Genesis AI, backed by Khosla Ventures, has unveiled GENE-26.5, its first foundational model for robotics, alongside a demonstration of robotic hands executing complex tasks.
Genesis AI revealed its initial offering after securing $105 million in seed funding to develop foundational AI systems for robotics applications.
The startup's debut model, GENE-26.5, represents the company's attempt to create generalized AI architecture applicable across robotic systems. Rather than task-specific solutions, foundational models aim to provide a flexible base layer that developers can adapt for various robotic applications.
The demonstration showcased robotic hands performing intricate manipulations, suggesting the model can handle dexterous control tasks. This "full stack" approach indicates Genesis AI is building both the AI models and the hardware integration needed to deploy them effectively.
Foundational models have transformed software development, with large language models like GPT serving as base layers for countless applications. Genesis AI's strategy applies this pattern to robotics, a field where hardware and software have traditionally remained siloed.
The company joins other ventures pursuing similar paths, including Boston Dynamics and Figure AI, which have also announced advances in general-purpose robotic systems. The robotics industry increasingly views foundational models as essential infrastructure for scaling autonomous systems beyond narrow, single-task applications.
Genesis AI's $105 million seed round places the startup among well-funded robotics ventures, though specific details on valuation and investor composition remain limited. The funding reflects investor confidence in foundational AI approaches for robotics, despite the field's history of slow commercialization timelines.
The GENE-26.5 demo serves as proof of concept for the model's capabilities, though typical development cycles suggest broader commercial deployment remains some distance ahead. The startup's next milestones will likely focus on expanding task diversity, improving reliability, and attracting enterprise partners willing to integrate the foundational model into production systems.
London-based AI infrastructure startup Nscale is targeting up to $3 billion in its US initial public offering, potentially launching as early as September, according to sources.
Adit Singh, an early investor in semiconductor startup Cerebras, has joined Mayfield as an infrastructure partner. He will focus on investments in semiconductors, cybersecurity, and physical AI.
Starcloud has raised $250 million to build data centers in orbit as competition for limited launch capacity intensifies. The funding underscores growing demand for space-based computing infrastructure.
TextTech has unveiled its annual curated list of the 200 most promising early-stage startups globally. The Startup Battlefield 200 recognizes emerging companies across sectors showing significant potential.