Case study
DeepL is a global AI company building the language infrastructure that powers global business. More than 200,000 business teams and millions of individuals use DeepL’s Language AI platform to communicate globally, collaborate and operate across languages in real time. The team share what their first greenhouse gas inventory revealed — and why 94% of the carbon cost of building AI sits where most don't think to look.
At DeepL we’ve always cared about our impact, but in the early years doing the right thing was mostly ad hoc. However, we knew sustainability had to be central from early days. We chose data center locations in Sweden and Iceland because it made sense for both performance and efficiency. We understood early that sustainability made sound business sense. We moved fast and spent less time measuring our decisions.
As we grew and matured, so did the scale of our responsibility. In 2025, we took the natural next step in completing our first Greenhouse Gas (GHG) Inventory, based on 2024 operational data. For us, this marked a turning point from moving fast with good intentions to measuring our impact and proving it to our stakeholders.
Uncovering the ‘hidden’ footprint
Our first audit revealed a total footprint of 11,468 tCO2e. It was a wake-up call that brought the true scale of our work into focus.
The real surprise? The gap between our physical presence and our digital backbone. While we ensured our data centers were powered by or matched with renewable energy (keeping our Scope 2 emissions at 669 tCO2e), the data showed that 94% of our impact (10,779 tCO2e) sits within Scope 3. Specifically, the high-performance hardware and capital goods required to build world-class AI. Now that we have these numbers, we finally know where to focus.
Precision and accountability
The biggest challenge wasn’t the ‘why’ behind our ESG work, but the ‘how’. Building a GHG inventory for the first time is a detective mission and means gathering accurate data across multiple teams in a fast-moving business. But it was time well spent: this precision is now a core part of our ESG strategy. With a quantified footprint we can move away from vague environmental claims and instead hold ourselves accountable with evidence-based transparency.
What’s next
Using 2024 as our baseline is just the start. We aren’t checking a box; we’re building sustainability into the way we do business. We’ll run this audit every year to track our progress, with a firm goal to set science-backed reduction targets within 2026.
This journey has taught us that sustainability isn't a side project- it’s a data-driven engineering challenge. By mapping the exact carbon cost of our growth, we’re bringing the same rigor to our footprint that we bring to our neural networks. We are still learning and improving, of course, but today we are doing that with the confidence that comes from audit-ready processes and data.