Doxray continues using HOSTKEY GPU servers for document processing software

Doxray, the AI company founded in Neede in 2014 that provides automated document processing and data extraction for European organisations, is continuing its partnership with hosting provider HOSTKEY. The company has been renting dedicated GPU servers from HOSTKEY for several years for the development and operation of its software.

Doxray processes documents using AI models that require significant computing power. For that GPU processing capacity, the company relies on HOSTKEY's infrastructure, which offers high-performance servers equipped with NVIDIA and AMD graphics cards, suitable for machine learning and other compute-intensive workloads, among other uses.

Wimdo Blaauboer of Doxray says he is satisfied with the partnership, particularly with the speed at which HOSTKEY implements new technical requirements and with the support the company receives, both during the initial setup and throughout ongoing use.

What Doxray does and what it needs

Doxray focuses on the automated processing and extraction of data from documents, a branch of AI that typically requires intensive model inference. This includes recognising text, classifying document types, and extracting structured data from unstructured sources. Such tasks place high demands on computing power, particularly on the graphics processors used for parallel computation.

The company, which serves clients across Europe, raised a financing round of €85 million in early 2025 through a Nordic Bond and a Series D Extension. That funding gives the company the means to further develop its technology and expand its customer base.

The choice of dedicated servers, rather than on-demand cloud capacity, suits applications where predictable performance and fixed costs per compute unit are relevant factors. Dedicated GPU hardware gives companies like Doxray direct access to a server's full capacity, without it being shared with other users.

What HOSTKEY provides

HOSTKEY was founded in 2006 and delivers dedicated servers from TIER III data centres in multiple countries, including the Netherlands, the United Kingdom, Germany, Finland, Iceland, Turkey, and the United States. The company positions itself specifically in the segment of compute-intensive workloads, with servers equipped with professional NVIDIA and AMD GPU cards.

For AI companies training or running large models, such servers offer an alternative to public cloud providers. The presence of data centres in multiple European countries also allows companies like Doxray to manage data residency, a consideration that can be relevant in the European document processing market due to privacy regulations such as the GDPR.

Continuation as a signal of infrastructure choices

The fact that Doxray is continuing its partnership with HOSTKEY after several years points to a deliberate infrastructure strategy. Many AI startups begin with public cloud solutions for the flexibility they offer in the early stages, but as their workloads become more predictable, they switch to dedicated or hybrid environments to manage costs.

Blaauboer cites the rapid implementation of new requirements as a concrete advantage. For a company that is actively developing its models, the speed with which a hosting partner can adapt to changing hardware needs is a practical factor alongside price and capacity.

In the broader Dutch AI scene, where a growing number of companies face the question of how to organise GPU capacity in a cost-efficient and GDPR-compliant way, Doxray's choice demonstrates that European dedicated hosting is a serious option alongside the large American cloud platforms. For investors and other founders weighing similar infrastructure questions, the Doxray–HOSTKEY combination provides a concrete example of how a Dutch AI player structures its computing infrastructure over the longer term.