Matthias Niessner, one of Europe’s leading researchers in AI-driven 3D modeling, has secured a substantial $13 million seed investment to pursue what many in the field consider the “holy grail”: interactive spatial foundation models capable of generating realistic, fully navigable 3D environments from simple text instructions.
Niessner, who previously co-founded the realistic AI avatar startup Synthesia (recently valued at over $2 billion), has temporarily stepped away from his academic post leading the visual computing and AI lab at the Technical University of Munich. He is now the CEO of the freshly launched SpAItial, a new startup specifically targeting groundbreaking advances in this emerging field.
Lead investor Earlybird Venture Capital, recognized for backing prominent tech companies such as UiPath and PeakGames, spearheaded the funding round, joined by Speedinvest and several angel investors. The size of the investment is noteworthy, particularly by European standards, given that SpAItial has yet to debut a comprehensive public demo beyond an initial teaser video illustrating its potential.
The strength and credibility of the founding team seem to have played a substantial role in drawing investors’ confidence. Alongside Niessner, the startup’s leadership includes Ricardo Martin-Brualla, formerly involved with Google’s 3D telepresence project Beam, and David Novotny, who previously spearheaded Meta’s text-to-3D asset creation initiative. Rounding out the founding team is Luke Rogers, a former executive at Cazoo tasked with driving business development efforts.
Despite the relative infancy of their technology, SpAItial’s founders see an enormous long-term opportunity. Their vision extends far beyond merely constructing static 3D worlds; instead, it entails creating fully interactive virtual spaces capable of realistic interactions, such as objects responding authentically to user manipulations.
Though the startup acknowledges that its ultimate ambition—empowering users to dynamically generate fully functional video game environments or digital twins from simple text prompts—is ambitious, the initial strategic focus will be on offering developers early access to SpAItial’s foundational models via APIs. Niessner hopes this approach will enable third-party developers to experiment and innovate, creating tailored applications for areas including entertainment, game design, architecture, construction, and perhaps even robotics training scenarios.
A crowded competitive landscape already exists in related areas, with startups like Odyssey—which raised $27 million for virtual entertainment experiences—and World Labs, an AI-centric venture founded by renowned researcher Fei-Fei Li currently valued over $1 billion, actively vying for dominance. Nevertheless, Niessner remains confident the market has room for multiple players, especially given the virtually limitless potential and complexity inherent in realistic spatial modeling.
Revenue generation is high on SpAItial’s short-term priority list, a notable contrast with the more typical growth-first approach common among well-funded AI newcomers. Immediate plans include targeted recruitment, expansion of computational infrastructure, and early-stage partnerships to refine their technology.
For now, SpAItial is determined to maintain a lean, expert-driven operation rather than scaling up excessively. According to Niessner, the goal isn’t initially massive staff expansions but product refinement—specifically, achieving environments detailed enough to allow, for example, realistic destruction of objects like glass in real-time interactions. Eventually, Niessner envisions a future where even a child could easily assemble their own fully operable 3D game environment virtually instantaneously, guided only by their creative descriptions typed into a simple interface.
Whether destined to revolutionize gaming, profoundly impact digital construction simulations, or transform the landscape of future augmented reality platforms, SpAItial’s founders are convinced the next era in interactive digital environment generation has only just begun.
