EN
Back to the archive

The archive · AI & Models · Strategic decision · 2017–2024

DeepL's translation bet: from free quality win to a $2B enterprise AI company

Cologne's DeepL bet neural machine translation could beat Google — 100,000+ business customers, a $2B valuation, and a $300M raise to push into enterprise.

DeepL

The betNeural machine translation done better than Google Translate could win on quality alone, and a free consumer hit would grow into a business-to-business language platform.Scaling

What the business is

DeepL Translator, a neural machine-translation service covering 32 languages, plus DeepL Write Pro, an AI writing assistant for business text.

Starting capital$100M+ round at a $1B valuation (January 2023), then $300M at a $2B post-money valuation, led by Index Ventures with ICONIQ Growth, Teachers' Venture Growth, IVP, Atomico and WiL (May 22, 2024).

How it started

DeepL grew out of the online dictionary Linguee and launched DeepL Translator from Cologne in 2017, founded by Jarek Kutylowski, betting that neural networks could make machine translation clearly better than Google Translate.

What happened

DeepL scaled on word of mouth to more than 100,000 business customers including Zendesk, Nikkei, Coursera and Deutsche Bahn, supporting 32 languages, while remaining unprofitable; it raised just over $100M at a $1B valuation in January 2023, then, seeing inbound demand flatten, raised $300M at $2B in May 2024 to build out sales and R&D.

How it ended up

Still independent and scaling: the $300M was to fund sales, marketing and research, including the enterprise writing assistant DeepL Write Pro, as the company pivoted from consumer inbound to outbound enterprise relationships.

Background

DeepL grew out of the online dictionary Linguee and launched DeepL Translator from Cologne in 2017, founded by Jarek Kutylowski. The bet was that neural machine translation could be made clearly better than Google Translate, and that quality alone — the company says its translations are preferred three times more often than competitors' — would pull users in.

The strategy worked as a funnel: DeepL scaled to more than 100,000 business and organizational customers, including Zendesk, Nikkei, Coursera and Deutsche Bahn, across 32 languages, while remaining unprofitable. It raised just over $100M at a $1B valuation in January 2023, then $300M led by Index Ventures at a $2B post-money valuation on May 22, 2024.

The 2024 round marked a strategic shift. CEO Kutylowski said inbound demand only takes you so far: 'at some point, you have to start solving the problems together with your customers.' The company planned to build an outbound sales function, expand sales and marketing, and push enterprise products like DeepL Write Pro, a writing assistant tailored for business.

What has to be true

  • DeepL bet its whole wedge on measurable translation quality, a dimension consumers could judge instantly, so the product spread without a sales force.
  • The freemium consumer base became a B2B lead source: businesses discovered the tool themselves and then wanted control over how their employees used it.
  • By May 2024 roughly 60% of staff were technologists, and the raise funded the missing non-technical half — sales and marketing — instead of more science.
  • Index-led investors bought the pivot at $2B (up from $1B in January 2023), betting enterprise willingness-to-pay, not consumer advertising, was the durable revenue.

What can be applied

A quality edge in a commodity product can build a unicorn — but the moat becomes a business only when you walk users from free consumer to enterprise contracts, a different company.

Aftermath

As of the May 22, 2024 raise, DeepL planned to deploy the $300M into sales, marketing and R&D, launch DeepL Write Pro for business writing, and hire more non-technical staff as it 'transforms quite a bit into this enterprise direction' (Kutylowski). Backers included returning investors IVP, Atomico and WiL alongside new money from Index Ventures, ICONIQ Growth and Teachers' Venture Growth.

Sources

spotted an error? The archive wants to know.

Your turn

You just read one. Describe what you are building, and see who is betting on the same thing.

Free account · 3 free questions · no card

Related cases