In a significant move that challenges the prevailing “bigger is better” philosophy of the artificial intelligence industry, a Copenhagen-based open-source startup has emerged to spearhead a shift toward a more decentralized, secure, and sustainable AI ecosystem. This initiative directly confronts the dominance of a few technology behemoths whose massive, cloud-based Large Language Models (LLMs) have defined the current landscape. By focusing on Small Language Models (SLMs) that operate locally on user devices, this new approach presents a fundamental alternative, one that embeds privacy and efficiency into its very architecture. Backed by €2 million in pre-seed funding, the company is not just developing a product but is building the critical infrastructure layer for a new generation of AI, aiming to democratize access and reassert digital sovereignty, particularly within the European technology sphere. This counter-movement signals a growing consensus that the future of AI may not reside in ever-larger, centralized brains in the cloud, but rather in smaller, more agile intelligence running at the edge, directly in the hands of users.
The High Cost of Centralized Intelligence
The prevailing architecture for advanced AI has been built upon a foundation of massive, cloud-hosted language models, a paradigm established and controlled by a handful of non-European tech giants. While powerful, this model carries significant and often hidden costs that extend far beyond subscription fees. Operationally, these LLMs require immense computational resources, translating into exorbitant energy consumption and a substantial carbon footprint that raises serious environmental concerns. The constant need for internet connectivity and data transfer to centralized servers creates an expensive and fragile dependency. For developers and businesses, this results in escalating cloud inference bills that can become prohibitive as an application scales, effectively creating a barrier to entry that stifles innovation and prices out startups, non-profits, and public-sector organizations. This economic model concentrates power, fostering a vendor lock-in that limits flexibility and forces organizations to cede control over their core technological infrastructure to a few dominant corporations.
Beyond the economic and environmental toll, the centralized cloud model presents a fundamental threat to data security and privacy. In order for these systems to function, vast quantities of user data, often containing sensitive personal or proprietary information, must be transmitted and processed on third-party servers located across the globe. This practice creates critical vulnerabilities, exposing data to potential breaches and unauthorized access. It represents a structural loss of data control for both individuals and organizations, a reality that stands in direct conflict with stringent data protection regulations like Europe’s GDPR. Instead of privacy being a core feature, it becomes an afterthought, managed through policies and agreements rather than being embedded in the technology itself. This reliance on a single, high-value target—the centralized cloud server—also makes the entire system less resilient, more susceptible to targeted attacks, and prone to systemic failures that can disrupt services for millions of users simultaneously.
A New Paradigm for On-Device AI
In direct response to these challenges, a fundamentally different architecture is being pioneered, one that shifts the locus of computation from the cloud to the individual device. NobodyWho is at the forefront of this movement, developing a powerful open-source engine that allows thousands of existing Small Language Models to run efficiently and directly on local hardware such as laptops and mobile phones. This “device-first” approach is not just an incremental improvement but a paradigm shift that redefines the relationship between users, their data, and AI. By processing all information locally, this model ensures that sensitive data never has to leave the user’s device. This inherently guarantees true data sovereignty and achieves “privacy by design,” giving individuals and organizations complete and uncompromised control. Furthermore, this decentralized architecture bolsters security and resilience. Instead of relying on a single point of failure, computation is distributed across thousands or even millions of individual devices, making the system far more robust against targeted attacks and systemic outages.
The economic and accessibility benefits of this on-device model are equally transformative, effectively democratizing access to advanced AI capabilities. A key innovation of this architecture is the radical alteration of the cost structure; it completely eliminates the escalating financial burden of cloud inference. Because users leverage their own hardware, an application can scale from a handful of users to millions without a corresponding increase in the developer’s cloud bill. This makes sophisticated AI both accessible and affordable for a much broader range of entities, including early-stage startups and public institutions that would otherwise be priced out of the market. Recognizing that most developers are not machine learning specialists, the core strategy is to abstract away the underlying complexity. The goal is to make the integration of a local language model as trivial as incorporating any other software dependency, ideally with just two lines of code. This focus on ease of use is designed to dismantle the traditional barriers to local inference and empower a wider community to build production-grade, private, and efficient AI applications.
Forging a Distinctly European Path
The business strategy underpinning this technological revolution is as innovative as the engine itself, operating on an “open-core” model that balances community-driven development with commercial sustainability. The foundational inference engine, libraries, and developer integrations are all provided as free and open-source software under the European Public License (EUPL) 1.2. This license was deliberately chosen for its explicit permission of commercial use, a critical factor in fostering a vibrant, practical ecosystem rather than a mere technical demonstration. Monetization is planned through managed fine-tuning services. Fine-tuning a model for specific tasks is a computationally expensive and operationally complex process. The company offers to handle this complexity for businesses, charging for the required compute time plus a service fee, providing a simpler and more cost-effective solution than attempting it in-house. Once a model is fine-tuned, it can be deployed to an unlimited number of end-users with zero additional inference cost, a key advantage of the local architecture. This model acknowledges that while massive LLMs retain a role for broad reasoning tasks, the vast majority of real-world business applications—such as specialized chatbots or internal support tools—are better served by more efficient and controllable SLMs.
This entire initiative is deeply rooted in a strategic vision for Europe’s role in the global AI race. Rather than attempting to compete in the resource-intensive “bigger is better” game dominated by American and Chinese corporations, this approach posits that Europe can and should compete differently. The opportunity lies in championing an AI paradigm that reflects the continent’s core values: stringent data security, GDPR compliance, environmental sustainability, and digital sovereignty. NobodyWho’s platform is an embodiment of this vision, building technology where these values are not afterthoughts but are embedded into the very architecture. This platform-agnostic, open-source approach also creates a distinct strategic advantage. While Big Tech competitors will likely develop their own local AI solutions, they are expected to be optimized exclusively for their closed ecosystems. In contrast, an open infrastructure layer creates a more inclusive and resilient ecosystem that reinforces European values in the digital age, positioning the region as a leader in responsible and human-centric AI innovation.
The Foundation for a Decentralized Future
With the backing of prominent investors and the rapid growth of its developer community, the company successfully validated its thesis that a significant market existed for a private, efficient, and decentralized AI infrastructure. The €2 million in pre-seed funding from firms like PSV Tech and The Footprint Firm provided the necessary capital, but it was the enthusiastic adoption by over 5,000 developers that signaled the true potential of the on-device AI movement. These early successes demonstrated that the industry was ready for an alternative to the monolithic cloud model. By providing the critical tools to make smaller, powerful language models “truly plug-and-play,” the initiative did more than just launch a product; it laid the essential groundwork for a new generation of artificial intelligence. This effort represented a pivotal first wave in decentralizing AI, making it more accessible, privacy-preserving, and sustainable for everyone.
