Allbirds has officially completed its strategic pivot from direct-to-consumer footwear to artificial intelligence infrastructure, rebranding as Smartbird and selling its existing shoe operations for $43 million. The transition, initially announced in April, culminates with the formal closure of the retail division and the appointment of Nadia Carlsten as chief executive. Carlsten, a former Amazon Web Services executive and recent leader at European compute firm DCAI, assumed her role following the retail shutdown and will operate from an Amsterdam-based office while recruiting a foundational leadership and engineering team. Smartbird aims to capture a specialized segment of the enterprise AI market by providing single-tenant, managed infrastructure that prioritizes data sovereignty and direct hardware control. Rather than competing with hyperscale cloud providers or general-purpose neoclouds on price or massive scale, the company will target industries with strict regulatory and compliance requirements, including pharmaceuticals, energy, finance, and the public sector. Carlsten emphasized that client needs typically range from hundreds to thousands of GPUs, focusing on cluster agility and full-stack control rather than massive chip procurement. The company expects to deploy initial compute clusters for multiple customers before the end of the year. The pivot was financed through a $100 million equity raise and followed the divestment of Allbirds legacy assets. The corporate restructuring also required dropping the company public benefit corporation charter, which previously codified its sustainability commitments, though Smartbird board has endorsed Carlsten long-term AI strategy. Carlsten compensation package includes a $700,000 base salary and approximately $9 million in equity awards. While the broader AI infrastructure sector is currently driven by massive capital expenditures and speculative scaling, Smartbird approach targets niche enterprise workloads that require dedicated hardware management and regulatory compliance over raw computational capacity.