DeepL, Germany’s most valuable AI startup, is not intimidated by the financial dominance of US tech giants and continues to rely on human expertise to refine its translation model. CEO Jaroslaw Kutylowski, while maintaining a cautious demeanor, has expressed confidence in the company’s future, even as competition from companies like Microsoft and OpenAI intensifies.
In an interview with WirtschaftsWoche in March 2023, Kutylowski was asked how he plans to compete with the billions of dollars at the disposal of Microsoft and ChatGPT creator OpenAI. His response was brief: “No comment.”
Since the launch of the generative AI-based language bot ChatGPT in late 2022, the pressure on DeepL has increased. DeepL’s translation algorithms operate on a similar principle, using language models and neural networks, much like ChatGPT. While DeepL has managed to keep powerful rivals like Google Translate at bay since its translation service launched in 2017, competitors are now closing in with improved AI language models.
Despite this, Kutylowski remains optimistic about maintaining DeepL’s lead over its larger American rivals. “We are working very hard on both the technical and linguistic sides to ensure it stays that way,” said the 41-year-old during the “Chefgespräch” podcast with WirtschaftsWoche. A key factor in this optimism is the human element. According to the company, DeepL collaborates with several thousand language experts to continually improve its language model. “We need to tell the neural network what we want from the technology,” Kutylowski explained. “We want a translation to sound like the average of the internet, and for that, we need human influence.”
Kutylowski is also unfazed by the seemingly unlimited financial resources of American tech giants. In May, DeepL completed a new funding round, raising around $300 million from international investors. With a valuation of $2 billion—double that of the previous year—DeepL is now Germany’s most valuable AI startup.
However, this is still far less than the amount Microsoft has invested in OpenAI. By the end of last year alone, an estimated $13 billion from Redmond had flowed to the San Francisco startup led by Sam Altman. Such vast sums are necessary because training increasingly complex language models requires enormous resources. For instance, Google reportedly spent nearly $200 million on the current version of its Gemini model.
To put it in perspective: OpenAI’s older GPT-3 model from 2020 reportedly cost just $4.3 million to develop.
But here, too, Kutylowski remains calm. His company has always operated efficiently, both financially and in its use of computing resources. “It’s like any other startup competing against big, seemingly overwhelming rivals,” he asserted. “It’s always about doing more with fewer resources, and with enough focus, hard work, and good prioritization, we manage it.”
OpenAI takes the lead in simultaneous translation
DeepL’s next goal is to launch a simultaneous translation tool. Kutylowski announced this in May, but the product has yet to be released. In this area, OpenAI has taken the lead. Their latest language model, GPT-4, introduced in mid-May, can chat like a human and act as an interpreter. Kutylowski asks for a bit more patience, explaining, “We are still in the testing phase and are working closely with our customers to bring the tool to market with the usual DeepL quality.”
Kutylowski places his trust in DeepL’s customer base of more than 100,000 companies and organizations, many of whom use the paid DeepL Pro subscription. As a result, the Cologne-based company has been profitable for years, albeit at a modest level. According to the most recent entry available in the commercial register, DeepL generated €28.3 million in revenue in 2021 and posted a net profit of €1.5 million.
For this reason, Kutylowski sees no need to sell his company. When asked how large a check Google would need to write to acquire DeepL, Kutylowski responded noncommittally: “I’m afraid I won’t be able to answer that question.” Once again, the reserved founder demonstrates his calm determination to chart his own course.
ⓘ The machine translation market is experiencing significant growth, driven by the increasing need for efficient and cost-effective translation solutions across various industries. As of 2024, the market size is estimated at USD 716.05 million and is projected to reach USD 927.01 million by 2029, growing at a compound annual growth rate (CAGR) of 5.30%. This growth is fueled by the rising demand for content localization and the adoption of advanced technologies like neural machine translation. Major players in the market include IBM Corporation, Microsoft Corporation, and Lionbridge Technologies Inc., among others. The market is also witnessing a shift from traditional human translation to machine translation due to its efficiency and cost benefits. Additionally, the integration of artificial intelligence and big data is enhancing the accuracy and speed of machine translation, making it a vital tool for global businesses. The Asia Pacific region is expected to be the fastest-growing market, driven by the increasing internet penetration and the need for multilingual content.
/cdn.vox-cdn.com/uploads/chorus_asset/file/25617319/flappy_bird_new.png?w=120&resize=120,86&ssl=1 "Flappy Bird game set to make a comeback after a decade")
