepL's specialized approach offers high accuracy, privacy, and security, making it a good choice for enterprise applications that care about this. This trend highlights the need for vertical models that excel in specific domains over "general-purpose" solutions. I've been reflecting on the insights from the book "Co-Intelligence." It suggests that we are in the dial-up era of GenAI, meaning we are still too early to make definitive decisions about its long-term impact. This perspective reminds us that the current generation of AI is the worst it will ever be. As the technology evolves, so will its capabilities and applications. Interestingly, while DeepL is making headlines, ModernMT, which offers similar translation quality, hasn't yet reached the same level of market recognition or investment. It's encouraging to see an LSP like DeepL, with its 900 employees, receiving such a strong market valuation. But the game is not over, I think.
Said Sürücü
2mo
I can't speak for other language pairs but for EN<->TR pair, it is really the most accurate translation tool on the market, just like advertised. I tried all major LLM's and MT tools. I like the feature of being able to select alternative translations anywhere on the output text in the app. I didn't see that on any competitor. Also, document translation feature doesn't have a character limit. I translated 100k words e-book once. It only took one of my 20 monthly document translation limit.
Philipp Meier
2mo
Too late to just buy them Konstantin Dranch? 😉
Darya Chernina
2mo
Interesting take, Konstantin! I always appeeciated DeepL’s high quality even in the freebie version and that was the top differentiator for me. Now reading the comments and understanding that their paid plans offered source data protection, if you will, I’d say that would be another big one for me. Focus on the quality of the output while not compromising client’s input. Win-win.
Alex Buran 🚀
1mo
My startup doctranslator.com successfully beats deepl in pdf translation. Bootstrapped. No billions of dollars raised. Can’t complain. Just shipping the code and keeping customers happy.
Laura Gutierrez
1mo
Thanks for your analysis! Always so spot on. I wonder what does Index Ventures say about the potential they see for the market here? Do you have some insights or released statement there?
Max Lobanov
2mo
Looks like soliciting :D
Ludek Janda
2mo
When I worked for a large corporation, our reason for buying DeepL were privacy concerns. Google and other generic MTs are pretty obscure about what happens to your data, whereas DeepL gave us the guarantee that we own our source and target translations and our data is not sold to third parties. However, I'm not sure if this is still the case.
Entrepreneur and Investor
2moWait wait… It is an exciting time, but let’s clarify a few things. 1) No. GPT-4o costs well more than DeepL and delivers lower quality and is 8x slower. The math on next Imminent Report. 2) Many IT people follow the company policy and get their default cloud provider. The smart ones, the early adopters, spend money on a solution that better fit the company needs. DeepL and ModernMT are used by many early adopters. Early adopters influence the early majority… 3) 30 languages covers 90% of the consumer and smb market need. Not a bad choice. 4) They had a great consumer product. All successful consumer products use word of mouth, not sales and marketing. I don’t think they need a lot of sales. They proved there is more money in consumer than enterprise. Chapeau. 5) Again, they are consumer focused. They will go to SMB at one point. Only enterprise needs those things. The question is: they got an evaluation of $2B, which is multiple times the MT market size. Are we going to see a big growth in the market? I hope yes, for all of us.